- Email: [email protected]
The motorcycle to passenger car ownership ratio and economic growth: A cross-country analysis
Journal of Transport Geography 46 (2015) 122–128
Contents lists available at ScienceDirect
Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo
The motorcycle to passenger car ownership ratio and economic growth: A cross-country analysis Teik Hua Law ⇑, Hussain Hamid, Chia Ning Goh Road Safety Research Centre, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
a r t i c l e
i n f o
Article history: Received 15 April 2014 Revised 2 June 2015 Accepted 3 June 2015 Available online 11 June 2015 Keywords: Motorcycle ownership Car ownership Economic growth Fixed effects panel linear regression
a b s t r a c t Cross-country statistics have revealed steady growth in the number of motorcycles in many less advanced economic countries (LAEC) with emerging economies due to increased urbanisation and personal wealth. In contrast, an opposite trend is occurring in advanced economic countries (AEC), with cars replacing motorcycles as income grows. Motor vehicle crashes and injuries are an inevitable consequence of a high motorcycle population. This study focused on understanding how economic growth affects the motorcycle to passenger car (MPC) ownership ratio and what factors underlie this relationship. The data used in this analysis contained a sample of 80 countries at various levels of economic developmental growth over the 48-year period between 1963 and 2010. The results pointed to an inverted U-shaped relationship between the MPC ownership ratio and the per capita Gross Domestic Product (GDP). Generally, the MPC ownership ratio increased with income at a lower level and decreased with income at a higher level. The evidence indicated that urbanisation, the total road length per thousand population, and a proxy for purchasing power with regard to vehicle purchases were the underlying factors that contributed to this relationship. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The motorcycle is a major means of transportation in many less advanced economic countries (LAEC), such as Vietnam, Malaysia and Cambodia. This is due mainly to the affordability of motorcycles, with more people in LAEC able to purchase them. It is also due to their high manoeuvrability on congested roads. Cross-country statistics have revealed steady growth in the number of motorcycles in many less advanced economic countries (LAEC) with emerging economies due to increased urbanisation and personal wealth. In contrast, an opposite trend is occurring in advanced economic countries (AEC), with cars replacing motorcycles as income grows (Lai et al., 2006; Yamamoto, 2009; Chiou et al., 2009; Pongthanaisawan and Sorapipatana, 2010). Motor vehicle crashes and injuries are an inevitable consequence of a high motorcycle population (Abdul Manan and Várhelyi, 2012; NHTSA, 2008; Preusser et al., 1995; Radin Umar et al., 1995; Ranney et al., 2010; Sharma, 2008). Vehicle ownership is typically influenced by socioeconomic factors (De Jong et al., 2004) and can be modelled by aggregated or disaggregated models. The aggregate model predicts changes ⇑ Corresponding author. E-mail addresses: [email protected] (T.H. Law), [email protected] (H. Hamid), [email protected] (C.N. Goh). http://dx.doi.org/10.1016/j.jtrangeo.2015.06.007 0966-6923/Ó 2015 Elsevier Ltd. All rights reserved.
in vehicle ownership, particularly for geographic regions during certain periods (Tanner, 1978; Khan and Willumsen, 1986; Button et al., 1993; Dargay and Gately, 1999; Sillaparcharn, 2007; Ingram and Liu, 1999). The advantage of this technique is that the models do not require extensive survey data. The alternative model uses disaggregate data to model vehicle ownership at the level of users (Whelan, 2007). This model relates an individual’s propensity to own a vehicle to various owners’ attributes, such as demographic, socioeconomic, household and geographical characteristics, vehicle price, and the availability of other transport modes (Delbosc, 2013; Chiou et al., 2009; Nolan, 2010). Yet, due to the need to collect extensive and detailed socioeconomic survey data, this model structure has rarely been adopted in developing countries. Over the last three decades, many researchers have investigated the relationship between car ownership and income growth. Linear and logarithmic functions were commonly employed to describe the long-term growth of car ownership in earlier studies (Khan and Willumsen, 1986). However, one of the limitations of these models was that they led to unreasonable vehicle ownership growth at higher income levels. In response to this limitation, a sigmoid curve function with the saturation level of car ownership was adopted (Tanner, 1983; Button et al., 1993; Ingram and Liu, 1999; Dargay and Gately, 1999; Whelan et al., 2000; Whelan, 2001; Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007). With this model, car ownership increases slowly at lower income
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
levels, then increases rapidly, and finally approaches saturation level. Previous studies of the relationship between population density and car ownership revealed that a higher spatial or urban population density was associated with lower car ownership (Yamamoto, 2009; Lam and Tam, 2002; Clark, 2007, 2009; Khan and Willumsen, 1986; Hess and Ong, 2002; Riley, 2002). This is probably due to greater congestion, increased parking constraints and costs, and a more efficient public transport system in densely populated areas (Dargay and Gately, 1999; De Jong and Van de Riet, 2008; Schwanen et al., 2004). Population density was found to have a positive effect on motorcycle ownership (RAND, 2004). According to Wong (2013), this was due to motorcycles offering a more efficient and cheaper means of transport than cars in population-dense areas. Several studies demonstrated that increases in road density were associated with growth in car ownership (Sehatzadeh et al., 2011; Cao and Huang, 2013; De Jong and Van De Riet, 2008). Cao and Huang (2013) indicated that a 1% increase in road density raised car ownership by 0.13%. Motoring cost was also identified as a significant determinant of car ownership, with studies showing that higher motoring costs reduced car ownership (Dargay and Vythoulkas, 1999; Dargay and Hanly, 2007; De Jong and Van de Riet, 2008). Romilly et al. (1998) estimated that a 1% increase in motoring cost decreased car ownership by 0.3% in the short term and 2.2% in the long term. Previous empirical evidence revealed that running cost had a greater impact than purchase costs on the elasticity of car ownership (Whelan et al., 2000; Dargay and Gately, 1999). The literature on motorcycle ownership is rather scarce, with most studies conducted in East Asian countries, such as Taiwan, Malaysia, Indonesia, Thailand, Japan and Vietnam (Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007; Tuan, 2011; Tuan and Shimizu, 2005; Sanko et al., 2009; Hsu, 2005; Nagai et al., 2003; Dao and Duc, 2005). Previous evidence revealed an inverted U-shaped relationship between motorcycle ownership and income growth (Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007; Tuan, 2011; Senbil et al., 2007; Nishitateno and Burke, 2014). Pongthanaisawan and Sorapipatana (2010) indicated that as a country developed, motorcycle ownership increased but that it fell once income level exceeded the threshold level. The study attributed this finding to motorcycle ownership growing with the increasing demand for transport in the early stages of economic growth. Eventually, probably due to the prestige, convenience, safety and comfort, people shifted from motorcycles to car ownership as their income grew. Although previous empirical studies have addressed the relationship between motorcycle ownership and income growth, there is little explanation provided for the mechanisms by which income growth leads to the inverted U-shaped relationship. The present study focused on understanding how economic growth affects the motorcycle to passenger car (MPC) ownership ratio and what factors underlie this relationship. In particular, several variables that are correlated with economic growth of a country, such as urbanisation, road density and a proxy for purchasing power with regards to vehicle purchases were used to explain this relationship. The findings of this study aim to enhance the understanding of the determinants of passenger car and motorcycle ownership and the mechanisms affecting the growth of passenger car and motorcycle ownership. This would allow policy makers to formulate more effective transport policies and strategies.
2. Data and variable description The data used in this analysis contained a sample of 80 countries at various levels of economic developmental growth over the 48-year period between 1963 and 2010. The unbalanced panel
123
data consisted of 1,934 annual observations; certain variables were missing for some countries and some years in the sample.1 The countries were classified into two groups: AEC (with the Human Development Index (HDI)2 in 2010 of 0.86 or greater) and LAEC. The list of countries included in the sample is presented in Tables 1 and 2. The dependent variable was the MPC ownership ratio, and it was derived by dividing the total number of motorcycles (including motorcycles and mopeds) by the total number of passenger cars for a particular country and the year. The data on motorcycle ownership and passenger car ownership were derived primarily from various editions of the International Road Federation (IRF) World Road Statistic annual yearbooks. Supplement data were derived from Asean Japan Transport Partnership Information Centre, United Nations Economic Commission for Europe Transport Division Database and International Road and Traffic Accident Database. The per capita Real Gross Domestic Product (GDP) (Int $ 2005 constant prices: Chain series) was used as a key variable and utilized as a proxy for income. The per capita GDP contributes to the change in the MPC ownership ratio because it determines the level of road users’ affordability in purchasing motor vehicles. The per capita GDP was obtained from the Penn World Table version 7.0 (Heston et al., 2011). Data on the urban population percentage, which was used to examine the impact of urban population density on the MPC ownership ratio, were obtained from the World Development Indicator (WDI, 2009). The consumer price index (CPI)3, a measure for inflation, was used as a proxy for purchasing power with regard to vehicle purchases. This data is drawn from the WDI and it is only available for the year 1960 onwards. The total road length per thousand population was used as another explanatory variable to determine the MPC ownership ratio. Previous studies have frequently used the total road length per thousand population to explain travel patterns and vehicle ownership (Bento et al., 2005; Ingram and Liu, 1999; Riley, 2002). Data on the this variable were obtained from the WDI and the IRF databases.
3. Methodology The use of panel data regression methods has become increasingly popular with greater availability of panel data for cross-country data sets. Empirical studies have found that panel data are generally more informative, which offers greater variability, less linearity between variables, and provides more efficient estimates (Elhorst, 2010; Greene, 2003; Hsiao, 2003). Because the data used in this study is a panel data set, we used the panel linear regression with exogenous covariates. A major concern of using panel data in regression analysis is the unobserved heterogeneity. Heterogeneity bias refers to the confounding effects of unmeasured time-invariant variables, which are omitted from regression models. Ignoring these effects could lead to inconsistent estimates of model parameters (Hsiao, 2003). Econometrically, the use of a fixed effects or random effects model could control for heterogeneity and provide consistent and efficient estimates of model parameters in the presence of 1 The study by Shao et al. (2011) indicates that it is common to have unbalanced panel data for two main reasons. First, the sampling design is not balanced. Second, some data are missing, although the original design is balanced. 2 The HDI, published annually by the United Nations, measures per capita income, life expectancy and educational achievement. 3 According to the WDI, CPI reflects changes in the cost to the average consumer of acquiring a basket of goods and services that may be fixed or changed at specified intervals, such as yearly.
124
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Table 1 The list of Advanced Economic Countries (AEC) and the available years of data. No.
Country
Observation
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Australia Austria Belgium Canada Cyprus Czech Rep Denmark Estonia Finland France Germany Greece Hong Kong Iceland Ireland Israel Italy Japan Korea Rep Luxembourg Malta Netherlands New Zealand Norway Portugal Slovakia Slovenia Spain Sweden Switzerland United Kingdom United State
30 45 47 30 41 18 30 15 47 38 15 35 14 46 39 22 40 48 33 27 10 41 41 46 17 15 19 41 33 48 36 40
Period From
To
1963 1963 1964 1969 1963 1993 1977 1996 1963 1963 1991 1971 1981 1963 1965 1987 1963 1963 1971 1975 1998 1963 1963 1963 1963 1995 1992 1967 1973 1963 1975 1970
2004 2010 2010 2009 2010 2010 2008 2010 2010 2008 2010 2008 1994 2008 2010 2008 2010 2010 2008 2009 2009 2010 2008 2010 2003 2010 2010 2010 2009 2010 2010 2010
heterogeneity (Bevan and Danbolt, 2004; Greene, 2003; Wooldridge, 2002). The fixed effects model assumes that the country-specific intercept is correlated with the explanatory variables, while the random effects model assumes that the country-specific intercept is part of the error term and uncorrelated with the explanatory variables (Greene, 2003). The Hausman test was used to determine whether the random effects model or fixed effects model was more appropriate under the null hypothesis that the country-specific intercept was uncorrelated with other explanatory variables in the model. With a correlation (the null hypothesis is rejected), the more appropriate model would be the fixed effects (Baltagi, 2005; Greene, 2003). The Hausman test indicated that the fixed effects model was more appropriate than the random effects model in all models, as shown in Table 6. The fixed effects panel linear regression model can be written as, 2
lnðMPCi;t Þ ¼ ai þ x1 lnðGDPi;t Þ þ x2 ðlnðGDPi;t ÞÞ þ uyear i;t þ b0 xi;t þ ei;t
ð1Þ
where the sub-index i denotes country, t denotes period of time; ln denotes natural logarithms; ai is the country-specific intercept (fixed effects); u, x and b are the model parameters; e is error term; year is the time trend; x are other explanatory variables. 4. Results and discussion Fig. 1 shows the relationship between the MPC ownership ratio and the per capita GDP. As shown in the figure, for LAEC, the MPC ownership ratio initially increased and then decreased as income rose. This implies that there is an inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP for LAEC. On the other hand, for AEC, the MPC ownership ratio and
the per capita GDP showed a negative relationship over most of the sample range. This implies that in AEC, increases in per capita GDP lead to decreases in the MPC ownership ratio. Tables 3 and 4 present the descriptive statistics for the explanatory variables used in this study for AEC and LAEC, respectively. As can be seen, the MPC ownership ratio of AEC is lower than that of LAEC. This suggests that economic development may be an important determinant of the MPC ownership ratio. Table 5 presents a pair-wise correlation matrix between various factors and the MPC ownership ratio. As shown in Table 5, there is a high correlation between the natural logarithm of per capita GDP and the urban population percentage (0.8592). However, the results shown in Model F of Table 6 indicate that excluding the natural logarithm of per capita GDP does not significantly affect the coefficient for the urban population percentage, indicating that multicollinearity is not a problem in the estimations. Table 6 presents the regression results. The explanatory variables used in the analysis are the per capita GDP, the CPI, the urban population percentage and the total road length per thousand
Table 2 The list of LAEC and the available years of data. No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
Country
Bahrain Bangladesh Benin Botswana Brazil Brunei Bulgaria Cambodia Cameron China Colombia Costa Rica Cote d’Ivoire Ecuador Egypt Ethiopia Hungary India Indonesia Jordan Kenya Laos Latvia Lithuania Malawi Malaysia Mauritius Mexico Mongolia Morocco Niger Nigeria Pakistan Panama Peru Philippines Poland Romania Senegal South Africa Sri Lanka Swaziland Syria Thailand Togo Tunisia Turkey Vietnam
Observation
21 9 5 18 3 16 20 6 12 15 16 20 10 14 17 23 29 37 31 6 28 12 14 16 3 46 34 10 10 30 3 11 38 15 5 18 41 14 18 26 20 14 16 35 20 11 41 10
Period From
To
1987 1990 1992 1975 2000 1991 1985 2005 1968 1990 1983 1984 1966 1984 1982 1973 1972 1965 1966 2003 1966 1990 1993 1995 1980 1963 1966 1990 1994 1969 2006 1973 1963 1977 2000 1981 1970 1990 1968 1967 1969 1987 1963 1967 1966 1983 1966 2000
2008 2003 1996 2005 2004 2010 2010 2010 2006 2008 2008 2008 2007 2007 2008 2007 2010 2008 2010 2008 2008 2002 2010 2010 1982 2010 2008 2008 2003 2007 2008 1996 2008 2008 2004 2002 2010 2004 2006 2001 2003 2003 2008 2010 2007 2008 2010 2009
125
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
ln (Motorcycles/Passenger Cars)
6 4 2 0
5
6
7
8
9
10
11
12
-2 -4 -6
ln (GDP)
-8
LAEC
AEC
Fig. 1. The relationship between the motorcycle to passenger car (MPC) ownership ratio and the per capita GDP.
Table 3 Descriptive statistic for Advanced Economic Countries (AEC). Variable
Unit
Obs.
Mean
Std. Dev.
Min
Max
MPC GDP CPI URB ROAD
Motorcycle to passenger car ownership ratio Per capita Real GDP (Int $ 2005 constant prices: Chain series) Consumer price index Urban population (percentage) Road density (km per 1000 population)
1047 1047 1047 1047 1047
0.206 23,579 54.36 72.51 16.74
0.349 9690 30.23 13.69 13.65
0.001 3018 0.06 36.14 0.21
5.614 73,243 105.69 99.91 82.86
Table 4 Descriptive statistic for Less Advanced Economic Countries (LAEC). Variable
Unit
Obs.
Mean
Std. Dev.
Min
Max
MPC GDP CPI URB ROAD
Motorcycle to passenger car ownership ratio Per capita Real GDP (Int $ 2005 constant prices: Chain series) Consumer price index (2005 = 100) Urban population (percentage) Road density (km per 1000 population)
887 887 887 887 887
1.884 5619 38.19 44.23 4.94
7.483 7203 31.78 19.56 5.34
0.002 349 0.0001 8.94 0.39
77.012 54,281 101.11 88.52 33.09
Table 5 Correlation matrix.
ln(MPC) ln(GDP) ln(CPI) URB ln(ROAD) Year a
ln(MPC)
ln(GDP)
ln(CPI)
URB
ln(ROAD)
Year
1.0000 0.4701 0.1526 0.5501 0.4655 0.1369
1.0000 0.3226 0.8592a 0.6915 0.2235
1.0000 0.3041 0.1408 0.5414
1.0000 0.5856 0.2391
1.0000 0.0512
1.0000
Correlation value > 0.8.
population. The natural logarithm of the explanatory variables are used to minimize heteroscedasticity and also allow easier interpretation of the relative elasticity value of the estimates. A quadratic term was added to the per capita GDP and the urban population percentage to account for the quadratic relationship between these two variables and the MPC ownership ratio. The CPI and the total length per thousand population interacted with the per capita GDP. This allows us to examine the effect of total road length per thousand population and CPI on the MPC ownership ratio as income increased. A time trend is included to capture the effects of other exogenous time dependent variables. Model A only includes linear and quadratic terms of per capita GDP to investigate the relationship between the MPC ownership ratio and the per capita GDP. Models B to F include the influence of additional explanatory variables on this relationship. The results
of Models A to E in Table 6 confirm a statistically significant inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP. Generally, the MPC ownership ratio increased with income at a lower level and decreased with income at a higher level. The results of Models A and B indicated that the turning point for the MPC ownership ratio was in the range of US$4895 to US$6060.4 Model G includes only data after year 1993 to examine the effect of per capita GDP on the MPC ownership ratio at a higher per capita GDP level. The average GDP per capita between 1993 and 2010 for LAEC and AEC was USD$7812 and USD$28,975, respectively. The results showed a significant U-shaped relationship between the MPC ownership ratio and the per capita GDP. Specifically, the MPC ownership ratio declined with per capita GDP at a lower per capita GDP level and increased with per capita GDP at a higher per capita GDP level.5 The turning point was reached at US$9215. This finding may be attributed to people with a higher income level using motorcycles in recreational and leisure activities, especially in AEC (Jamson and Chorlton, 2009). The estimated results showed that the coefficient for the CPI was positive and statistically significant, whereas the coefficient on its interaction with the per capita GDP was estimated to be negative. 4 The turning point is calculated by, exp(x1/(2.x2)) where x1 is the coefficient on the log of GDP and x2 is the coefficient on the log GDP, whole quantity squared. 5 Model G in Table 6 was also estimated based on data after year 1984, and similar results were found (not reported).
126
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Table 6 Analysis results.
Variables ln(GDP) (ln(GDP))2 ln(CPI) ln(CPI) ⁄ ln(GDP) ln(ROAD) ln(ROAD) ⁄ ln(GDP) URB URB2 Year Constant No. Obs. No. Groups Overall model significance Hausman test (Chi-square)
Model A
Model B
Model C
Model D
Model E
6.151*** 0.362***
6.027*** 0.346***
5.613*** 0.318*** 0.231* 0.034**
3.557*** 0.237*** 0.259* 0.037** 1.035*** 0.093**
0.0012 28.284*** 1934 80 132*** 19.98***
0.0028 18.451*** 1934 80 126*** 18.89**
5.233*** 0.297*** 0.681*** 0.085*** 0. 602* 0.051 0.069*** 0.0004*** 0.0029 35.834*** 1934 80 109*** 18.36**
27.264*** 1934 80 130*** 27.29***
0.0052** 17. 203*** 1934 80 131*** 32.34***
Model F
Model G 1.917** 0.105**
1.389*** 0.164*** 1.365*** 0.131*** 0.051*** 0.0002*** 1.847*** 1934 80 107*** 24.65***
6.703* 821 79 205*** 18.99***
Model G includes only data after year 1993. * Significant at 10%. ** Significant at 5%. *** Significant at 1%.
Several studies have shown that inflation increases the price of goods and services, thereby eroding purchasing power (Manzoor et al., 2011; Molana, 1990). Therefore, under high inflation circumstances, commuters tend to opt for vehicles with lower purchase and running costs, such as motorcycles and mopeds, and this leads to an increase in the MPC ownership ratio at a lower income level. Nonetheless, the effect of CPI on the MPC ownership ratio declined as income continued to rise. This is probably because cars become a necessity 6and are more affordable at a higher income level. The quadratic coefficient for urbanisation was positive and significant in Models E and F, pointing to a convex relationship between urbanisation and the MPC ownership ratio. At a lower urban population density level, the scale of economic activities is small and less varied in scope than at a higher urban population density. Therefore, there is less of a demand for long-distance travel, and motorcycles are preferred to passenger cars may be because they are more suited to short-distance travel. However, as the urban population density and urban area increase, long-distance travel by passenger car is expected to increase. On the other hand, a higher level of urban population growth may give rise to traffic congestion, as well as strong competition for parking spaces. Karathodorou et al. (2010) and Ng et al. (2010) indicated that traffic congestion and limited parking areas discouraged potential passenger car owners. Nishitateno and Burke (2014) indicated that some commuters may choose to ride a motorcycle to avoid traffic congestion and limited parking bays. As a result, the MPC ownership ratio increases. Models D and F showed that the total road length per thousand population had a positive and statistically significant effect on the MPC ownership ratio, whereas the coefficient for its interaction with per capita GDP was estimated to be negative. These results suggest that the MPC ownership ratio increases with total road length per thousand population at a low income level. However, this effect declined with per capita GDP growth over time. A possible explanation for this finding is that road network development was mainly limited to urban areas and low-performance roads (such as unpaved and lower number of lanes) in the early stages of road network development. At this development phase, a higher road density was associated with enhanced local accessibility. Motorcycle ownership can be expected to rise with road density at this development stage. However, as an economy grows, more
6 Individuals who own cars are more probable to be employed (Ong, 1996; Holzer et al., 1994).
extensive high-performance road networks (such as expressway) are built to connect the large cities. At this stage of road development, motorcycles can be expected to be less preferable than passenger cars, possibly because motorcycles are less safe and less comfortable than cars for longer-distance travel. Consequently, the MPC ownership ratio declines. Model B indicated that the time trend variable was negative and significant, implying that some unmeasured effects caused a downward trend in the MPC ownership ratio over time. However, the inclusion of the CPI and its interaction with the per capita GDP eliminated the significance of the time trend variable in Models C, D and E, suggesting that these variables played a role in reducing MPC ownership ratio. 5. Conclusions This study focused on the effect of economic growth on the MPC ownership ratio. The estimated results pointed to an inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP. The evidence indicated that urbanisation, the total road length per thousand population, and the CPI were the underlying factors that contributed to this relationship. In view of the relationship, it could be assumed that at a low income level, motorcycle ownership is popular because the scale of the economic activities is relatively low and simple. Moreover, cars are less affordable than motorcycles at a low income level. As the level of income increases, purchasing power also rises. At this stage, more people likely switch from motorcycles to passenger cars probably for their prestige, convenience and safety. At a higher income level, more resources are also available for the government to invest in improvements of road network infrastructures. Advances in road network infrastructures are likely to lead to increased passenger car ownership. The latter is likely probably due to passenger cars being safer than motorcycles and more suitable for long-distance travel. A limitation of this study is that the models omitted other transportation modes, such as public transportation, bicycling and walking. Earlier studies have consistently pointed to a higher demand for public transportation in densely populated areas (Lam and Tam, 2002; Clark, 2007, 2009; Khan and Willumsen, 1986). This is probably due to the higher level of congestion, greater parking constraints and costs, and more organized public transportation systems in these areas (Dargay and Gately, 1999; De Jong and Van de Riet, 2008; Schwanen et al., 2004). Previous studies also showed that public transportation tended to impose lower risks on
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
passengers (Brenac and Clabaux, 2005; Elvik and Bjørnskau, 2005; Evans, 1994). In this study, we assumed that motorcycles and passenger cars were the two major private transportation modes, with commuters having to choose one mode of travel. In reality, they could select other private transportation modes, such as bicycling and walking, to meet their mobility needs. The factors that influence the use of these private transportation modes may differ from those that influence the use of motorcycles and passenger cars. Therefore, further research is needed to examine the modal shift from motorcycles to other transportation modes and to determine what factors contribute to this change. The key finding of this study is that there are complex pathways from economic growth to the change in the MPC ownership ratio, such as urbanisation, advances in road network infrastructures, and vehicle purchasing power. However, there is always a time lag between economic growth and the MPC ownership ratio. That is, economic growth may be rapid but the effects of transport policies and measures may take longer time to be felt, particularly for some newly-developed cities in LAEC. This may have implications on transport policies and strategies. Future research to address this issue is warranted and should be pursued. Many of the LAEC in the sample have per capita GDP below the turning point of the inverted U-curve for the MPC ownership ratio. This implies that these countries are still at a stage where further economic growth will result in an increase in motorcycle ownership. As reported by Law et al. (2009), growth in income at lower income levels appeared to be associated with a rise in motorcycle deaths, with the increase partly explained by the large number of motorcycles and partly by the high degree of mobility of the motorcyclists. To enhance motorcycle safety, these countries could focus on the provision of specific injury-prevention countermeasures. Such measures could include improving public transportation services and simultaneously encouraging motorcyclists to use public transportation instead of riding motorcycles, as well as designing roads with exclusive lanes for motorcycles and increasing the stringency of motorcycle helmet laws and enforcement. As indicated in the present study, the MPC ownership ratio declined once the per capita GDP level exceeded a threshold level. This result suggests that developed countries, particularly those with increased urbanisation and a higher road network density, should devote resources to accommodating the increased demand for passenger cars, without causing excessive congestion. The empirical evidence found in this study also showed that the reduction in the MPC ownership ratio after the first turning point was not permanent, with a second turning point occurring at a higher per capita GDP level. Jamson and Chorlton (2009) indicated that at higher income levels, more mature motorcycle riders take up motorcycling as a leisure pursuit rather than as a means of transport. Therefore, more emphasis should be placed on motorcycling training specifically for leisure motorcycle riders. Acknowledgement This research was funded by Universiti Putra Malaysia Research University Grant (Grant No. 05-05-10-1108RU). The funding source played absolutely no role in the study design nor any collection, analysis, or interpretation of the data. References Abdul Manan, M.M., Várhelyi, A., 2012. Motorcycle fatalities in Malaysia. IATSS Res. 36 (1), 30–39. Baltagi, B.H., 2005. Econometric Analysis of Panel Data. John Wiley and Sons Inc., New York. Bento, A.M., Cropper, M.L., Mobarak, A.M., Vinha, K., 2005. The effects of urban spatial structure on travel demand in the United States. Rev. Econ. Stat. 87 (3), 466–478.
127
Bevan, A.A., Danbolt, J., 2004. Testing for inconsistencies in the estimation of UK capital structure determinants. Appl. Financ. Econ. 14 (1), 55–66. Brenac, T., Clabaux, N., 2005. The indirect involvement of buses in traffic accident processes. Saf. Sci. 43 (10), 835–843. Button, K., Ngoe, N., Hine, J., 1993. Modelling vehicle ownership and use in low income countries. J. Transport Econ. Policy 27 (1), 57–67. Cao, X., Huang, X., 2013. City-level determinants of private car ownership in China. Asian Geogr. 30 (1), 37–53. Chiou, Y.C., Wen, C.H., Tsai, S.H., Wang, W.Y., 2009. Integrated modelling of car/motorcycle ownership, type and usage for estimating energy consumption and emissions. Transport. Res. Part A: Policy Pract. 43 (7), 665–684. Clark, S., 2007. Estimating local car ownership models. J. Transport Geogr. 15 (3), 184–197. Clark, S., 2009. Charactering and predicting car ownership using rough sets. Transport. Res. Part C: Emerg. Technol. 17 (4), 381–393. Dao, N.X., Duc, N.H., 2005. Study on the ownership of motorcycle in Vietnam. In: Proceeding the 6th, International Conference of Eastern Asia Society for Transportation Studies, Bangkok. Dargay, J., Gately, D., 1999. Income’s effect on car and vehicle ownership, worldwide: 1960–2015. Transport. Res. Part A: Policy Pract. 33 (2), 101–138. Dargay, J., Hanly, M., 2007. Volatility of car ownership, commuting mode and time in the UK. Transport. Res. Part A: Policy Pract. 41 (10), 934–948. Dargay, J.M., Vythoulkas, P.C., 1999. Estimation of a dynamic car ownership model: a pseudo-panel approach. J. Transport Econ. Policy 33 (3), 287–302. De Jong, G.C., Van de Riet, O., 2008. The driving factors of passenger transport. Eur. J. Transport Infrastruct. Res. 8 (3), 227–250. De Jong, G., Fox, J., Pieters, M., Daly, A.J., Smit, R., 2004. Comparison of car ownership models. Transport Rev. 24 (4), 379–408. Delbosc, A., 2013. Household composition and within-household car saturation in Melbourne. Transport Policy 25, 94–100. Elhorst, J.P., 2010. Spatial panel data models. In: Fischer, M.M., Getis, A. (Eds.), Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications. Springer, Berlin, pp. 377–407. Elvik, R., Bjørnskau, T., 2005. How accurately does the public perceive differences in transport risks?: an exploratory analysis of scales representing perceived risk. Accid. Anal. Prevent. 37 (6), 1005–1011. Evans, A.W., 1994. Evaluating public transport and road safety measures. Accid. Anal. Prevent. 26 (4), 411–428. Greene, W.H., 2003. Econometric Analysis. Prentice Education, Upper Saddle River, NJ. Hess, D.B., Ong, P.M., 2002. Traditional neighbourhoods and automobile ownership. Transport. Res. Rec. 1805, 35–44. Heston, A., Summers, R., Aten, B., 2011. Penn world table version 7.0. Centre for International Comparisons of Production, Income and Prices. University of Pennsylvania. Holzer, H.J., Ihlanfeldt, K.R., Sjoquist, D.L., 1994. Work, search and travel among white and black youth. J. Urban Econ. 35, 320–345. Hsiao, C., 2003. Analysis of Panel Data, second ed. Cambridge University Press, Cambridge, UK. Hsu, T.P., 2005. Comparative study on motorcycle ownership forecasting model of Asian Countries-Taiwan, Malaysia and Vietnam. In: Proceedings the 6th International Conference of Eastern Asia Society for Transpiration Studies, Bangkok. Ingram, G.K., Liu, Z., 1999. Determinants of motorization and road provision. World Bank Policy Research Working Paper No. 2042. The World Bank, Washington, D.C. Jamson, S., Chorlton, K., 2009. The changing nature of motorcycling: patterns of use and rider characteristics. Transport. Res. Part F: Traffic Psychol. Behav. 12 (4), 335–346. Karathodorou, N., Graham, D.J., Noland, R.B., 2010. Estimating the effect of urban density on fuel demand. Energy Econ. 32 (1), 86–92. Khan, A., Willumsen, L., 1986. Modelling car ownership and use in developing countries. Traffic Eng. Control 27 (11), 554–560. Lai, W.T., Lu, J.L., Chiang, Y.S., 2006. The study of multi-vehicle household mix demand model on auto–motorcycle ownership and use in Taiwan area. Transport. Plann. J. 35 (3), 309–336. Lam, W., Tam, M., 2002. Reliability of territory-wide car ownership estimates in Hong Kong. J. Transport Geogr. 10 (1), 51–60. Law, T.H., Noland, R.B., Evans, A.W., 2009. Factors associated with the relationship between motorcycle deaths and economic growth. Accid. Anal. Prevent. 41 (2), 234–240. Manzoor, M.M., Siddiqui, A.A., Sattar, A., Fahim, M., Rasheed, S., 2011. Impact of inflation on household consumption – a case of Pakistan. Int. Res. J. Finance Econ. 68, 161–167. Molana, H., 1990. A note on the effect of inflation on consumers’ expenditure. Empirical Econ. 15 (1), 1–15. Nagai, Y., Fukuda, A., Okada, Y., Hashino, Y., 2003. Two-wheeled vehicle ownership trends and issues in the Asian region. J. Eastern Asia Soc. Transport. Stud. 5, 135–146. Ng, W.S., Schipper, L., Chen, Y., 2010. China motorization trends: new directions for crowded cities. J. Transport Land Use 3 (3), 5–25. National Highway Transportation Safety Administration (NHTSA), 2008. Traffic Safety Facts 2007 Data: Motorcycles (Report No. DOT HS 810 990). NHTSA, National Centre for Statistics and Analysis, Washington, D.C. Nishitateno, S., Burke, P.J., 2014. The motorcycle Kuznets curve. J. Transport Geogr. 36, 116–123.
128
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Nolan, A., 2010. A dynamic analysis of household car ownership. Transport. Res. Part A: Policy Pract. 44 (6), 446–455. Ong, P., 1996. Work and automobile ownership among welfare recipients. Soc. Work Res. 30 (4), 255–262. Pongthanaisawan, J., Sorapipatana, C., 2010. Relationship between level of economic development and motorcycle and car ownerships and their impacts on fuel consumption and greenhouse gas emission in Thailand. Renew. Sustain. Energy Rev. 14 (9), 2966–2975. Preusser, D.F., Williams, A.F., Ulmer, R.G., 1995. Analysis of fatal motorcycle crashes: crash typing. Accid. Anal. Prevent. 27 (6), 845–851. Radin Umar, R.S., Mackay, G.M., Hills, B.L., 1995. Preliminary analysis of motorcycle accidents: short-term impacts of the running headlights campaign and regulation in Malaysia. J. Traffic Med. 23, 17–28. RAND Europe and WSP, 2004. Motorcycles and Congestion: The Effect of Modal Split – Full Summary, Department for Transport (with support from Transport for London). Ranney, M.L., Mello, M.J., Baird, J.B., Chai, P.R., Clark, M.A., 2010. Correlates of motorcycle helmet use among recent graduates of a motorcycle training course. Accid. Anal. Prevent. 42 (6), 2057–2062. Riley, K., 2002. Motor vehicles in China: the impact of demographic and economic changes. Populat. Environ. 23 (5), 479–494. Romilly, P., Song, H., Liu, X., 1998. Modelling and forecasting car ownership in Britain: a cointegration and general to specific approach. J. Transport Econ. Policy 132 (2), 165–185. Sanko, N., Maesoba, H., Dissanayake, D., Yamamoto, T., Kurauchi, S., 2009. Intertemporal analysis of household car and motorcycle ownership behaviours: the case in Nagoya Metropolitan Area of Japan, 1981–2001. IATSS Res. 33 (2), 39–53. Schwanen, T., Dijst, M., Dieleman, F.M., 2004. Policies for urban form and their impact on travel: the Netherlands experience. Urban Stud. 41 (3), 579–603. Sehatzadeh, B., Noland, R.B., Weiner, M.D., 2011. Walking frequency, cars, dogs and the built environment. Transport. Res. Part A: Policy Pract. 45 (8), 741–754. Senbil, M., Zhang, J., Fujiwara, A., 2007. Motorization in Asia – 14 countries and three metropolitan areas. Int. Assoc. Traffic Saf. Sci. Res. 31 (1), 46–58.
Shao, J., Xiao, Z., Xu, R., 2011. Estimation with unbalanced panel data having covariate measurement error. J. Stat. Plann. Inference 141 (2), 800–808. Sharma, B.R., 2008. Road traffic injuries: a major global public health crisis. Public Health 122 (12), 1399–1406. Sillaparcharn, P., 2007. Vehicle ownership and trip generation modelling: a case study of Thailand. J. Int. Assoc. Traffic Saf. Sci. 31 (2), 17–26. Tanner, J.C., 1978. Long-term forecasting of vehicle ownership and road traffic. J. Roy. Stat. Soc., Ser. A 141 (1), 13–63. Tanner, J.S., 1983. International comparisons of car ownership and car usage. Transport and Road Research Laboratory Report 1070. Department of the Environment and Transport, Crowthorne, Berkshire. Tuan, V.A., 2011. Dynamic interactions between private passenger car and motorcycle ownership in Asia: a cross-country analysis. J. Eastern Asia Soc. Transport. Stud. 9, 541–556. Tuan, V.A., Shimizu, T., 2005. Modelling of household motorcycle ownership behaviour in Hanoi city. J. Eastern Asia Soc. Transport. Stud. 6, 1751–1765. World Development Indicators (WDI), 2009. World Bank World Development Indicators January 2009, ESDS International, (MIMAS) University of Manchester. Whelan, G., 2001. Methodological advances in modelling and forecasting car ownership in Great Britain. In: Paper for European Transport Conference 2001, PTRC, Cambridge. Whelan, G., 2007. Modelling car ownership in Great Britain. Transport. Res. Part A: Policy Pract. 41 (3), 205–219. Whelan, G., Wardman, M., Daly, A., 2000. Is there a limit to car ownership growth? An exploration of household saturation levels using two novel approaches. In: Paper presented at European Transport Conference 2000, PTRC, Cambridge. Wong, K.I., 2013. An analysis of car and motorcycle ownerships in Macao. Int. J. Sustain. Transport. 7 (3), 204–225. Wooldridge, J.M., 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press, Cambridge, MA. Yamamoto, T., 2009. Comparative analysis of household car, motorcycle and bicycle ownership between Osaka metropolitan area, Japan and Kuala Lumpur, Malaysia. Transportation 36 (3), 351–366.
Contents lists available at ScienceDirect
Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo
The motorcycle to passenger car ownership ratio and economic growth: A cross-country analysis Teik Hua Law ⇑, Hussain Hamid, Chia Ning Goh Road Safety Research Centre, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
a r t i c l e
i n f o
Article history: Received 15 April 2014 Revised 2 June 2015 Accepted 3 June 2015 Available online 11 June 2015 Keywords: Motorcycle ownership Car ownership Economic growth Fixed effects panel linear regression
a b s t r a c t Cross-country statistics have revealed steady growth in the number of motorcycles in many less advanced economic countries (LAEC) with emerging economies due to increased urbanisation and personal wealth. In contrast, an opposite trend is occurring in advanced economic countries (AEC), with cars replacing motorcycles as income grows. Motor vehicle crashes and injuries are an inevitable consequence of a high motorcycle population. This study focused on understanding how economic growth affects the motorcycle to passenger car (MPC) ownership ratio and what factors underlie this relationship. The data used in this analysis contained a sample of 80 countries at various levels of economic developmental growth over the 48-year period between 1963 and 2010. The results pointed to an inverted U-shaped relationship between the MPC ownership ratio and the per capita Gross Domestic Product (GDP). Generally, the MPC ownership ratio increased with income at a lower level and decreased with income at a higher level. The evidence indicated that urbanisation, the total road length per thousand population, and a proxy for purchasing power with regard to vehicle purchases were the underlying factors that contributed to this relationship. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The motorcycle is a major means of transportation in many less advanced economic countries (LAEC), such as Vietnam, Malaysia and Cambodia. This is due mainly to the affordability of motorcycles, with more people in LAEC able to purchase them. It is also due to their high manoeuvrability on congested roads. Cross-country statistics have revealed steady growth in the number of motorcycles in many less advanced economic countries (LAEC) with emerging economies due to increased urbanisation and personal wealth. In contrast, an opposite trend is occurring in advanced economic countries (AEC), with cars replacing motorcycles as income grows (Lai et al., 2006; Yamamoto, 2009; Chiou et al., 2009; Pongthanaisawan and Sorapipatana, 2010). Motor vehicle crashes and injuries are an inevitable consequence of a high motorcycle population (Abdul Manan and Várhelyi, 2012; NHTSA, 2008; Preusser et al., 1995; Radin Umar et al., 1995; Ranney et al., 2010; Sharma, 2008). Vehicle ownership is typically influenced by socioeconomic factors (De Jong et al., 2004) and can be modelled by aggregated or disaggregated models. The aggregate model predicts changes ⇑ Corresponding author. E-mail addresses: [email protected] (T.H. Law), [email protected] (H. Hamid), [email protected] (C.N. Goh). http://dx.doi.org/10.1016/j.jtrangeo.2015.06.007 0966-6923/Ó 2015 Elsevier Ltd. All rights reserved.
in vehicle ownership, particularly for geographic regions during certain periods (Tanner, 1978; Khan and Willumsen, 1986; Button et al., 1993; Dargay and Gately, 1999; Sillaparcharn, 2007; Ingram and Liu, 1999). The advantage of this technique is that the models do not require extensive survey data. The alternative model uses disaggregate data to model vehicle ownership at the level of users (Whelan, 2007). This model relates an individual’s propensity to own a vehicle to various owners’ attributes, such as demographic, socioeconomic, household and geographical characteristics, vehicle price, and the availability of other transport modes (Delbosc, 2013; Chiou et al., 2009; Nolan, 2010). Yet, due to the need to collect extensive and detailed socioeconomic survey data, this model structure has rarely been adopted in developing countries. Over the last three decades, many researchers have investigated the relationship between car ownership and income growth. Linear and logarithmic functions were commonly employed to describe the long-term growth of car ownership in earlier studies (Khan and Willumsen, 1986). However, one of the limitations of these models was that they led to unreasonable vehicle ownership growth at higher income levels. In response to this limitation, a sigmoid curve function with the saturation level of car ownership was adopted (Tanner, 1983; Button et al., 1993; Ingram and Liu, 1999; Dargay and Gately, 1999; Whelan et al., 2000; Whelan, 2001; Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007). With this model, car ownership increases slowly at lower income
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
levels, then increases rapidly, and finally approaches saturation level. Previous studies of the relationship between population density and car ownership revealed that a higher spatial or urban population density was associated with lower car ownership (Yamamoto, 2009; Lam and Tam, 2002; Clark, 2007, 2009; Khan and Willumsen, 1986; Hess and Ong, 2002; Riley, 2002). This is probably due to greater congestion, increased parking constraints and costs, and a more efficient public transport system in densely populated areas (Dargay and Gately, 1999; De Jong and Van de Riet, 2008; Schwanen et al., 2004). Population density was found to have a positive effect on motorcycle ownership (RAND, 2004). According to Wong (2013), this was due to motorcycles offering a more efficient and cheaper means of transport than cars in population-dense areas. Several studies demonstrated that increases in road density were associated with growth in car ownership (Sehatzadeh et al., 2011; Cao and Huang, 2013; De Jong and Van De Riet, 2008). Cao and Huang (2013) indicated that a 1% increase in road density raised car ownership by 0.13%. Motoring cost was also identified as a significant determinant of car ownership, with studies showing that higher motoring costs reduced car ownership (Dargay and Vythoulkas, 1999; Dargay and Hanly, 2007; De Jong and Van de Riet, 2008). Romilly et al. (1998) estimated that a 1% increase in motoring cost decreased car ownership by 0.3% in the short term and 2.2% in the long term. Previous empirical evidence revealed that running cost had a greater impact than purchase costs on the elasticity of car ownership (Whelan et al., 2000; Dargay and Gately, 1999). The literature on motorcycle ownership is rather scarce, with most studies conducted in East Asian countries, such as Taiwan, Malaysia, Indonesia, Thailand, Japan and Vietnam (Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007; Tuan, 2011; Tuan and Shimizu, 2005; Sanko et al., 2009; Hsu, 2005; Nagai et al., 2003; Dao and Duc, 2005). Previous evidence revealed an inverted U-shaped relationship between motorcycle ownership and income growth (Pongthanaisawan and Sorapipatana, 2010; Sillaparcharn, 2007; Tuan, 2011; Senbil et al., 2007; Nishitateno and Burke, 2014). Pongthanaisawan and Sorapipatana (2010) indicated that as a country developed, motorcycle ownership increased but that it fell once income level exceeded the threshold level. The study attributed this finding to motorcycle ownership growing with the increasing demand for transport in the early stages of economic growth. Eventually, probably due to the prestige, convenience, safety and comfort, people shifted from motorcycles to car ownership as their income grew. Although previous empirical studies have addressed the relationship between motorcycle ownership and income growth, there is little explanation provided for the mechanisms by which income growth leads to the inverted U-shaped relationship. The present study focused on understanding how economic growth affects the motorcycle to passenger car (MPC) ownership ratio and what factors underlie this relationship. In particular, several variables that are correlated with economic growth of a country, such as urbanisation, road density and a proxy for purchasing power with regards to vehicle purchases were used to explain this relationship. The findings of this study aim to enhance the understanding of the determinants of passenger car and motorcycle ownership and the mechanisms affecting the growth of passenger car and motorcycle ownership. This would allow policy makers to formulate more effective transport policies and strategies.
2. Data and variable description The data used in this analysis contained a sample of 80 countries at various levels of economic developmental growth over the 48-year period between 1963 and 2010. The unbalanced panel
123
data consisted of 1,934 annual observations; certain variables were missing for some countries and some years in the sample.1 The countries were classified into two groups: AEC (with the Human Development Index (HDI)2 in 2010 of 0.86 or greater) and LAEC. The list of countries included in the sample is presented in Tables 1 and 2. The dependent variable was the MPC ownership ratio, and it was derived by dividing the total number of motorcycles (including motorcycles and mopeds) by the total number of passenger cars for a particular country and the year. The data on motorcycle ownership and passenger car ownership were derived primarily from various editions of the International Road Federation (IRF) World Road Statistic annual yearbooks. Supplement data were derived from Asean Japan Transport Partnership Information Centre, United Nations Economic Commission for Europe Transport Division Database and International Road and Traffic Accident Database. The per capita Real Gross Domestic Product (GDP) (Int $ 2005 constant prices: Chain series) was used as a key variable and utilized as a proxy for income. The per capita GDP contributes to the change in the MPC ownership ratio because it determines the level of road users’ affordability in purchasing motor vehicles. The per capita GDP was obtained from the Penn World Table version 7.0 (Heston et al., 2011). Data on the urban population percentage, which was used to examine the impact of urban population density on the MPC ownership ratio, were obtained from the World Development Indicator (WDI, 2009). The consumer price index (CPI)3, a measure for inflation, was used as a proxy for purchasing power with regard to vehicle purchases. This data is drawn from the WDI and it is only available for the year 1960 onwards. The total road length per thousand population was used as another explanatory variable to determine the MPC ownership ratio. Previous studies have frequently used the total road length per thousand population to explain travel patterns and vehicle ownership (Bento et al., 2005; Ingram and Liu, 1999; Riley, 2002). Data on the this variable were obtained from the WDI and the IRF databases.
3. Methodology The use of panel data regression methods has become increasingly popular with greater availability of panel data for cross-country data sets. Empirical studies have found that panel data are generally more informative, which offers greater variability, less linearity between variables, and provides more efficient estimates (Elhorst, 2010; Greene, 2003; Hsiao, 2003). Because the data used in this study is a panel data set, we used the panel linear regression with exogenous covariates. A major concern of using panel data in regression analysis is the unobserved heterogeneity. Heterogeneity bias refers to the confounding effects of unmeasured time-invariant variables, which are omitted from regression models. Ignoring these effects could lead to inconsistent estimates of model parameters (Hsiao, 2003). Econometrically, the use of a fixed effects or random effects model could control for heterogeneity and provide consistent and efficient estimates of model parameters in the presence of 1 The study by Shao et al. (2011) indicates that it is common to have unbalanced panel data for two main reasons. First, the sampling design is not balanced. Second, some data are missing, although the original design is balanced. 2 The HDI, published annually by the United Nations, measures per capita income, life expectancy and educational achievement. 3 According to the WDI, CPI reflects changes in the cost to the average consumer of acquiring a basket of goods and services that may be fixed or changed at specified intervals, such as yearly.
124
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Table 1 The list of Advanced Economic Countries (AEC) and the available years of data. No.
Country
Observation
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Australia Austria Belgium Canada Cyprus Czech Rep Denmark Estonia Finland France Germany Greece Hong Kong Iceland Ireland Israel Italy Japan Korea Rep Luxembourg Malta Netherlands New Zealand Norway Portugal Slovakia Slovenia Spain Sweden Switzerland United Kingdom United State
30 45 47 30 41 18 30 15 47 38 15 35 14 46 39 22 40 48 33 27 10 41 41 46 17 15 19 41 33 48 36 40
Period From
To
1963 1963 1964 1969 1963 1993 1977 1996 1963 1963 1991 1971 1981 1963 1965 1987 1963 1963 1971 1975 1998 1963 1963 1963 1963 1995 1992 1967 1973 1963 1975 1970
2004 2010 2010 2009 2010 2010 2008 2010 2010 2008 2010 2008 1994 2008 2010 2008 2010 2010 2008 2009 2009 2010 2008 2010 2003 2010 2010 2010 2009 2010 2010 2010
heterogeneity (Bevan and Danbolt, 2004; Greene, 2003; Wooldridge, 2002). The fixed effects model assumes that the country-specific intercept is correlated with the explanatory variables, while the random effects model assumes that the country-specific intercept is part of the error term and uncorrelated with the explanatory variables (Greene, 2003). The Hausman test was used to determine whether the random effects model or fixed effects model was more appropriate under the null hypothesis that the country-specific intercept was uncorrelated with other explanatory variables in the model. With a correlation (the null hypothesis is rejected), the more appropriate model would be the fixed effects (Baltagi, 2005; Greene, 2003). The Hausman test indicated that the fixed effects model was more appropriate than the random effects model in all models, as shown in Table 6. The fixed effects panel linear regression model can be written as, 2
lnðMPCi;t Þ ¼ ai þ x1 lnðGDPi;t Þ þ x2 ðlnðGDPi;t ÞÞ þ uyear i;t þ b0 xi;t þ ei;t
ð1Þ
where the sub-index i denotes country, t denotes period of time; ln denotes natural logarithms; ai is the country-specific intercept (fixed effects); u, x and b are the model parameters; e is error term; year is the time trend; x are other explanatory variables. 4. Results and discussion Fig. 1 shows the relationship between the MPC ownership ratio and the per capita GDP. As shown in the figure, for LAEC, the MPC ownership ratio initially increased and then decreased as income rose. This implies that there is an inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP for LAEC. On the other hand, for AEC, the MPC ownership ratio and
the per capita GDP showed a negative relationship over most of the sample range. This implies that in AEC, increases in per capita GDP lead to decreases in the MPC ownership ratio. Tables 3 and 4 present the descriptive statistics for the explanatory variables used in this study for AEC and LAEC, respectively. As can be seen, the MPC ownership ratio of AEC is lower than that of LAEC. This suggests that economic development may be an important determinant of the MPC ownership ratio. Table 5 presents a pair-wise correlation matrix between various factors and the MPC ownership ratio. As shown in Table 5, there is a high correlation between the natural logarithm of per capita GDP and the urban population percentage (0.8592). However, the results shown in Model F of Table 6 indicate that excluding the natural logarithm of per capita GDP does not significantly affect the coefficient for the urban population percentage, indicating that multicollinearity is not a problem in the estimations. Table 6 presents the regression results. The explanatory variables used in the analysis are the per capita GDP, the CPI, the urban population percentage and the total road length per thousand
Table 2 The list of LAEC and the available years of data. No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
Country
Bahrain Bangladesh Benin Botswana Brazil Brunei Bulgaria Cambodia Cameron China Colombia Costa Rica Cote d’Ivoire Ecuador Egypt Ethiopia Hungary India Indonesia Jordan Kenya Laos Latvia Lithuania Malawi Malaysia Mauritius Mexico Mongolia Morocco Niger Nigeria Pakistan Panama Peru Philippines Poland Romania Senegal South Africa Sri Lanka Swaziland Syria Thailand Togo Tunisia Turkey Vietnam
Observation
21 9 5 18 3 16 20 6 12 15 16 20 10 14 17 23 29 37 31 6 28 12 14 16 3 46 34 10 10 30 3 11 38 15 5 18 41 14 18 26 20 14 16 35 20 11 41 10
Period From
To
1987 1990 1992 1975 2000 1991 1985 2005 1968 1990 1983 1984 1966 1984 1982 1973 1972 1965 1966 2003 1966 1990 1993 1995 1980 1963 1966 1990 1994 1969 2006 1973 1963 1977 2000 1981 1970 1990 1968 1967 1969 1987 1963 1967 1966 1983 1966 2000
2008 2003 1996 2005 2004 2010 2010 2010 2006 2008 2008 2008 2007 2007 2008 2007 2010 2008 2010 2008 2008 2002 2010 2010 1982 2010 2008 2008 2003 2007 2008 1996 2008 2008 2004 2002 2010 2004 2006 2001 2003 2003 2008 2010 2007 2008 2010 2009
125
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
ln (Motorcycles/Passenger Cars)
6 4 2 0
5
6
7
8
9
10
11
12
-2 -4 -6
ln (GDP)
-8
LAEC
AEC
Fig. 1. The relationship between the motorcycle to passenger car (MPC) ownership ratio and the per capita GDP.
Table 3 Descriptive statistic for Advanced Economic Countries (AEC). Variable
Unit
Obs.
Mean
Std. Dev.
Min
Max
MPC GDP CPI URB ROAD
Motorcycle to passenger car ownership ratio Per capita Real GDP (Int $ 2005 constant prices: Chain series) Consumer price index Urban population (percentage) Road density (km per 1000 population)
1047 1047 1047 1047 1047
0.206 23,579 54.36 72.51 16.74
0.349 9690 30.23 13.69 13.65
0.001 3018 0.06 36.14 0.21
5.614 73,243 105.69 99.91 82.86
Table 4 Descriptive statistic for Less Advanced Economic Countries (LAEC). Variable
Unit
Obs.
Mean
Std. Dev.
Min
Max
MPC GDP CPI URB ROAD
Motorcycle to passenger car ownership ratio Per capita Real GDP (Int $ 2005 constant prices: Chain series) Consumer price index (2005 = 100) Urban population (percentage) Road density (km per 1000 population)
887 887 887 887 887
1.884 5619 38.19 44.23 4.94
7.483 7203 31.78 19.56 5.34
0.002 349 0.0001 8.94 0.39
77.012 54,281 101.11 88.52 33.09
Table 5 Correlation matrix.
ln(MPC) ln(GDP) ln(CPI) URB ln(ROAD) Year a
ln(MPC)
ln(GDP)
ln(CPI)
URB
ln(ROAD)
Year
1.0000 0.4701 0.1526 0.5501 0.4655 0.1369
1.0000 0.3226 0.8592a 0.6915 0.2235
1.0000 0.3041 0.1408 0.5414
1.0000 0.5856 0.2391
1.0000 0.0512
1.0000
Correlation value > 0.8.
population. The natural logarithm of the explanatory variables are used to minimize heteroscedasticity and also allow easier interpretation of the relative elasticity value of the estimates. A quadratic term was added to the per capita GDP and the urban population percentage to account for the quadratic relationship between these two variables and the MPC ownership ratio. The CPI and the total length per thousand population interacted with the per capita GDP. This allows us to examine the effect of total road length per thousand population and CPI on the MPC ownership ratio as income increased. A time trend is included to capture the effects of other exogenous time dependent variables. Model A only includes linear and quadratic terms of per capita GDP to investigate the relationship between the MPC ownership ratio and the per capita GDP. Models B to F include the influence of additional explanatory variables on this relationship. The results
of Models A to E in Table 6 confirm a statistically significant inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP. Generally, the MPC ownership ratio increased with income at a lower level and decreased with income at a higher level. The results of Models A and B indicated that the turning point for the MPC ownership ratio was in the range of US$4895 to US$6060.4 Model G includes only data after year 1993 to examine the effect of per capita GDP on the MPC ownership ratio at a higher per capita GDP level. The average GDP per capita between 1993 and 2010 for LAEC and AEC was USD$7812 and USD$28,975, respectively. The results showed a significant U-shaped relationship between the MPC ownership ratio and the per capita GDP. Specifically, the MPC ownership ratio declined with per capita GDP at a lower per capita GDP level and increased with per capita GDP at a higher per capita GDP level.5 The turning point was reached at US$9215. This finding may be attributed to people with a higher income level using motorcycles in recreational and leisure activities, especially in AEC (Jamson and Chorlton, 2009). The estimated results showed that the coefficient for the CPI was positive and statistically significant, whereas the coefficient on its interaction with the per capita GDP was estimated to be negative. 4 The turning point is calculated by, exp(x1/(2.x2)) where x1 is the coefficient on the log of GDP and x2 is the coefficient on the log GDP, whole quantity squared. 5 Model G in Table 6 was also estimated based on data after year 1984, and similar results were found (not reported).
126
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Table 6 Analysis results.
Variables ln(GDP) (ln(GDP))2 ln(CPI) ln(CPI) ⁄ ln(GDP) ln(ROAD) ln(ROAD) ⁄ ln(GDP) URB URB2 Year Constant No. Obs. No. Groups Overall model significance Hausman test (Chi-square)
Model A
Model B
Model C
Model D
Model E
6.151*** 0.362***
6.027*** 0.346***
5.613*** 0.318*** 0.231* 0.034**
3.557*** 0.237*** 0.259* 0.037** 1.035*** 0.093**
0.0012 28.284*** 1934 80 132*** 19.98***
0.0028 18.451*** 1934 80 126*** 18.89**
5.233*** 0.297*** 0.681*** 0.085*** 0. 602* 0.051 0.069*** 0.0004*** 0.0029 35.834*** 1934 80 109*** 18.36**
27.264*** 1934 80 130*** 27.29***
0.0052** 17. 203*** 1934 80 131*** 32.34***
Model F
Model G 1.917** 0.105**
1.389*** 0.164*** 1.365*** 0.131*** 0.051*** 0.0002*** 1.847*** 1934 80 107*** 24.65***
6.703* 821 79 205*** 18.99***
Model G includes only data after year 1993. * Significant at 10%. ** Significant at 5%. *** Significant at 1%.
Several studies have shown that inflation increases the price of goods and services, thereby eroding purchasing power (Manzoor et al., 2011; Molana, 1990). Therefore, under high inflation circumstances, commuters tend to opt for vehicles with lower purchase and running costs, such as motorcycles and mopeds, and this leads to an increase in the MPC ownership ratio at a lower income level. Nonetheless, the effect of CPI on the MPC ownership ratio declined as income continued to rise. This is probably because cars become a necessity 6and are more affordable at a higher income level. The quadratic coefficient for urbanisation was positive and significant in Models E and F, pointing to a convex relationship between urbanisation and the MPC ownership ratio. At a lower urban population density level, the scale of economic activities is small and less varied in scope than at a higher urban population density. Therefore, there is less of a demand for long-distance travel, and motorcycles are preferred to passenger cars may be because they are more suited to short-distance travel. However, as the urban population density and urban area increase, long-distance travel by passenger car is expected to increase. On the other hand, a higher level of urban population growth may give rise to traffic congestion, as well as strong competition for parking spaces. Karathodorou et al. (2010) and Ng et al. (2010) indicated that traffic congestion and limited parking areas discouraged potential passenger car owners. Nishitateno and Burke (2014) indicated that some commuters may choose to ride a motorcycle to avoid traffic congestion and limited parking bays. As a result, the MPC ownership ratio increases. Models D and F showed that the total road length per thousand population had a positive and statistically significant effect on the MPC ownership ratio, whereas the coefficient for its interaction with per capita GDP was estimated to be negative. These results suggest that the MPC ownership ratio increases with total road length per thousand population at a low income level. However, this effect declined with per capita GDP growth over time. A possible explanation for this finding is that road network development was mainly limited to urban areas and low-performance roads (such as unpaved and lower number of lanes) in the early stages of road network development. At this development phase, a higher road density was associated with enhanced local accessibility. Motorcycle ownership can be expected to rise with road density at this development stage. However, as an economy grows, more
6 Individuals who own cars are more probable to be employed (Ong, 1996; Holzer et al., 1994).
extensive high-performance road networks (such as expressway) are built to connect the large cities. At this stage of road development, motorcycles can be expected to be less preferable than passenger cars, possibly because motorcycles are less safe and less comfortable than cars for longer-distance travel. Consequently, the MPC ownership ratio declines. Model B indicated that the time trend variable was negative and significant, implying that some unmeasured effects caused a downward trend in the MPC ownership ratio over time. However, the inclusion of the CPI and its interaction with the per capita GDP eliminated the significance of the time trend variable in Models C, D and E, suggesting that these variables played a role in reducing MPC ownership ratio. 5. Conclusions This study focused on the effect of economic growth on the MPC ownership ratio. The estimated results pointed to an inverted U-shaped relationship between the MPC ownership ratio and the per capita GDP. The evidence indicated that urbanisation, the total road length per thousand population, and the CPI were the underlying factors that contributed to this relationship. In view of the relationship, it could be assumed that at a low income level, motorcycle ownership is popular because the scale of the economic activities is relatively low and simple. Moreover, cars are less affordable than motorcycles at a low income level. As the level of income increases, purchasing power also rises. At this stage, more people likely switch from motorcycles to passenger cars probably for their prestige, convenience and safety. At a higher income level, more resources are also available for the government to invest in improvements of road network infrastructures. Advances in road network infrastructures are likely to lead to increased passenger car ownership. The latter is likely probably due to passenger cars being safer than motorcycles and more suitable for long-distance travel. A limitation of this study is that the models omitted other transportation modes, such as public transportation, bicycling and walking. Earlier studies have consistently pointed to a higher demand for public transportation in densely populated areas (Lam and Tam, 2002; Clark, 2007, 2009; Khan and Willumsen, 1986). This is probably due to the higher level of congestion, greater parking constraints and costs, and more organized public transportation systems in these areas (Dargay and Gately, 1999; De Jong and Van de Riet, 2008; Schwanen et al., 2004). Previous studies also showed that public transportation tended to impose lower risks on
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
passengers (Brenac and Clabaux, 2005; Elvik and Bjørnskau, 2005; Evans, 1994). In this study, we assumed that motorcycles and passenger cars were the two major private transportation modes, with commuters having to choose one mode of travel. In reality, they could select other private transportation modes, such as bicycling and walking, to meet their mobility needs. The factors that influence the use of these private transportation modes may differ from those that influence the use of motorcycles and passenger cars. Therefore, further research is needed to examine the modal shift from motorcycles to other transportation modes and to determine what factors contribute to this change. The key finding of this study is that there are complex pathways from economic growth to the change in the MPC ownership ratio, such as urbanisation, advances in road network infrastructures, and vehicle purchasing power. However, there is always a time lag between economic growth and the MPC ownership ratio. That is, economic growth may be rapid but the effects of transport policies and measures may take longer time to be felt, particularly for some newly-developed cities in LAEC. This may have implications on transport policies and strategies. Future research to address this issue is warranted and should be pursued. Many of the LAEC in the sample have per capita GDP below the turning point of the inverted U-curve for the MPC ownership ratio. This implies that these countries are still at a stage where further economic growth will result in an increase in motorcycle ownership. As reported by Law et al. (2009), growth in income at lower income levels appeared to be associated with a rise in motorcycle deaths, with the increase partly explained by the large number of motorcycles and partly by the high degree of mobility of the motorcyclists. To enhance motorcycle safety, these countries could focus on the provision of specific injury-prevention countermeasures. Such measures could include improving public transportation services and simultaneously encouraging motorcyclists to use public transportation instead of riding motorcycles, as well as designing roads with exclusive lanes for motorcycles and increasing the stringency of motorcycle helmet laws and enforcement. As indicated in the present study, the MPC ownership ratio declined once the per capita GDP level exceeded a threshold level. This result suggests that developed countries, particularly those with increased urbanisation and a higher road network density, should devote resources to accommodating the increased demand for passenger cars, without causing excessive congestion. The empirical evidence found in this study also showed that the reduction in the MPC ownership ratio after the first turning point was not permanent, with a second turning point occurring at a higher per capita GDP level. Jamson and Chorlton (2009) indicated that at higher income levels, more mature motorcycle riders take up motorcycling as a leisure pursuit rather than as a means of transport. Therefore, more emphasis should be placed on motorcycling training specifically for leisure motorcycle riders. Acknowledgement This research was funded by Universiti Putra Malaysia Research University Grant (Grant No. 05-05-10-1108RU). The funding source played absolutely no role in the study design nor any collection, analysis, or interpretation of the data. References Abdul Manan, M.M., Várhelyi, A., 2012. Motorcycle fatalities in Malaysia. IATSS Res. 36 (1), 30–39. Baltagi, B.H., 2005. Econometric Analysis of Panel Data. John Wiley and Sons Inc., New York. Bento, A.M., Cropper, M.L., Mobarak, A.M., Vinha, K., 2005. The effects of urban spatial structure on travel demand in the United States. Rev. Econ. Stat. 87 (3), 466–478.
127
Bevan, A.A., Danbolt, J., 2004. Testing for inconsistencies in the estimation of UK capital structure determinants. Appl. Financ. Econ. 14 (1), 55–66. Brenac, T., Clabaux, N., 2005. The indirect involvement of buses in traffic accident processes. Saf. Sci. 43 (10), 835–843. Button, K., Ngoe, N., Hine, J., 1993. Modelling vehicle ownership and use in low income countries. J. Transport Econ. Policy 27 (1), 57–67. Cao, X., Huang, X., 2013. City-level determinants of private car ownership in China. Asian Geogr. 30 (1), 37–53. Chiou, Y.C., Wen, C.H., Tsai, S.H., Wang, W.Y., 2009. Integrated modelling of car/motorcycle ownership, type and usage for estimating energy consumption and emissions. Transport. Res. Part A: Policy Pract. 43 (7), 665–684. Clark, S., 2007. Estimating local car ownership models. J. Transport Geogr. 15 (3), 184–197. Clark, S., 2009. Charactering and predicting car ownership using rough sets. Transport. Res. Part C: Emerg. Technol. 17 (4), 381–393. Dao, N.X., Duc, N.H., 2005. Study on the ownership of motorcycle in Vietnam. In: Proceeding the 6th, International Conference of Eastern Asia Society for Transportation Studies, Bangkok. Dargay, J., Gately, D., 1999. Income’s effect on car and vehicle ownership, worldwide: 1960–2015. Transport. Res. Part A: Policy Pract. 33 (2), 101–138. Dargay, J., Hanly, M., 2007. Volatility of car ownership, commuting mode and time in the UK. Transport. Res. Part A: Policy Pract. 41 (10), 934–948. Dargay, J.M., Vythoulkas, P.C., 1999. Estimation of a dynamic car ownership model: a pseudo-panel approach. J. Transport Econ. Policy 33 (3), 287–302. De Jong, G.C., Van de Riet, O., 2008. The driving factors of passenger transport. Eur. J. Transport Infrastruct. Res. 8 (3), 227–250. De Jong, G., Fox, J., Pieters, M., Daly, A.J., Smit, R., 2004. Comparison of car ownership models. Transport Rev. 24 (4), 379–408. Delbosc, A., 2013. Household composition and within-household car saturation in Melbourne. Transport Policy 25, 94–100. Elhorst, J.P., 2010. Spatial panel data models. In: Fischer, M.M., Getis, A. (Eds.), Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications. Springer, Berlin, pp. 377–407. Elvik, R., Bjørnskau, T., 2005. How accurately does the public perceive differences in transport risks?: an exploratory analysis of scales representing perceived risk. Accid. Anal. Prevent. 37 (6), 1005–1011. Evans, A.W., 1994. Evaluating public transport and road safety measures. Accid. Anal. Prevent. 26 (4), 411–428. Greene, W.H., 2003. Econometric Analysis. Prentice Education, Upper Saddle River, NJ. Hess, D.B., Ong, P.M., 2002. Traditional neighbourhoods and automobile ownership. Transport. Res. Rec. 1805, 35–44. Heston, A., Summers, R., Aten, B., 2011. Penn world table version 7.0. Centre for International Comparisons of Production, Income and Prices. University of Pennsylvania. Holzer, H.J., Ihlanfeldt, K.R., Sjoquist, D.L., 1994. Work, search and travel among white and black youth. J. Urban Econ. 35, 320–345. Hsiao, C., 2003. Analysis of Panel Data, second ed. Cambridge University Press, Cambridge, UK. Hsu, T.P., 2005. Comparative study on motorcycle ownership forecasting model of Asian Countries-Taiwan, Malaysia and Vietnam. In: Proceedings the 6th International Conference of Eastern Asia Society for Transpiration Studies, Bangkok. Ingram, G.K., Liu, Z., 1999. Determinants of motorization and road provision. World Bank Policy Research Working Paper No. 2042. The World Bank, Washington, D.C. Jamson, S., Chorlton, K., 2009. The changing nature of motorcycling: patterns of use and rider characteristics. Transport. Res. Part F: Traffic Psychol. Behav. 12 (4), 335–346. Karathodorou, N., Graham, D.J., Noland, R.B., 2010. Estimating the effect of urban density on fuel demand. Energy Econ. 32 (1), 86–92. Khan, A., Willumsen, L., 1986. Modelling car ownership and use in developing countries. Traffic Eng. Control 27 (11), 554–560. Lai, W.T., Lu, J.L., Chiang, Y.S., 2006. The study of multi-vehicle household mix demand model on auto–motorcycle ownership and use in Taiwan area. Transport. Plann. J. 35 (3), 309–336. Lam, W., Tam, M., 2002. Reliability of territory-wide car ownership estimates in Hong Kong. J. Transport Geogr. 10 (1), 51–60. Law, T.H., Noland, R.B., Evans, A.W., 2009. Factors associated with the relationship between motorcycle deaths and economic growth. Accid. Anal. Prevent. 41 (2), 234–240. Manzoor, M.M., Siddiqui, A.A., Sattar, A., Fahim, M., Rasheed, S., 2011. Impact of inflation on household consumption – a case of Pakistan. Int. Res. J. Finance Econ. 68, 161–167. Molana, H., 1990. A note on the effect of inflation on consumers’ expenditure. Empirical Econ. 15 (1), 1–15. Nagai, Y., Fukuda, A., Okada, Y., Hashino, Y., 2003. Two-wheeled vehicle ownership trends and issues in the Asian region. J. Eastern Asia Soc. Transport. Stud. 5, 135–146. Ng, W.S., Schipper, L., Chen, Y., 2010. China motorization trends: new directions for crowded cities. J. Transport Land Use 3 (3), 5–25. National Highway Transportation Safety Administration (NHTSA), 2008. Traffic Safety Facts 2007 Data: Motorcycles (Report No. DOT HS 810 990). NHTSA, National Centre for Statistics and Analysis, Washington, D.C. Nishitateno, S., Burke, P.J., 2014. The motorcycle Kuznets curve. J. Transport Geogr. 36, 116–123.
128
T.H. Law et al. / Journal of Transport Geography 46 (2015) 122–128
Nolan, A., 2010. A dynamic analysis of household car ownership. Transport. Res. Part A: Policy Pract. 44 (6), 446–455. Ong, P., 1996. Work and automobile ownership among welfare recipients. Soc. Work Res. 30 (4), 255–262. Pongthanaisawan, J., Sorapipatana, C., 2010. Relationship between level of economic development and motorcycle and car ownerships and their impacts on fuel consumption and greenhouse gas emission in Thailand. Renew. Sustain. Energy Rev. 14 (9), 2966–2975. Preusser, D.F., Williams, A.F., Ulmer, R.G., 1995. Analysis of fatal motorcycle crashes: crash typing. Accid. Anal. Prevent. 27 (6), 845–851. Radin Umar, R.S., Mackay, G.M., Hills, B.L., 1995. Preliminary analysis of motorcycle accidents: short-term impacts of the running headlights campaign and regulation in Malaysia. J. Traffic Med. 23, 17–28. RAND Europe and WSP, 2004. Motorcycles and Congestion: The Effect of Modal Split – Full Summary, Department for Transport (with support from Transport for London). Ranney, M.L., Mello, M.J., Baird, J.B., Chai, P.R., Clark, M.A., 2010. Correlates of motorcycle helmet use among recent graduates of a motorcycle training course. Accid. Anal. Prevent. 42 (6), 2057–2062. Riley, K., 2002. Motor vehicles in China: the impact of demographic and economic changes. Populat. Environ. 23 (5), 479–494. Romilly, P., Song, H., Liu, X., 1998. Modelling and forecasting car ownership in Britain: a cointegration and general to specific approach. J. Transport Econ. Policy 132 (2), 165–185. Sanko, N., Maesoba, H., Dissanayake, D., Yamamoto, T., Kurauchi, S., 2009. Intertemporal analysis of household car and motorcycle ownership behaviours: the case in Nagoya Metropolitan Area of Japan, 1981–2001. IATSS Res. 33 (2), 39–53. Schwanen, T., Dijst, M., Dieleman, F.M., 2004. Policies for urban form and their impact on travel: the Netherlands experience. Urban Stud. 41 (3), 579–603. Sehatzadeh, B., Noland, R.B., Weiner, M.D., 2011. Walking frequency, cars, dogs and the built environment. Transport. Res. Part A: Policy Pract. 45 (8), 741–754. Senbil, M., Zhang, J., Fujiwara, A., 2007. Motorization in Asia – 14 countries and three metropolitan areas. Int. Assoc. Traffic Saf. Sci. Res. 31 (1), 46–58.
Shao, J., Xiao, Z., Xu, R., 2011. Estimation with unbalanced panel data having covariate measurement error. J. Stat. Plann. Inference 141 (2), 800–808. Sharma, B.R., 2008. Road traffic injuries: a major global public health crisis. Public Health 122 (12), 1399–1406. Sillaparcharn, P., 2007. Vehicle ownership and trip generation modelling: a case study of Thailand. J. Int. Assoc. Traffic Saf. Sci. 31 (2), 17–26. Tanner, J.C., 1978. Long-term forecasting of vehicle ownership and road traffic. J. Roy. Stat. Soc., Ser. A 141 (1), 13–63. Tanner, J.S., 1983. International comparisons of car ownership and car usage. Transport and Road Research Laboratory Report 1070. Department of the Environment and Transport, Crowthorne, Berkshire. Tuan, V.A., 2011. Dynamic interactions between private passenger car and motorcycle ownership in Asia: a cross-country analysis. J. Eastern Asia Soc. Transport. Stud. 9, 541–556. Tuan, V.A., Shimizu, T., 2005. Modelling of household motorcycle ownership behaviour in Hanoi city. J. Eastern Asia Soc. Transport. Stud. 6, 1751–1765. World Development Indicators (WDI), 2009. World Bank World Development Indicators January 2009, ESDS International, (MIMAS) University of Manchester. Whelan, G., 2001. Methodological advances in modelling and forecasting car ownership in Great Britain. In: Paper for European Transport Conference 2001, PTRC, Cambridge. Whelan, G., 2007. Modelling car ownership in Great Britain. Transport. Res. Part A: Policy Pract. 41 (3), 205–219. Whelan, G., Wardman, M., Daly, A., 2000. Is there a limit to car ownership growth? An exploration of household saturation levels using two novel approaches. In: Paper presented at European Transport Conference 2000, PTRC, Cambridge. Wong, K.I., 2013. An analysis of car and motorcycle ownerships in Macao. Int. J. Sustain. Transport. 7 (3), 204–225. Wooldridge, J.M., 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press, Cambridge, MA. Yamamoto, T., 2009. Comparative analysis of household car, motorcycle and bicycle ownership between Osaka metropolitan area, Japan and Kuala Lumpur, Malaysia. Transportation 36 (3), 351–366.