- Email: [email protected]
The effect of parking charges and time limit to car usage and parking behaviour
Transport Policy 30 (2013) 125–131
Contents lists available at ScienceDirect
Transport Policy journal homepage: www.elsevier.com/locate/tranpol
The effect of parking charges and time limit to car usage and parking behaviour Jelena Simićević, Smiljan Vukanović, Nada Milosavljević n University of Belgrade, Faculty of Transport and Traffic Engineering, Serbia, Vojvode Stepe 305, 11000 Belgrade, Serbia
art ic l e i nf o
a b s t r a c t
Available online 28 September 2013
Parking policies are considered a powerful tool for solving parking problems as well as problems of the transportation system in general (traffic congestion, modal split, etc.). To define parking policy properly, its effects must be estimated and predicted. In this paper, based on stated preference data and using a logistic regression, a model to predict the effects of introducing or changing the parking price and time limitation was developed. The results show that parking prices affect car usage, while time limitations determine the type of parking used (on-street or off-street). A positive finding for policy makers is that users with work are more sensitive to parking measures than are other users, so parking measures can be used to manage user categories. Although there is a concern that parking policy can jeopardise the attractiveness and efficiency of a zone, the results show that a very small number of users would give up travelling into the zone. & 2013 Elsevier Ltd. All rights reserved.
Keywords: Parking charge Time-limited parking Stated preferences Multinomial logit model Direct Effects
1. Introduction The main objective of parking management is to balance the parking supply with the parking demand. However, parking policy has a strong impact not only on the operation of the parking subsystem but also on the entire transportation system and the city in general. Possible driver responses to parking policy (primarily to the parking charge and time limitation) are complex and varied. These include a change in the parking type, parking location, transportation mode, car occupancy, destination, travel frequency, travel time (with possible consequences on the parking duration) and route (Scholefield et al., 1997). This mechanism of influence allows parking policy to be used to achieve objectives beyond this subsystem. For example, studies have shown that the most important factor in reducing car usage is the parking price (Higgins, 1992). Thus, parking policy can be the most effective policy for achieving the desired modal split (Victorian Competition and Efficiency Commission (VCEC), 2006). Furthermore, the parking charge is considered to be the second best measure for solving traffic congestion after congestion charging (Albert and Mahalel, 2006; Kelly and Clinch, 2006), but it is used far more often because of its relatively simple implementation (Marsden, 2006; Verhoef et al., 1995).
n
Corresponding author. Tel.: þ 381 63 602142; fax: þ 381 11 2468120. E-mail addresses: [email protected] (J. Simićević), [email protected] (S. Vukanović), [email protected] (N. Milosavljević). 0967-070X/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tranpol.2013.09.007
Although good parking policy has many positive implications for sustainable transportation, poor parking policy can have the opposite effect. For example, analysing 16 studies from 11 international cities showed that approximately 30% of the traffic volume are vehicles cruising for parking, i.e., result of poor parking management (Shoup, 2005). In addition, recently, there is concern that parking policy could negatively impact the competitiveness and business efficiency in an area (D’Acierno et al., 2006). To properly set the parking policy and define the appropriate measures, i.e., to ensure that the objectives are met without adverse impact on the transportation system and other systems of a city, the effects of the policy must be predicted. Originally, models for the prediction of parking policy impacts were aggregate, i.e., based on group behaviour. Conventional models were later replaced by disaggregate models because it was recognised that the individual impact must be examined and included (Kelly and Clinch, 2006). The user response to time limitation can be relatively easily predicted; it depends on the parking duration and the possibility of shortening the duration (which is associated with trip purpose (Transit Cooperative Research Programme (TCRP), 2005)). However, the prediction of the user response to the parking price is very complex and not accurately known. It is particularly complex to determine the impact of several measures (the time limitation and parking price of on-street parking and the parking price of off-street parking) because of their synergistic effect. For this reason, it is not appropriate to estimate the impacts of the measures individually; instead, this should be performed simultaneously (see for example Ibeas et al., 2011).
126
J. Simićević et al. / Transport Policy 30 (2013) 125–131
To investigate the parameters of significance for parking decision making, a logistic regression of the stated preference data is usually conducted (Hess, 2001; Shiftan and Burd-Eden, 2001; Tsamboulas, 2001; Shiftan and Golani, 2005; Albert and Mahalel, 2006; Khodaii et al., 2010; Simićević et al., 2012a). Thus, some of the socio-economic and trip characteristics significant for decision making are identified. Lately, more and more researchers are interested in this topic. In a review paper on parking policy, Marsden (2006) noted that “We do not understand nearly enough about how individuals respond to parking policy interventions nor how these responses interact with local circumstances, the availability of alternative transport modes or alternative destinations,” and among parking topics that require further research, he highlighted “the importance of out-of-vehicle costs and in particular walk-times on parking behaviour”. This paper is testing the hypotheses: (1) that time limit and parking price influence parking behaviour (2) that these influences differ for specific user subgroups (depending on certain characteristics of users and trips). The aim of this paper is to forecast user behaviour in the conditions of change in price and/or time limit of parking, and further to forecast direct effects of such measures. The forecast was made by using multinomial logit model fitted with the data gathered by approaches of revealed and stated preferences. The structure of the paper is as follows: in Section 2, the state of parking in the central area of Belgrade is described. In Section 3, the problem this paper addresses is presented, and the procedure for its solution is described. To test this procedure, necessary data were gathered as shown in Section 4, while the collected data are presented in Section 6. In the 7th and final section, final considerations are summarised.
The period of regime validity is every day from 7 a.m. to 9 p.m. and on Saturdays from 7 a.m. to 2 p.m. Residents and businesses in the area are entitled to a parking permit (PP), which does not guarantee a vacant parking space to its holder; however, once the user finds a vacant parking space, the user can park there without any time limitation. The price of PP for residents is 480 RSD per month (5.05 EUR), while the price for businesses depends on the zone the company is located in (9,130 RSD (86.95 EUR), 6,176 RSD (58,82 EUR) and 4,106 RSD (39.10 EUR) per month, respectively). Disabled persons can park at specially marked parking spaces (3% of the total number of parking spaces (Milosavljević et al., 2009)), which the parking regime does not take into account. On-street parking spaces can be reserved for state institutions, city institutions, public services, diplomatic and other foreign representatives, businesses and entrepreneurs. The city administration approves reservations based on previously prescribed conditions. Approximately 10% of the total number of parking spaces in the central area is reserved (Milosavljević et al., 2009). Parking at parking lots and garages is charged every day for 24 h. The parking price varies from facility to facility and is 2–3 times higher than the on-street parking price. In addition to paying for parking per commenced hour, it is possible to pay for parking per month (types and fees vary among facilities, but they are also far higher than those for on-street parking). During increased area attractiveness, all on-street parking spaces are occupied, and even illegal parking occurs, making it hard to find a vacant parking space. On the other hand, off-street car parks and parking garages are never 100% occupied, and it can be assumed that at every moment, a vacant parking space can be found.
3. Problem statement and proposed solution 2. Parking in Belgrade Belgrade is the capital of Serbia. The urban part of the city occupies an area of about 77,000 ha and has approximately 1.5 million inhabitants. About 96,000 inhabitants live in the city centre, which has an approximate area of 440 ha. Based on the traffic survey, the inhabitants of Belgrade make approximately three million trips per day. In the modal split, passenger cars account for 22% and public transport for 52% of all daily trips. Coverage with the public transport network is approximately 2.1 km/km2; headways are between 6 and 20 min. Public transport users assess the quality of service as very good (mark near 4 out of a maximum of 5) (Jović and Djorić, 2009). The parking problem in Belgrade is present in almost all its urban area. The parking problem arises as a result of the disproportion between the parking demand and the number of available parking spaces. The disproportion is a result of historically formed city structures, flows or omissions in the planning and a lack of good parking supply management. The basic characteristic of Belgrade is the insufficient off-street public parking capacity, so the majority of vehicles are parked on the streets. All parking spaces are owned by and under the jurisdiction of the city administration, which is, therefore, responsible for making the appropriate parking policy. A restrictive parking element is implemented within the central area for on-street parking spaces. The area is divided into three zones (red, yellow and green) that differ in the following regime attributes: time limitation (1, 2 and 3 h, respectively) and parking price (56 RSD (0.53 EUR), 38 RSD (0.36 EUR) and 31 RSD (0.30 EUR) per commenced hour, respectively). Most visitors pay parking by mobile phone, although other technologies are in use, such as parking metres, parking tickets and electronic tickets.
As already mentioned, the aim of this paper is to predict the effects of the introduction or change of the parking policy, namely, time limitation and the parking price. These policies refer only to users who park at public, nonreserved parking spaces and who are subject to the parking regime (so-called visitors). It is therefore necessary, in the first step, to isolate this demand and use it for the study. To survey visitors' responses to parking policy changes, the stated preference method is used, where through interviews, various hypothetical situations are presented to the respondents, and they declare how they would behave in such situations. This implies that there are scenarios with different combinations of the time limitation and parking price levels. If the number of all possible scenarios (full factorial) is so great that it is impractical to examine them all, it is possible to examine only a subset, chosen to represent a full set with certainty (fractional factorial) (Hensher and King, 2001; Wong, 2006). The possible choices for visitors are as follows: no change in behaviour, driving to the zone and changing the type of parking space (on-street instead of offstreet or vice versa), parking at the fringe of the zone and continuing the journey by public transport or on foot, using other transport modes, and so on. To better explain the response visitors faced with parking policy changes, it is necessary to gather a wide set of parameters that are assumed to have a strong impact on the travel decision. These are some socio-economic characteristics of the visitors and trip characteristics. All data will be used to fit the multinomial logit (MNL) model, which will predict the probability of choosing each of the stated alternatives for each visitor for any parking price and time limitation. At the end of the modelling process, it is necessary to
J. Simićević et al. / Transport Policy 30 (2013) 125–131
aggregate the demand in order to analyse the results and the effects on the general level (Train, 2002).
4. Survey methodology The proposed procedure will be tested in the Belgrade central area in the red zone for parking. At 811 non-reserved on-street parking spaces in this zone, parking is time-limited to one hour and costs 56 RSD (0.53 EUR). In the zone, there are two parking garages and one off-street parking lot, with a total of 1145 parking spaces. The charge of parking is 75 RSD (0.71 EUR) for the first hour and 90 RSD (0.86 EUR) for each additional hour. Eight street sections were selected to represent on-street parking, while the parking garage Obilićev Venac was chosen to represent off-street parking (Fig. 1). The locations were chosen in such way that based on the previous studies of parking and users characteristics, as well as on the basis of land use in their neighbourhood it can be deemed with certainty that they represent the entire zone. The survey was conducted by face-to-face interviews of users who had just parked or who were leaving the parking space. After a pilot study, which confirmed that both the interviewers and the respondents understood the questions and which assisted in defining the levels of the parking price and time limitation to be presented in the interview, the survey was carried out during 5 days in November and December 2011. The time of the survey was the validity period of the regime (from 7 a.m. to 9 p.m.). On average, each interview lasted 5 min. The interview was conducted by students of the University of Belgrade, Faculty of Transport and Traffic Engineering. Great attention was paid to the interviewers' training and their control in the field. Interviewers were carefully trained on how to present themselves and how to interview. Before the interview, their task was to familiarise the respondent with the research. They emphasised that the research was conducted by the Faculty of Transport and Traffic Engineering, not by the city administration, for scientific purposes. Thus, respondents could not “see through” the intension of the city administration to increase the parking price
Fig. 1. Study area.
127
and could therefore give inaccurate responses (Kelly and Clinch, 2006). Furthermore, an effort was made to overcome the tendency of the respondents to detect and confirm interviewer attitudes, which is considered harmful to the stated preference technique (Tsamboulas, 2001). In total, 438 users were interviewed. For aggregation, the on-street and off-street parking volume is necessary. These data are not surveyed but taken from the study “Parking management strategy”(Milosavljević et al., 2009).
5. The collected data The first question in the interview defined the user category (visitor or other). This question could be eliminated because the further procedure applies only to the category of visitors. Based on theoretical expectations and previous experience, the parameters that were expected to influence the visitors' choice were selected (Tsamboulas, 2001; Coppola, 2002; Shiftan and Golani, 2005; Kelly and Clinch, 2006; Van der Waerden et al., 2006; Khodaii et al., 2010, Simićević et al., 2012a). The following parameters were surveyed: age, gender, origin and destination, car occupancy, parking purpose, parking frequency, parking duration, parking search time, whether the employer pays for parking and whether the respondent is car dependent. Previous studies have shown that respondents often refuse to answer about income or give an incorrect answer (Shiftan and Burd-Eden, 2001). Therefore, in this paper, instead of income, two proxies were surveyed: age of the car and engine size. One of the potential parameters is whether a visitor uses on- or off-street parking. This parameter is considered because different measures are implemented for these types of parking. On the other hand, the visitor's previous choice should depict the visitor's preference towards a certain type of parking (on- or off-street). The second part of the interview was hypothetical scenarios, i.e., different combinations of time limitation and parking price levels. Although the on-street parking price is currently lower than the off-street parking price, in the scenarios, these prices are equal. The reasons why we opt for this restriction are as follows: Because of the negative effects of on-street parking spaces and, in particular, the parking search on the transportation system and environment, the on-street parking price should be higher than or equal to the off-street parking price. However, currently in Belgrade, the situation is the opposite, leading to underutilised off-street parking and over utilised on-street parking. Therefore, we decided for this initial step to set the prices to be equal. An additional reason is the small number of off-street parking spaces in the total parking supply (Section 2). This paper examines the mitigation and tightening of parking measures. Based on the results of the pilot study the values of time limit were chosen, as well as the prices for which visitors' responses will be studied. Five prices in the range of 30 RSD (0.29 EUR) to 200 RSD (1.90 EUR) were taken, and three values of the time limitation, 30, 60 and 120 min, were taken. All possible combinations of the stated values of regime attributes (full factorial) were studied, i.e. in total 15 scenarios (5 3) were tested. However, visitors' responses in certain scenarios are obvious (can be predicted with high probability). These are scenarios in which the regime attributes of the space where the visitor is parked are unchanged or mitigated, and the attributes of another type of parking are unchanged or even tightened. To reduce the number of scenarios which were investigated, we decided not to investigate these “known” scenarios but to assume their outcome. Thus, four scenarios were eliminated (for on-street and off-street parking interviews). That is, we had 11 scenarios to
128
J. Simićević et al. / Transport Policy 30 (2013) 125–131
investigate. Because it is unreasonable to present such a large number of scenarios to one respondent (Hensher and King, 2001) because of the lapse of concentration and the possibility of withdrawal from the interview, we conducted three types of interviews containing three or four scenarios that were randomly selected, and one or two “known” scenarios were added later. The order of the presented scenarios was such that the parking price increased gradually (Kelly and Clinch, 2006) as visitors slowly considered the threshold to which they were willing to pay for parking. It would be wrong to suddenly present a situation in which parking is up to 250% more expensive because this situation is difficult to conceive, and the stated response would be questionable (Hensher and King, 2001). An example of a combination of scenarios is shown in Fig. 2.
6. Results 6.1. MNL model The gathered data were used to fit the MNL model. Respondents had four alternatives and the possibility to write down an alternative if it was not listed (“other”), as shown in Fig. 2. However, because of the small number of choices of some alternatives and because a large number of dependent variable categories make modelling much more complicated, alternatives are grouped into the following three categories: (1) on-street parking, (2) off-street parking or (3) not coming to the zone by car. The probabilities to choose these alternatives are marked with P1, P2 and P3, respectively. By definition, the sum of these three probabilities is equal to one (Hess, 2001): P 1 þ P 2 þP 3 ¼ 1
ð1Þ
The adjusted model is given by two equations: log
P1 ¼ αa þ β1a x1 þ ::: þ βia xi P3
ð2Þ
log
P2 ¼ αb þ β1b x1 þ ::: þ βib xi P3
ð3Þ
where α – intercepts, x – independent variables relevant to the choice, β –parameter estimates. The independent variables were selected from the set of surveyed parameters. Their parameter estimates were estimated by the maximum likelihood method. The final model included five independent variables. In addition to the parking price and time limitation, there are car dependency, parking purpose and current choice. It should be noted that Wald statistics revealed the significance of some other variables, namely, whether the employer pays for parking, length of the trip and income proxies (age of the car and engine size). However, their inclusion would not significantly contribute to the model adjustment to the observed data. For this reason and because of the intension to create a realistic model that does not require too much data (Ortuzar and Willumsen, 2000), these variables were not included. The fitted model is shown in Table 1. For model presentation, the following characteristics are selected: variable name, parameter estimates (β), standard error, significance and exp(β) (Field, 2005). The test of the full model compared to the intercept-only model is statistically significant, indicating that the set of independent variables reliably distinguish between the choices of onstreet parking, off-street parking and not coming to the zone by car (model χ2 ¼1047; df¼ 10; p o0.000). Nagelkerke's R2 of 0.60
19a. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 30 RSD/h time limitation (on-street): ½ hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19b. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 100 RSD/h time limitation (on-street): 2 hours YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19c. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 150 RSD/h time limitation (on-street): ½ hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19d. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 200 RSD/h time limitation (on-street): 1 hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
Fig. 2. Example of SP scenarios.
Table 1 MNL model results. Alternative
On-street parking
Variable
Parameter estimate (β) (Std. error)
Sig.
Exp(β)
Parameter estimate (β) (Std. error)
Sig.
Exp(β)
Intercept Car dependency Purpose work Park on-street Parking price (RSD/h) Time limitation (min.)
0.208 1.758 0.662 2.244 0.028 0.020
0.611 0.000 0.014 0.000 0.000 0.000
5.801 0.516 9.434 0.972 1.020
3.773 1.085 0.439 1.604 0.020 0.003
0.000 0.000 0.049 0.000 0.000 0.190
2.960 0.644 0.201 0.980 0.997
Number of observations ¼1407.
(0.409) (0.236) (0.270) (0.243) (0.002) (0.003)
Off-street parking
(0.346) (0.203) (0.223) (0.185) (0.002) (0.002)
J. Simićević et al. / Transport Policy 30 (2013) 125–131
On-street parking
129
Off-street parking
Parking demand (vehicles)
Not coming to zone by car 700 600 500 400 300 200 100 0 30
70
110 150 Parking price (RSD/h)
On-street parking
190
230
Off-street parking
Parking demand (vehicles)
Not coming to zone by car 800 700 600 500 400 300 200 100 0 30
60 90 Time limitation (RSD/h)
120
Fig. 3. The model based effects of (a) the parking price and (b) time limitation on the parking demand.
indicates that the independent variables explained the most variation of the dependent variable. The likelihood ratio index is equal to 0.65 and shows good performance of the model. The likelihood ratio tests showed that all included independent variables contribute significantly to the prediction. Category “not coming to zone by car” is redundant. Generally, the effects of the independent variables on the dependant variables are logical and expected. Car dependency is determined by the question “Do you use a car for every trip in the city?”, where a positive answer determines car dependent visitor. The results show that car dependent visitors are more likely to park in the zone then to give up driving to the zone compared to visitors who are not car dependent. The next variable is also dummy equal to 1 if the purpose of parking is work and is 0 otherwise. The reason for this division is that visitors with purposeful work, because of their travel time and long parking duration, are unwanted in central areas. Unlike them, visitors with other purposes (business, shopping, leisure…) are essential to the vitality and attractiveness of the area. It should be noted that despite a 1 h time limitation, work is present at red zone on-street parking spaces. The reason for this is the change in its character, i.e., these are employees who need to use cars during working hours. On the other hand, a longer duration can be ensured by system abuse; for example, one can pay for parking
more than once but in different ways to avoid the prescribed time limitation (Simićević et al., 2012b). Parameter estimates (Table 1) reveal that visitors with work are more sensitive to a parking policy change, and it is more likely that these visitors will give up parking in the zone. A visitor's previous choice depicts the visitor's preference to a specific parking type (on- or off-street), so, for example, visitors who use on-street parking are more likely to choose this type of parking in the future compared to those who currently use offstreet parking. As the parking price in the zone increases, the odds of parking in the zone decreases. With every RSD of increase, the odds of parking on-street decrease by 0.97 and of parking off-street decrease by 0.98. As the on-street parking time limitation increases, the odds of parking on-street increase. With every 1 min, the odds increase by 1.02. On the other hand, the odds of parking off-street decrease (by 0.997). In case that the visitors, due to strict parking measures in the zone, decide not to park in such zone (3rd alternative), the ways in which they would come into the city centre are investigated. The results show that in such case the majority of visitors would shift to public transport (51%) or would park on the zone fringe (27%). Not a single visitor stated that he/she would choose car pool.
J. Simićević et al. / Transport Policy 30 (2013) 125–131
800 600 400 200 90
0 30
70
800 600 400 200 120 90
0 30
70
110 150 g pric e (RS D/h)
Parkin
Parking demand (vehicles)
30 230
60 30
190
Time limitation (min)
Parking demand (vehicles)
110 150 190 Parking price (RSD/h)
Time limitation (min)
1000
230
600 500 400 300 200 100 0
120 90 60 30
70
110
30
150
Parking price (RSD/h
190
Time limitation (min)
Parking demand (vehicles)
130
230
)
Fig. 4. The model based effects of parking policies on the parking demand. (a) On-street parking, (b) Off-street parking and (c) Not parking in the zone.
The relations shown in the figures are logical and have already been depicted through parameter estimates (Table 1). Increasing the parking price decreases the demand in on-street and off-street parking, while the share of visitors who will give up driving to the red zone increases. For a parking price up to approximately 110 RSD (1.05 EUR), the off-street parking demand curve is completely horizontal, showing visitors' propensity to a parking price increase. This is not surprising because of the existing off-street parking price. After this threshold, the curve has a similar (although somewhat lower) decline as for the on-street parking curve. Therefore, 110 RSD (1.05 EUR) can be considered the threshold for off-street parking. Unlike the parking price, the time limitation has no significant effect on the abandonment of the car in the area, although there is a logical trend. The reason for this is primarily because of the time limitation in the subject area (1 h), and therefore, only short-term visitors mostly use on-street parking. On the other hand, visitors always have an alternative, off-street parking, where parking is not time limited. Therefore, this is proof that visitors, if they do not fit the time limitation, would rather use off-street parking than give up coming to the zone by car. However, although it is proven that this attribute has little effect on the amount of parking demand in the zone, it affects the parking demand redistribution by the type of parking (on- and offstreet parking). By tightening the time limitation, the share of visitors who will park off-street (where there is no time limitation) increases. This number also increases because of the great share of on-street parking visitors with work, such as business and personal business (78%), who cannot shorten the parking duration to adopt to the new situation (TCRP, 2005). Furthermore, by mitigating this attribute, some visitors who currently use off-street parking because of the time limitation will switch to on-street parking. This is proven by the opposite signs of the parameter estimates (Table 1). Fig. 4 shows that the parking price and time limitation have great influence on visitors' behaviour and, therefore, on the parking demand. For the most stringent tested scenario (price of 200 RSD/h and time limitation of 30 min), 54% of visitors would give up coming to the zone by car. It should be noted that a complete picture can be obtained by a comprehensive survey (interview of households), which would examine the generation of new demand. In this paper, because of financial constraints, it is not done.
6.2. Demand prediction 7. Conclusion The fitted MNL model is further used to calculate the probability of the selection of alternatives, i.e., determining the visitors' sensitivity to parking policies. This is done by solving the MNL equations for probability using a wide range of parking price and time limitation values and fixing the values of the other variables. After calculating the probabilities for each visitor individually, the results are aggregated. The sample consists of on-street and off-street parking visitors. As in the current situation, different regimes and parking prices are implemented, and visitors' sensitivities are different. Therefore, because of a driver's preference towards a previous choice, aggregation is performed by dividing the sample into two segments, visitors who currently park on-street and visitors who currently park off-street. Within each segment, the probabilities of choosing each alternative are summed, and the results are extrapolated to the level of the daily parking volume. The results for the segments are then summed. The individual effects of the parking price and time limitation change are shown in Fig. 3(a and b), respectively, and the impacts of both policies at the same time are presented in Fig. 4.
Impacts of parking price and time limit on behaviour of users and on car use are studied in this paper. The setup hypotheses are proven: parking regime attributes influence parking behaviour and these influences differ for specific user subgroups. In this regard, the MNL model was fitted to predict the effects of introducing or changing the parking price and time limitation levels. The parameters included in the model are, in addition to the parking price and time limitation, car dependency, parking purpose and previous choice. All relations of the independent variables to the dependant variable are logical and expected. All statistics show that the model is very good at fitting the data. The model results confirm that the parking price can affect the amount of parking demand and, therefore, parking utilisation. The impact on car use can be used to fulfil some sustainable transportation objectives such as addressing/mitigating traffic congestion problems, achieving the desired modal split, etc. Tightening time limits in areas where limits already exist does not lead to a significant reduction in the parking demand. However, it enables the parking demand to be managed according to
J. Simićević et al. / Transport Policy 30 (2013) 125–131
the parking type. This can be desirable, among other reasons, to reduce the parking search time, which is a consequence of unevenly utilised types of parking. A typical example of this is the Belgrade red zone (and entire central area), where a vacant offstreet parking space can be found at any time, but visitors spend approximately 342 h daily searching for a vacant on-street parking space (here are included visitors who, when failing to find a vacant on-street parking space, eventually park off-street). Visitors with work are much more sensitive to parking policy interventions than are others. This is a very desirable finding for policy makers because these visitors are unwanted in the city central areas. This enables managing user categories. Finally it should be noted that in this paper, we opted to group alternatives in this way, but depending on research objectives, grouping can be done differently. For example, a policy maker may be interested in what visitors who give up parking in the zone would do. In this example, the most common responses of visitors who would give up parking in the zone are using public transport (51%) and parking at the fringe of the zone (27%). This response leads to the need to monitor the public transport quality of the service and state of parking at the fringe of the zone when changing parking policy. Only 2% of visitors who give up would change the trip destination, which means that parking policies will not significantly jeopardise zone attractiveness and efficiency. References Albert, G., Mahalel, D., 2006. Congestion tolls and parking fees: a comparison of the potential effect on travel behaviour. Transport Policy 13, 496–502. Coppola, P., 2002. A joint model of mode/parking choice with elastic parking demand, Transportation Planning. Kluwer Academic Publishers, Netherlands. D'Acierno, L., Gallo, M., Montella, B., 2006. Optimisation models for the urban parking pricing problem. Transport Policy 13, 34–48. Field, A., 2005. Discovering Statistic Using SPSS. Sage Publication, London. Hensher, D.A., King, J., 2001. Parking demand and responsiveness to supply, pricing and location in the Sydney central business district. Transportation Research Part A 35, 177–196. Hess, D.B., 2001. Effect of free parking on commuter mode choice: Evidence form travel diary data. Transportation Research Record 1753, 35–42. Higgins, D., 1992. Parking taxes: effectiveness, legality and implementation, some general considerations. Transportation 19 (3), 221–230.
131
Ibeas, A., Cordera, R., Dell'Olio, L., Moura, J.L., 2011. Modelling demand in Restricted Parking Zones. Transportation Research Part A: Policy and Practice 45, 485–498. Jović, J., Djorić, V., 2009. Application of transport demand modelling in pollution estimation of a street network. Thermal Science 13 (3), 229–243. Kelly, J.A., Clinch, J.P., 2006. Influence of varied parking tariffs on parking occupancy levels by trip purpose. Transport Policy 16, 487–495. Khodaii, A., Aflaki, E., Moradkhani, A., 2010. Modelling the effect of parking fare on personal car use. Transaction A: Civil Engineering 17 (3), 209–216. Marsden, G., 2006. Evidence base for parking policies—a review. Transport Policy 13, 447–457. Milosavljević, N., Simićević, J., Maletić, G., et al., 2009. Parking Management Strategy [Prilozistrategijiupravljanjaparkiranjem]. Institute of the Faculty of Transport and Traffic Engineering, Belgrade. Ortuzar, J. de D., Willumsen, L.G., 2000. Modelling Transport. John Wiley and sons, UK. Scholefield, G., Bradley, R., Skinner, A., 1997. Study of parking and traffic demand: a traffic restraint analysus model (TRAM). In: Transportation Planning Methods: proceedings of Seminar E held at the European Transport Forum Annual Meeting, Brunel University, England, 1–5 September 1997, vol. p 414. Shiftan, Y., Burd-Eden, R., 2001. Modelling the response to parking policy. Transportation Research Record 1765, 27–34. Shiftan, Y., Golani, A., 2005. Effect of auto restraint policies on travel behaviour. Transportation Research Record: Journal of the Transportation Research Board, 1932, Transportation Research Board of the National Academies, Washington, D.C., pp. 156–163. Shoup, D., 2005. The High Cost of Free Parking. Planers and Press, Chicago. Simićević, J., Milosavljević, N., Maletić, G., 2012a. Influence of parking price on parking garage users’ behaviour. Promet—Traffic&Transportation 24 (5), 413–423. Simićević, J., Milosavljević, N., Maletić, G., Kaplanović, S., 2012b. Defining parking price based on users' attitudes. Transport Policy 23, 70–78. Train, K., 2002. Discrete Choice Methods with Simulation. CambridgeUniversity Press. Transit Cooperative Research Program, 2005. Traveler Response to Transportation System Changes. Transportation Research Board, Washington DC. Tsamboulas, D., 2001. Parking fare thresholds: a policy tool. Transport Policy 8, 115–124. Van der Waerden, P., Borgers, A., Timmermans, H., 2006. Attitudes and behavioral responses to parking measures. European Journal of Transport and Infrastructure Research 6 (4), 301–312. Verhoef, E., Nijkamp, P., Rietveld, P., 1995. The economics of regulatory parking policies: The (im)possibilities of parking policies in traffic regulation. Transportation Research Part A 29 (2), 141–156. Victorian Competition and Efficiency Commission, 2006. International Approaches to Tackling Transport Congestion’, Paper 2 (Final): Parking Restraint Measures, Booz Allen Hamilton. Wong, S.T., 2006. Disaggregate Analyses of Stated Preference Data for Capturing Parking Choice Behaviour. Department of Civil Engineering, The University of Hong Kong. (Master thesis).
Contents lists available at ScienceDirect
Transport Policy journal homepage: www.elsevier.com/locate/tranpol
The effect of parking charges and time limit to car usage and parking behaviour Jelena Simićević, Smiljan Vukanović, Nada Milosavljević n University of Belgrade, Faculty of Transport and Traffic Engineering, Serbia, Vojvode Stepe 305, 11000 Belgrade, Serbia
art ic l e i nf o
a b s t r a c t
Available online 28 September 2013
Parking policies are considered a powerful tool for solving parking problems as well as problems of the transportation system in general (traffic congestion, modal split, etc.). To define parking policy properly, its effects must be estimated and predicted. In this paper, based on stated preference data and using a logistic regression, a model to predict the effects of introducing or changing the parking price and time limitation was developed. The results show that parking prices affect car usage, while time limitations determine the type of parking used (on-street or off-street). A positive finding for policy makers is that users with work are more sensitive to parking measures than are other users, so parking measures can be used to manage user categories. Although there is a concern that parking policy can jeopardise the attractiveness and efficiency of a zone, the results show that a very small number of users would give up travelling into the zone. & 2013 Elsevier Ltd. All rights reserved.
Keywords: Parking charge Time-limited parking Stated preferences Multinomial logit model Direct Effects
1. Introduction The main objective of parking management is to balance the parking supply with the parking demand. However, parking policy has a strong impact not only on the operation of the parking subsystem but also on the entire transportation system and the city in general. Possible driver responses to parking policy (primarily to the parking charge and time limitation) are complex and varied. These include a change in the parking type, parking location, transportation mode, car occupancy, destination, travel frequency, travel time (with possible consequences on the parking duration) and route (Scholefield et al., 1997). This mechanism of influence allows parking policy to be used to achieve objectives beyond this subsystem. For example, studies have shown that the most important factor in reducing car usage is the parking price (Higgins, 1992). Thus, parking policy can be the most effective policy for achieving the desired modal split (Victorian Competition and Efficiency Commission (VCEC), 2006). Furthermore, the parking charge is considered to be the second best measure for solving traffic congestion after congestion charging (Albert and Mahalel, 2006; Kelly and Clinch, 2006), but it is used far more often because of its relatively simple implementation (Marsden, 2006; Verhoef et al., 1995).
n
Corresponding author. Tel.: þ 381 63 602142; fax: þ 381 11 2468120. E-mail addresses: [email protected] (J. Simićević), [email protected] (S. Vukanović), [email protected] (N. Milosavljević). 0967-070X/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tranpol.2013.09.007
Although good parking policy has many positive implications for sustainable transportation, poor parking policy can have the opposite effect. For example, analysing 16 studies from 11 international cities showed that approximately 30% of the traffic volume are vehicles cruising for parking, i.e., result of poor parking management (Shoup, 2005). In addition, recently, there is concern that parking policy could negatively impact the competitiveness and business efficiency in an area (D’Acierno et al., 2006). To properly set the parking policy and define the appropriate measures, i.e., to ensure that the objectives are met without adverse impact on the transportation system and other systems of a city, the effects of the policy must be predicted. Originally, models for the prediction of parking policy impacts were aggregate, i.e., based on group behaviour. Conventional models were later replaced by disaggregate models because it was recognised that the individual impact must be examined and included (Kelly and Clinch, 2006). The user response to time limitation can be relatively easily predicted; it depends on the parking duration and the possibility of shortening the duration (which is associated with trip purpose (Transit Cooperative Research Programme (TCRP), 2005)). However, the prediction of the user response to the parking price is very complex and not accurately known. It is particularly complex to determine the impact of several measures (the time limitation and parking price of on-street parking and the parking price of off-street parking) because of their synergistic effect. For this reason, it is not appropriate to estimate the impacts of the measures individually; instead, this should be performed simultaneously (see for example Ibeas et al., 2011).
126
J. Simićević et al. / Transport Policy 30 (2013) 125–131
To investigate the parameters of significance for parking decision making, a logistic regression of the stated preference data is usually conducted (Hess, 2001; Shiftan and Burd-Eden, 2001; Tsamboulas, 2001; Shiftan and Golani, 2005; Albert and Mahalel, 2006; Khodaii et al., 2010; Simićević et al., 2012a). Thus, some of the socio-economic and trip characteristics significant for decision making are identified. Lately, more and more researchers are interested in this topic. In a review paper on parking policy, Marsden (2006) noted that “We do not understand nearly enough about how individuals respond to parking policy interventions nor how these responses interact with local circumstances, the availability of alternative transport modes or alternative destinations,” and among parking topics that require further research, he highlighted “the importance of out-of-vehicle costs and in particular walk-times on parking behaviour”. This paper is testing the hypotheses: (1) that time limit and parking price influence parking behaviour (2) that these influences differ for specific user subgroups (depending on certain characteristics of users and trips). The aim of this paper is to forecast user behaviour in the conditions of change in price and/or time limit of parking, and further to forecast direct effects of such measures. The forecast was made by using multinomial logit model fitted with the data gathered by approaches of revealed and stated preferences. The structure of the paper is as follows: in Section 2, the state of parking in the central area of Belgrade is described. In Section 3, the problem this paper addresses is presented, and the procedure for its solution is described. To test this procedure, necessary data were gathered as shown in Section 4, while the collected data are presented in Section 6. In the 7th and final section, final considerations are summarised.
The period of regime validity is every day from 7 a.m. to 9 p.m. and on Saturdays from 7 a.m. to 2 p.m. Residents and businesses in the area are entitled to a parking permit (PP), which does not guarantee a vacant parking space to its holder; however, once the user finds a vacant parking space, the user can park there without any time limitation. The price of PP for residents is 480 RSD per month (5.05 EUR), while the price for businesses depends on the zone the company is located in (9,130 RSD (86.95 EUR), 6,176 RSD (58,82 EUR) and 4,106 RSD (39.10 EUR) per month, respectively). Disabled persons can park at specially marked parking spaces (3% of the total number of parking spaces (Milosavljević et al., 2009)), which the parking regime does not take into account. On-street parking spaces can be reserved for state institutions, city institutions, public services, diplomatic and other foreign representatives, businesses and entrepreneurs. The city administration approves reservations based on previously prescribed conditions. Approximately 10% of the total number of parking spaces in the central area is reserved (Milosavljević et al., 2009). Parking at parking lots and garages is charged every day for 24 h. The parking price varies from facility to facility and is 2–3 times higher than the on-street parking price. In addition to paying for parking per commenced hour, it is possible to pay for parking per month (types and fees vary among facilities, but they are also far higher than those for on-street parking). During increased area attractiveness, all on-street parking spaces are occupied, and even illegal parking occurs, making it hard to find a vacant parking space. On the other hand, off-street car parks and parking garages are never 100% occupied, and it can be assumed that at every moment, a vacant parking space can be found.
3. Problem statement and proposed solution 2. Parking in Belgrade Belgrade is the capital of Serbia. The urban part of the city occupies an area of about 77,000 ha and has approximately 1.5 million inhabitants. About 96,000 inhabitants live in the city centre, which has an approximate area of 440 ha. Based on the traffic survey, the inhabitants of Belgrade make approximately three million trips per day. In the modal split, passenger cars account for 22% and public transport for 52% of all daily trips. Coverage with the public transport network is approximately 2.1 km/km2; headways are between 6 and 20 min. Public transport users assess the quality of service as very good (mark near 4 out of a maximum of 5) (Jović and Djorić, 2009). The parking problem in Belgrade is present in almost all its urban area. The parking problem arises as a result of the disproportion between the parking demand and the number of available parking spaces. The disproportion is a result of historically formed city structures, flows or omissions in the planning and a lack of good parking supply management. The basic characteristic of Belgrade is the insufficient off-street public parking capacity, so the majority of vehicles are parked on the streets. All parking spaces are owned by and under the jurisdiction of the city administration, which is, therefore, responsible for making the appropriate parking policy. A restrictive parking element is implemented within the central area for on-street parking spaces. The area is divided into three zones (red, yellow and green) that differ in the following regime attributes: time limitation (1, 2 and 3 h, respectively) and parking price (56 RSD (0.53 EUR), 38 RSD (0.36 EUR) and 31 RSD (0.30 EUR) per commenced hour, respectively). Most visitors pay parking by mobile phone, although other technologies are in use, such as parking metres, parking tickets and electronic tickets.
As already mentioned, the aim of this paper is to predict the effects of the introduction or change of the parking policy, namely, time limitation and the parking price. These policies refer only to users who park at public, nonreserved parking spaces and who are subject to the parking regime (so-called visitors). It is therefore necessary, in the first step, to isolate this demand and use it for the study. To survey visitors' responses to parking policy changes, the stated preference method is used, where through interviews, various hypothetical situations are presented to the respondents, and they declare how they would behave in such situations. This implies that there are scenarios with different combinations of the time limitation and parking price levels. If the number of all possible scenarios (full factorial) is so great that it is impractical to examine them all, it is possible to examine only a subset, chosen to represent a full set with certainty (fractional factorial) (Hensher and King, 2001; Wong, 2006). The possible choices for visitors are as follows: no change in behaviour, driving to the zone and changing the type of parking space (on-street instead of offstreet or vice versa), parking at the fringe of the zone and continuing the journey by public transport or on foot, using other transport modes, and so on. To better explain the response visitors faced with parking policy changes, it is necessary to gather a wide set of parameters that are assumed to have a strong impact on the travel decision. These are some socio-economic characteristics of the visitors and trip characteristics. All data will be used to fit the multinomial logit (MNL) model, which will predict the probability of choosing each of the stated alternatives for each visitor for any parking price and time limitation. At the end of the modelling process, it is necessary to
J. Simićević et al. / Transport Policy 30 (2013) 125–131
aggregate the demand in order to analyse the results and the effects on the general level (Train, 2002).
4. Survey methodology The proposed procedure will be tested in the Belgrade central area in the red zone for parking. At 811 non-reserved on-street parking spaces in this zone, parking is time-limited to one hour and costs 56 RSD (0.53 EUR). In the zone, there are two parking garages and one off-street parking lot, with a total of 1145 parking spaces. The charge of parking is 75 RSD (0.71 EUR) for the first hour and 90 RSD (0.86 EUR) for each additional hour. Eight street sections were selected to represent on-street parking, while the parking garage Obilićev Venac was chosen to represent off-street parking (Fig. 1). The locations were chosen in such way that based on the previous studies of parking and users characteristics, as well as on the basis of land use in their neighbourhood it can be deemed with certainty that they represent the entire zone. The survey was conducted by face-to-face interviews of users who had just parked or who were leaving the parking space. After a pilot study, which confirmed that both the interviewers and the respondents understood the questions and which assisted in defining the levels of the parking price and time limitation to be presented in the interview, the survey was carried out during 5 days in November and December 2011. The time of the survey was the validity period of the regime (from 7 a.m. to 9 p.m.). On average, each interview lasted 5 min. The interview was conducted by students of the University of Belgrade, Faculty of Transport and Traffic Engineering. Great attention was paid to the interviewers' training and their control in the field. Interviewers were carefully trained on how to present themselves and how to interview. Before the interview, their task was to familiarise the respondent with the research. They emphasised that the research was conducted by the Faculty of Transport and Traffic Engineering, not by the city administration, for scientific purposes. Thus, respondents could not “see through” the intension of the city administration to increase the parking price
Fig. 1. Study area.
127
and could therefore give inaccurate responses (Kelly and Clinch, 2006). Furthermore, an effort was made to overcome the tendency of the respondents to detect and confirm interviewer attitudes, which is considered harmful to the stated preference technique (Tsamboulas, 2001). In total, 438 users were interviewed. For aggregation, the on-street and off-street parking volume is necessary. These data are not surveyed but taken from the study “Parking management strategy”(Milosavljević et al., 2009).
5. The collected data The first question in the interview defined the user category (visitor or other). This question could be eliminated because the further procedure applies only to the category of visitors. Based on theoretical expectations and previous experience, the parameters that were expected to influence the visitors' choice were selected (Tsamboulas, 2001; Coppola, 2002; Shiftan and Golani, 2005; Kelly and Clinch, 2006; Van der Waerden et al., 2006; Khodaii et al., 2010, Simićević et al., 2012a). The following parameters were surveyed: age, gender, origin and destination, car occupancy, parking purpose, parking frequency, parking duration, parking search time, whether the employer pays for parking and whether the respondent is car dependent. Previous studies have shown that respondents often refuse to answer about income or give an incorrect answer (Shiftan and Burd-Eden, 2001). Therefore, in this paper, instead of income, two proxies were surveyed: age of the car and engine size. One of the potential parameters is whether a visitor uses on- or off-street parking. This parameter is considered because different measures are implemented for these types of parking. On the other hand, the visitor's previous choice should depict the visitor's preference towards a certain type of parking (on- or off-street). The second part of the interview was hypothetical scenarios, i.e., different combinations of time limitation and parking price levels. Although the on-street parking price is currently lower than the off-street parking price, in the scenarios, these prices are equal. The reasons why we opt for this restriction are as follows: Because of the negative effects of on-street parking spaces and, in particular, the parking search on the transportation system and environment, the on-street parking price should be higher than or equal to the off-street parking price. However, currently in Belgrade, the situation is the opposite, leading to underutilised off-street parking and over utilised on-street parking. Therefore, we decided for this initial step to set the prices to be equal. An additional reason is the small number of off-street parking spaces in the total parking supply (Section 2). This paper examines the mitigation and tightening of parking measures. Based on the results of the pilot study the values of time limit were chosen, as well as the prices for which visitors' responses will be studied. Five prices in the range of 30 RSD (0.29 EUR) to 200 RSD (1.90 EUR) were taken, and three values of the time limitation, 30, 60 and 120 min, were taken. All possible combinations of the stated values of regime attributes (full factorial) were studied, i.e. in total 15 scenarios (5 3) were tested. However, visitors' responses in certain scenarios are obvious (can be predicted with high probability). These are scenarios in which the regime attributes of the space where the visitor is parked are unchanged or mitigated, and the attributes of another type of parking are unchanged or even tightened. To reduce the number of scenarios which were investigated, we decided not to investigate these “known” scenarios but to assume their outcome. Thus, four scenarios were eliminated (for on-street and off-street parking interviews). That is, we had 11 scenarios to
128
J. Simićević et al. / Transport Policy 30 (2013) 125–131
investigate. Because it is unreasonable to present such a large number of scenarios to one respondent (Hensher and King, 2001) because of the lapse of concentration and the possibility of withdrawal from the interview, we conducted three types of interviews containing three or four scenarios that were randomly selected, and one or two “known” scenarios were added later. The order of the presented scenarios was such that the parking price increased gradually (Kelly and Clinch, 2006) as visitors slowly considered the threshold to which they were willing to pay for parking. It would be wrong to suddenly present a situation in which parking is up to 250% more expensive because this situation is difficult to conceive, and the stated response would be questionable (Hensher and King, 2001). An example of a combination of scenarios is shown in Fig. 2.
6. Results 6.1. MNL model The gathered data were used to fit the MNL model. Respondents had four alternatives and the possibility to write down an alternative if it was not listed (“other”), as shown in Fig. 2. However, because of the small number of choices of some alternatives and because a large number of dependent variable categories make modelling much more complicated, alternatives are grouped into the following three categories: (1) on-street parking, (2) off-street parking or (3) not coming to the zone by car. The probabilities to choose these alternatives are marked with P1, P2 and P3, respectively. By definition, the sum of these three probabilities is equal to one (Hess, 2001): P 1 þ P 2 þP 3 ¼ 1
ð1Þ
The adjusted model is given by two equations: log
P1 ¼ αa þ β1a x1 þ ::: þ βia xi P3
ð2Þ
log
P2 ¼ αb þ β1b x1 þ ::: þ βib xi P3
ð3Þ
where α – intercepts, x – independent variables relevant to the choice, β –parameter estimates. The independent variables were selected from the set of surveyed parameters. Their parameter estimates were estimated by the maximum likelihood method. The final model included five independent variables. In addition to the parking price and time limitation, there are car dependency, parking purpose and current choice. It should be noted that Wald statistics revealed the significance of some other variables, namely, whether the employer pays for parking, length of the trip and income proxies (age of the car and engine size). However, their inclusion would not significantly contribute to the model adjustment to the observed data. For this reason and because of the intension to create a realistic model that does not require too much data (Ortuzar and Willumsen, 2000), these variables were not included. The fitted model is shown in Table 1. For model presentation, the following characteristics are selected: variable name, parameter estimates (β), standard error, significance and exp(β) (Field, 2005). The test of the full model compared to the intercept-only model is statistically significant, indicating that the set of independent variables reliably distinguish between the choices of onstreet parking, off-street parking and not coming to the zone by car (model χ2 ¼1047; df¼ 10; p o0.000). Nagelkerke's R2 of 0.60
19a. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 30 RSD/h time limitation (on-street): ½ hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19b. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 100 RSD/h time limitation (on-street): 2 hours YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19c. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 150 RSD/h time limitation (on-street): ½ hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
19d. IF THE SITUATION IN THE RED ZONE WAS THE FOLLOWING: Parking price (on- and off-street): 200 RSD/h time limitation (on-street): 1 hour YOU WOULD: 1) park on-street 2) park off-street 3) park at the fringe of the zone 4) switch to public transport 5) other_________________
Fig. 2. Example of SP scenarios.
Table 1 MNL model results. Alternative
On-street parking
Variable
Parameter estimate (β) (Std. error)
Sig.
Exp(β)
Parameter estimate (β) (Std. error)
Sig.
Exp(β)
Intercept Car dependency Purpose work Park on-street Parking price (RSD/h) Time limitation (min.)
0.208 1.758 0.662 2.244 0.028 0.020
0.611 0.000 0.014 0.000 0.000 0.000
5.801 0.516 9.434 0.972 1.020
3.773 1.085 0.439 1.604 0.020 0.003
0.000 0.000 0.049 0.000 0.000 0.190
2.960 0.644 0.201 0.980 0.997
Number of observations ¼1407.
(0.409) (0.236) (0.270) (0.243) (0.002) (0.003)
Off-street parking
(0.346) (0.203) (0.223) (0.185) (0.002) (0.002)
J. Simićević et al. / Transport Policy 30 (2013) 125–131
On-street parking
129
Off-street parking
Parking demand (vehicles)
Not coming to zone by car 700 600 500 400 300 200 100 0 30
70
110 150 Parking price (RSD/h)
On-street parking
190
230
Off-street parking
Parking demand (vehicles)
Not coming to zone by car 800 700 600 500 400 300 200 100 0 30
60 90 Time limitation (RSD/h)
120
Fig. 3. The model based effects of (a) the parking price and (b) time limitation on the parking demand.
indicates that the independent variables explained the most variation of the dependent variable. The likelihood ratio index is equal to 0.65 and shows good performance of the model. The likelihood ratio tests showed that all included independent variables contribute significantly to the prediction. Category “not coming to zone by car” is redundant. Generally, the effects of the independent variables on the dependant variables are logical and expected. Car dependency is determined by the question “Do you use a car for every trip in the city?”, where a positive answer determines car dependent visitor. The results show that car dependent visitors are more likely to park in the zone then to give up driving to the zone compared to visitors who are not car dependent. The next variable is also dummy equal to 1 if the purpose of parking is work and is 0 otherwise. The reason for this division is that visitors with purposeful work, because of their travel time and long parking duration, are unwanted in central areas. Unlike them, visitors with other purposes (business, shopping, leisure…) are essential to the vitality and attractiveness of the area. It should be noted that despite a 1 h time limitation, work is present at red zone on-street parking spaces. The reason for this is the change in its character, i.e., these are employees who need to use cars during working hours. On the other hand, a longer duration can be ensured by system abuse; for example, one can pay for parking
more than once but in different ways to avoid the prescribed time limitation (Simićević et al., 2012b). Parameter estimates (Table 1) reveal that visitors with work are more sensitive to a parking policy change, and it is more likely that these visitors will give up parking in the zone. A visitor's previous choice depicts the visitor's preference to a specific parking type (on- or off-street), so, for example, visitors who use on-street parking are more likely to choose this type of parking in the future compared to those who currently use offstreet parking. As the parking price in the zone increases, the odds of parking in the zone decreases. With every RSD of increase, the odds of parking on-street decrease by 0.97 and of parking off-street decrease by 0.98. As the on-street parking time limitation increases, the odds of parking on-street increase. With every 1 min, the odds increase by 1.02. On the other hand, the odds of parking off-street decrease (by 0.997). In case that the visitors, due to strict parking measures in the zone, decide not to park in such zone (3rd alternative), the ways in which they would come into the city centre are investigated. The results show that in such case the majority of visitors would shift to public transport (51%) or would park on the zone fringe (27%). Not a single visitor stated that he/she would choose car pool.
J. Simićević et al. / Transport Policy 30 (2013) 125–131
800 600 400 200 90
0 30
70
800 600 400 200 120 90
0 30
70
110 150 g pric e (RS D/h)
Parkin
Parking demand (vehicles)
30 230
60 30
190
Time limitation (min)
Parking demand (vehicles)
110 150 190 Parking price (RSD/h)
Time limitation (min)
1000
230
600 500 400 300 200 100 0
120 90 60 30
70
110
30
150
Parking price (RSD/h
190
Time limitation (min)
Parking demand (vehicles)
130
230
)
Fig. 4. The model based effects of parking policies on the parking demand. (a) On-street parking, (b) Off-street parking and (c) Not parking in the zone.
The relations shown in the figures are logical and have already been depicted through parameter estimates (Table 1). Increasing the parking price decreases the demand in on-street and off-street parking, while the share of visitors who will give up driving to the red zone increases. For a parking price up to approximately 110 RSD (1.05 EUR), the off-street parking demand curve is completely horizontal, showing visitors' propensity to a parking price increase. This is not surprising because of the existing off-street parking price. After this threshold, the curve has a similar (although somewhat lower) decline as for the on-street parking curve. Therefore, 110 RSD (1.05 EUR) can be considered the threshold for off-street parking. Unlike the parking price, the time limitation has no significant effect on the abandonment of the car in the area, although there is a logical trend. The reason for this is primarily because of the time limitation in the subject area (1 h), and therefore, only short-term visitors mostly use on-street parking. On the other hand, visitors always have an alternative, off-street parking, where parking is not time limited. Therefore, this is proof that visitors, if they do not fit the time limitation, would rather use off-street parking than give up coming to the zone by car. However, although it is proven that this attribute has little effect on the amount of parking demand in the zone, it affects the parking demand redistribution by the type of parking (on- and offstreet parking). By tightening the time limitation, the share of visitors who will park off-street (where there is no time limitation) increases. This number also increases because of the great share of on-street parking visitors with work, such as business and personal business (78%), who cannot shorten the parking duration to adopt to the new situation (TCRP, 2005). Furthermore, by mitigating this attribute, some visitors who currently use off-street parking because of the time limitation will switch to on-street parking. This is proven by the opposite signs of the parameter estimates (Table 1). Fig. 4 shows that the parking price and time limitation have great influence on visitors' behaviour and, therefore, on the parking demand. For the most stringent tested scenario (price of 200 RSD/h and time limitation of 30 min), 54% of visitors would give up coming to the zone by car. It should be noted that a complete picture can be obtained by a comprehensive survey (interview of households), which would examine the generation of new demand. In this paper, because of financial constraints, it is not done.
6.2. Demand prediction 7. Conclusion The fitted MNL model is further used to calculate the probability of the selection of alternatives, i.e., determining the visitors' sensitivity to parking policies. This is done by solving the MNL equations for probability using a wide range of parking price and time limitation values and fixing the values of the other variables. After calculating the probabilities for each visitor individually, the results are aggregated. The sample consists of on-street and off-street parking visitors. As in the current situation, different regimes and parking prices are implemented, and visitors' sensitivities are different. Therefore, because of a driver's preference towards a previous choice, aggregation is performed by dividing the sample into two segments, visitors who currently park on-street and visitors who currently park off-street. Within each segment, the probabilities of choosing each alternative are summed, and the results are extrapolated to the level of the daily parking volume. The results for the segments are then summed. The individual effects of the parking price and time limitation change are shown in Fig. 3(a and b), respectively, and the impacts of both policies at the same time are presented in Fig. 4.
Impacts of parking price and time limit on behaviour of users and on car use are studied in this paper. The setup hypotheses are proven: parking regime attributes influence parking behaviour and these influences differ for specific user subgroups. In this regard, the MNL model was fitted to predict the effects of introducing or changing the parking price and time limitation levels. The parameters included in the model are, in addition to the parking price and time limitation, car dependency, parking purpose and previous choice. All relations of the independent variables to the dependant variable are logical and expected. All statistics show that the model is very good at fitting the data. The model results confirm that the parking price can affect the amount of parking demand and, therefore, parking utilisation. The impact on car use can be used to fulfil some sustainable transportation objectives such as addressing/mitigating traffic congestion problems, achieving the desired modal split, etc. Tightening time limits in areas where limits already exist does not lead to a significant reduction in the parking demand. However, it enables the parking demand to be managed according to
J. Simićević et al. / Transport Policy 30 (2013) 125–131
the parking type. This can be desirable, among other reasons, to reduce the parking search time, which is a consequence of unevenly utilised types of parking. A typical example of this is the Belgrade red zone (and entire central area), where a vacant offstreet parking space can be found at any time, but visitors spend approximately 342 h daily searching for a vacant on-street parking space (here are included visitors who, when failing to find a vacant on-street parking space, eventually park off-street). Visitors with work are much more sensitive to parking policy interventions than are others. This is a very desirable finding for policy makers because these visitors are unwanted in the city central areas. This enables managing user categories. Finally it should be noted that in this paper, we opted to group alternatives in this way, but depending on research objectives, grouping can be done differently. For example, a policy maker may be interested in what visitors who give up parking in the zone would do. In this example, the most common responses of visitors who would give up parking in the zone are using public transport (51%) and parking at the fringe of the zone (27%). This response leads to the need to monitor the public transport quality of the service and state of parking at the fringe of the zone when changing parking policy. Only 2% of visitors who give up would change the trip destination, which means that parking policies will not significantly jeopardise zone attractiveness and efficiency. References Albert, G., Mahalel, D., 2006. Congestion tolls and parking fees: a comparison of the potential effect on travel behaviour. Transport Policy 13, 496–502. Coppola, P., 2002. A joint model of mode/parking choice with elastic parking demand, Transportation Planning. Kluwer Academic Publishers, Netherlands. D'Acierno, L., Gallo, M., Montella, B., 2006. Optimisation models for the urban parking pricing problem. Transport Policy 13, 34–48. Field, A., 2005. Discovering Statistic Using SPSS. Sage Publication, London. Hensher, D.A., King, J., 2001. Parking demand and responsiveness to supply, pricing and location in the Sydney central business district. Transportation Research Part A 35, 177–196. Hess, D.B., 2001. Effect of free parking on commuter mode choice: Evidence form travel diary data. Transportation Research Record 1753, 35–42. Higgins, D., 1992. Parking taxes: effectiveness, legality and implementation, some general considerations. Transportation 19 (3), 221–230.
131
Ibeas, A., Cordera, R., Dell'Olio, L., Moura, J.L., 2011. Modelling demand in Restricted Parking Zones. Transportation Research Part A: Policy and Practice 45, 485–498. Jović, J., Djorić, V., 2009. Application of transport demand modelling in pollution estimation of a street network. Thermal Science 13 (3), 229–243. Kelly, J.A., Clinch, J.P., 2006. Influence of varied parking tariffs on parking occupancy levels by trip purpose. Transport Policy 16, 487–495. Khodaii, A., Aflaki, E., Moradkhani, A., 2010. Modelling the effect of parking fare on personal car use. Transaction A: Civil Engineering 17 (3), 209–216. Marsden, G., 2006. Evidence base for parking policies—a review. Transport Policy 13, 447–457. Milosavljević, N., Simićević, J., Maletić, G., et al., 2009. Parking Management Strategy [Prilozistrategijiupravljanjaparkiranjem]. Institute of the Faculty of Transport and Traffic Engineering, Belgrade. Ortuzar, J. de D., Willumsen, L.G., 2000. Modelling Transport. John Wiley and sons, UK. Scholefield, G., Bradley, R., Skinner, A., 1997. Study of parking and traffic demand: a traffic restraint analysus model (TRAM). In: Transportation Planning Methods: proceedings of Seminar E held at the European Transport Forum Annual Meeting, Brunel University, England, 1–5 September 1997, vol. p 414. Shiftan, Y., Burd-Eden, R., 2001. Modelling the response to parking policy. Transportation Research Record 1765, 27–34. Shiftan, Y., Golani, A., 2005. Effect of auto restraint policies on travel behaviour. Transportation Research Record: Journal of the Transportation Research Board, 1932, Transportation Research Board of the National Academies, Washington, D.C., pp. 156–163. Shoup, D., 2005. The High Cost of Free Parking. Planers and Press, Chicago. Simićević, J., Milosavljević, N., Maletić, G., 2012a. Influence of parking price on parking garage users’ behaviour. Promet—Traffic&Transportation 24 (5), 413–423. Simićević, J., Milosavljević, N., Maletić, G., Kaplanović, S., 2012b. Defining parking price based on users' attitudes. Transport Policy 23, 70–78. Train, K., 2002. Discrete Choice Methods with Simulation. CambridgeUniversity Press. Transit Cooperative Research Program, 2005. Traveler Response to Transportation System Changes. Transportation Research Board, Washington DC. Tsamboulas, D., 2001. Parking fare thresholds: a policy tool. Transport Policy 8, 115–124. Van der Waerden, P., Borgers, A., Timmermans, H., 2006. Attitudes and behavioral responses to parking measures. European Journal of Transport and Infrastructure Research 6 (4), 301–312. Verhoef, E., Nijkamp, P., Rietveld, P., 1995. The economics of regulatory parking policies: The (im)possibilities of parking policies in traffic regulation. Transportation Research Part A 29 (2), 141–156. Victorian Competition and Efficiency Commission, 2006. International Approaches to Tackling Transport Congestion’, Paper 2 (Final): Parking Restraint Measures, Booz Allen Hamilton. Wong, S.T., 2006. Disaggregate Analyses of Stated Preference Data for Capturing Parking Choice Behaviour. Department of Civil Engineering, The University of Hong Kong. (Master thesis).