Travel behaviour and land use in the information society : Literature review and problem discussion

Full text

(1)VTI notat 48A • 2003. VTI notat 48A-2003. Travel behaviour and land use in the information society Literature review and problem discussion. Author. Mattias Haraldsson. Research division Transport Economics Project number. 92063. Project name. Travel behaviour and land use in the information society. Sponsor. Vinnova.

(2) Contents 1 2 3 3.1 3.2 3.3. Introduction Research questions Earlier studies Tele commuting Tele conferencing e-shopping. 3 3 4 5 6 6. 4 4.1 4.2 4.2.1 4.3 4.4. Activity based transport models The activity-travel pattern Modelling the activity-travel pattern Non compensatory models Interaction between ICT and the activity-travel pattern Example model. 6 7 8 8 9. 5 5.1 5.1.1 5.1.2 5.1.3 5.2. Data sources KOM First trip Distance and duration Number of trips Norwegian time use data. 11 11 12 12 13 14. 6. Econometric analysis of the temporal dimension. 15. 7. Conclusion. 16. Bibliography. VTI notat 48A-2003. 17.

(3) 1. Introduction. In recent years there has been a lot of interest focused on the use of information and communications technologies (ICT) and its possible impact on activity and travel behaviour. The relevance of a study focusing on these questions can be tracked to two sources. The first is that in a time when transports are the source of local problems (congestion) as well as (possible) global climate changes, there has been a hope for ICT use to reduce transportation needs profoundly (Ben-akiva et al., 1996) (Viswanathan et al., 2001) (Engström and Johanson, 1996). One reason for studying the impacts of ICT use is thus to evaluate the realism in the hope for ICT to decrease traffic and the associated social costs (Mokhtarian and Salomon, 2002). Unfortunately the strong appeal of these potential solutions has probably made forecasts overly optimistic (Golob, 2001) (Zumkeller, 2002). There is also a tendency to over estimate the importance of technology but to neglect social factors (Salomon, 1998). The second more general purpose of such a study is the belief in ICT as a factor with great power to transform everyday life. Accessibility to information through internet connected computers, mobile phones and other digital equipment are seen as a means to resolve spatial, temporal and organizational constraints. For example working arrangements can be altered both within companies (tele commuting) (Rapp and Rapp, 1999) and by horizontal disintegration (i.e. outsourcing). There might thus be a potential for ICT to overcome physical distances so that localization of businesses and housing could be done in a geographically less constrained way. Additionally, other activities such as shopping could to some extent be made on the Internet. Although there is a number of studies about the relationship between ICT usage and travelling, to test whether telecommunication are a substitute to travel or just a complement are pointed out as a policy relevant future research issue (Hensher and Golob, 2002) (Hjorthol, 2002). The purpose of this report is to review literature relevant for studies of the impact of ICT on transportation. The review will be used as a starting point for a research project at Swedish national road and transport research institute (VTI) and the department of economics, Uppsala university. The purpose and focus of this project has been outlined in (Ramjerdi, 1999). To the relevant literature belongs of course empirical studies. But for the planned econometric studies it is maybe even more important to find suitable theoretical models. Since the activity-travel patterns of individuals are very complex, support from well structured model, based on realistic assumptions is indispensable (Ben-akiva et al., 1996). Finally the report contains a brief description of data sources that can be used to test different hypothesises about the implications of ICT.. 2. Research questions. The question that are posed in this project is if ICT has any implications for the activity-travel pattern. We can expect the relationship between ICT use and the activity-travel pattern to be either substitution, generation, modification or neutrality. The relationship can be assessed at micro or macro level and also with different time horizons (Mokhtarian and Salomon, 2002). The question can be studied at different levels of aggregation; how does ICT usage in general affects activities and travel and more specifically, what are the implications from certain forms of ICT use on specific parts of the activity travel pattern? Even if limited or specific approaches are interesting, the relevance can be higher for more general studies (Mokhtarian VTI notat 48A-2003. 3.

(4) and Salomon, 2002) (Golob, 2001). In many cases though, studies will have a narrower scope, but even in these cases the relevance issue should be kept in mind. The time aspect is important since decisions about activity-travel pattern in the short run can be assumed to be made given the infrastructure, urban structure, place of living etc., factors subject to slow changes. Some researchers predicts that the information technology in the long run will resolve spatial constraints of dwelling, work and leisure activities (Viswanathan et al., 2001). On the other hand geographical proximity to education, research, subcontractors, financial institutions etc. are increasingly important location factors that leads not to a more dispersed industry landscape but domination by a large clusters (Dignon, 2001). Relatively dense population enables interaction between individuals, which has been proven not only to increase economic growth, but also human happiness (Glaeser, 2000). If the true effects from ICT on the activity travel pattern are to be assessed, it is of course important to control for other factors. At an aggregate level some of the observed relationship between ICT and travel can for instance be ”spurius”, since both ICT and transports are related to economic activity (Mokhtarian and Salomon, 2002). Statistical relation measures such as correlation are neutral in terms of causality. To draw conclusions about causality we need reasonable theoretical models. This has also been formulated as an effort to understand rather than just investigate various relationships (Sk˚amedal, 1999). What aspects of ICT is then interesting? ICT is used as a very broadly defined concepts including access as well as usage of to different kind of technologies. When assessing the impacts from ICT on the activity-travel pattern this distinction is obviously crucial. Further the literature is full of studies about concepts that are actually defined both in terms of ICT access or usage and specific features of the activity-travel pattern. Among these concepts we find telecommuting, tele conferencing and e-shopping. All these examples by definition implies relationships between ICT usage and activity travel pattern and thus their pure existence confirm the existence of relationships between ICT usage and the geographical pattern of work, shopping and so on. If a person works at home with intensive usage of internet connected computer for instance, then there is a relationship. In such cases the research question should be rephrased. Does these confirmed interactions between ICT and activity-travel pattern implies changes in other aspects of the activity-travel pattern? For example: if a person telecommutes, does he or she spend more or less time on shopping trips?. 3. Earlier studies. In this section some empirical results from studies of the relationship between ICT usage and travel patterns will be presented. There is an abundance of micro level case studies of various specific relationships between ICT usage and transportation or activities, while studies dealing with the general implications on a macro level seem to be few (Mokhtarian and Salomon, 2002). The major share of these studies focus on specific aspects of the ICT-activity-travel pattern relation. Although this is an active research field, most studies seem to lack theoretic foundation. An effect of the absence of a theoretical foundation might be faulty designed studies and biased results (Salomon, 1998). In a recent review focused on empirical findings about the relationship between information technology and travel, the conclusion is that ICT use really affects 4. VTI notat 48A-2003.

(5) transportation, but that the direction of this impact is unclear. All possible forms of relationship, i.e. substitution, generation, modification and neutrality can be found, although in different contexts. In studies that focuses on transports in general at a macro level, the findings is that there is a substitution effect for consumers, but a complementary effect among industries. Studies focusing on single activities at the micro level suggests that tele commuting is a substitute for travel, that teleconferencing are mainly used as a complement and that the effects of telephone is ambiguous. Finally, on the micro level the relationship between ICT use and travel in general seem to be complementary (Mokhtarian and Salomon, 2002). Findings from an empirical study with data collected in Germany and South Korea suggest that complementary effects are dominating and that there is ”little room for substitution theories or hopes of decreasing traffic volumes” (Zumkeller, 2002). According to Viswanathan et al., 2001 the interpretation of studies about the impacts from ICT, is that substitution is most likely in the short run. In the long run though, complementary effects is the most likely.. 3.1. Tele commuting. Commuting is a major source of private travel. One of the main possible impacts on travelling from ICT is therefore trough a reduced need for commuting (Engström and Johanson, 1996) (Mokhtarian and Salomon, 2002) (Sk˚amedal, 2001). Consequently, there is a vast literature around what has been called tele commuting (TC), i.e. arrangements that allows individuals to work from home (or other places that are not the main working place) with support from computer, phone and other electronic media. The impacts from ICT might in many instances be restricted by institutional factors, but tele commuting is an organizational arrangement that enables exploitation of the potential of ICT (Engström and Johanson, 1996). Although the appealing concept, tele commuting are not yet very common. In a review of theoretical and empirical studies about TC it is emphasized that the potential for tele commuting to resolve time and space constraints imposed by traditional working arrangements has for almost 30 years lead to expectations about large substitution effects (Sk˚amedal, 2001). Still, the share of the Swedish work force who uses this possibility is not larger than 7 percent and most of them tele commutes only a few times every month (SIKA, 1998). In Europe the share are estimated to between four and eight percent (Hjorthol, 2001). Even if TC should help reducing commute trips, it will probably have travel generating effects as well. A widespread opinion seem to be that TC also will generate new trips, that to some extent counterbalance the potential substitutive effects. TC changes the conditions for trip chaining and trips with several purpose and therefore it has been expected that tele commuters will make a larger number of single purpose trips. But the findings from empirical studies is that TC does not increase non-commuting trips, since tele commuters and their families tend to adopt to more local travel patterns during day time (Sk˚amedal, 2001). It has been shown however that persons who tele commute tends to spend more time on shopping activities, maybe because they do not have the possibility to combine shopping trips with work trips (Gould and Golob, 1997). Many tele commuters works at home only part of days, and thus commutes as much as workers that spend all day at the working place. The anticipated result from such part-day TC is thus not reduced traffic, but a redistribution to off-peak VTI notat 48A-2003. 5.

(6) hours. Studies has confirmed that many people that tele commutes avoid rush hours (Sk˚amedal, 2001). The propensity to tele commute for just half working days also increases car usage, partly as a result of dissolved car pools (Mokhtarian and Salomon, 2002) (Sk˚amedal, 2001). A potentially traffic generating effect triggered by TC is that less constrained working arrangement in the long term could increase the average distance between home and the working place (Golob, 2001). Improved communication facilities has made the land use less dense throughout history. (Sk˚amedal, 2001) If this trend continues we can expect trips actually performed to sum up to a large distance even if the number are limited. This is however long term effects that to a large extent remains to be captured in empirical studies (Mokhtarian and Salomon, 2002) (Sk˚amedal, 2001) (Hjorthol, 2001). From a broad review of TC studies, Mokhtarian and Salomon, 2002 draws the conclusion that the short term net effect are reduced demand for travel. This is also the result from Hjorthol, 2001. Due to potential generative effects from residential location etc., in the long run the substitution effect will probably decrease (Mokhtarian and Salomon, 2002).. 3.2. Tele conferencing. Tele conferencing is an interactive technology that enables people to have meetings face to face without actually being at the same place. Until today equipment for teleconferencing has been expensive and has not had any significant effect on business travelling (Golob, 2001). However, SIKA predicts that tele-conferences has a high probability of a breakthrough in a quite near future (SIKA, 2001). Increased time cost and a more favorable price relation between travel and video conferences might together with a more user friendly technology result in increased usage in the future (Hjorthol, 2001).. 3.3. e-shopping. Shopping behaviour and shopping related travelling might also be affected by the increases accessibility to electronic communication technology (Golob, 2001). Although e-shopping doesn’t have any significant impacts on the transports at the moment, it might have in the future (SIKA, 2001). There is reason to believe that this transformation is impeded by tax structures and other institutional arrangements (Svensson and Haraldsson, 2002). Another factor is the responses from vendors that use the traditional distribution channels. If e-shopping should become more popular we might thus expect reactions aiming to encourage customers to visit shopping malls (Mokhtarian and Salomon, 2002).. 4. Activity based transport models. The purpose of traditional transport models, i.e. the four-step approach, is to predict traffic load on network links. It would of course be possible to perform the project within this framework, as a means to improve traffic forecasts (Ben-akiva et al., 1996). To extend forecast models with a module, where ICT is treated as an additional mode is pointed out as an urgent need (Handy and Mokhtarian, 1996). Research in this direction is presented in (Börjesson, 2003). The four step model was constructed as a means to solve practical problems, weather the focus of activity based models are more on human behaviour (Recker 6. VTI notat 48A-2003.

(7) et al., 1986a). The school of activity based transport models has a different focus. If the main purpose of traditional models is prediction, activity based models are motivated by the endeavor to understand human activity and travel behaviour (McNally, 2000) (Recker et al., 1986a). Because the purpose of the project is to find out if and how ICT affects the activity-travel pattern, the activity-based approach seems to be a good choice, not the least since researchers has pointed out the integration of studies about the impacts of ICT and activity-based models as a potentially fruitful direction for further research (Golob, 2001). The core idea of activity based models is that transport is a derived demand, i.e., the amount of transports performed are seen as a function of the activities that people take part in. Other characteristics of activity based models are the focus on trip chaining and activity sequences, timing and duration of activities (de Palma and Fontan, 2001) (Misra, 1999). Interaction between individuals, which imposes constraints on the possibilities is also a main feature of the activity based approach. Many of these features seem to follow the time geography, a model developed by the Swedish geographer Torsten Hägerstrand (Carlestam and Sollbe, 1991). The activity based modelling rests on a set of assumptions about human activity-travel behaviour. Still a widespread opinion is that this research field is theoretically underdeveloped. This might be a result from the diverging applications of activity based models, and the fact that the features of activity-based models are defined in fairly broad terms. Attempts has been made to unify different aspects of the activity travel pattern in general analysis frameworks (Recker et al., 1986a) (Recker et al., 1986b) (Recker, 1994) (Misra, 1999) (Bhat and Koppelman, 1993) (Benakiva et al., 1996).. 4.1. The activity-travel pattern. Since the purpose of this study is to test and quantify the relationship between ICT usage and the activity-travel pattern of individuals, there is a need for a definition of activity-travel pattern. The activity travel pattern can be represented by a vector of characteristics (Kitamura et al., 1997) (Kitamura and Fujii, 1998). (Xi , Ti , Li ) = (Xi0 , Xi1 , . . . , Xin ; Ti0 , Ti1 , . . . , Tin ; Li0 , Li1 , . . . , Lin ). (4.1). where Xij Tij Lij. = type of j:th activity by individual i = duration of j:th activity by individual i = location of j:th activity by individual i. Activity-based analysis deals with a range of concepts that all can be derived from the activity-travel vector. The subparts of the activity-travel patterns are characterized by different attributes, such as activity type, travel mode etc. From the set of all possible combinations of the subparts, each individual chooses a specific subset which defines his/her activity-travel pattern. This choice process should to largest possible extent be used as a starting point of any empirical analysis.. VTI notat 48A-2003. 7.

(8) 4.2. Modelling the activity-travel pattern. Each individual has to choose one of a potentially unlimited number of possible activity-travel vectors. A really optimal choice is possible only if all vectors are evaluated and compared to each other in a simultaneous process. This kind of decision making are analysed with compensatory models, maybe more known as utility maximization models where an individual chooses the option that maximizes his/hers (expected) utility and is feasible under specific budget constraints. The choices are based on weighted sums of attribute values, where good attributes in some sense compensates for bad attributes. Utility maximization has however been criticized from researchers that emphasizes the importance of interpersonal constraints, choice complexity and imperfect information (Recker et al., 1986a). Besides the fact that joint decision is an unrealistic assumption, simplifications are necessary for practical reasons. As the number of possible activity-travel patterns are almost unlimited joint modelling is infeasible due to complexity (Bowman and Ben-Akiva, 1997). 4.2.1. Non compensatory models. Imposing some structure on the activity-travel vector results in non compensatory models where individuals are assumed to use simple rules- heuristics to reduce the choice set. Similar models is the result if some alternatives are unfeasible due to different restrictions. Here the decision process will be sequential, based on some measure of importance (Eriksson, 1996) (Eriksson, 1997). A crucial issue is then the order of the sub decisions. The order can be either assumed in theories, derived from data or defined by given restrictions. The division of the decision process into a sequence of sub decisions reduces the probability to make utility maximizing decisions, since all alternatives are not evaluated (Arentze and Timmermans, 1998) (Gärling, 1997). One commonly used non compensatory model is the heuristic based computational process models (CPM) (Kitamura and Fujii, 1998). Although CPM:s and compensatory models differs fundamentally, There is a continuum between the two extremes of compensatory models and CPM, consisting of weak CPM. Weak CPM are models that uses heuristics to determine the order decisions are made in, but uses utility maximization in each sub decision (Arentze and Timmermans, 1998). Non compensatory decision rules can be classified in dominance, satisfaction and lexicographic rules (Ben-Akiva and Lerman, 1985). A dominant alternative is better than other alternatives for at least one attribute and no worse on other attributes. If satisfaction is used as a decision rule, the decision maker chooses alternatives that meet specified requirements on some attribute and are no worse than other alternatives on other attributes. With lexicographic rules the decision maker puts the attribute in order of importance and then chooses the alternative that are best on the most important attribute. Gärling, 1997 and Gärling et al., 1997 describes the relationship between sequential decisions in terms of goal/instrument, where a subordinate decision is a instrument for a higher decision. In general most approaches seem to assume that activities and travels are performed as part of pursuit of some specific objective (Arentze and Timmermans, 1998). Since the departure of activity-based transport research is that travel demand is derived from the activities performed, the most natural is to condition travel on the activity scheme. 8. VTI notat 48A-2003.

(9) An alternative to assumed rules or rules derived from theory, is rules based on empiric analysis. To find these rules quantitative methods outside the scope of traditional econometric and statistical methods, such as different adaptive data mining algorithms has been tried (Timmermans et al., 2000). On example of a transportation simulation based on empirically determined decision trees is ALBATROSS (Arentze and Timmermans, 1998) Predetermined or mandatory activities, i.e. interaction with other people in time and space, puts firm constraints of the set of feasible activity patterns. According to the time-geography theory developed by Hägerstrand, time-space constraints restricts the activity-travel pattern of individuals (Golob, 2001). The concept of timespace constraints are used in several models and studies, although the terminology differs. A common assumption is that people makes decisions about prioritized activities first. The most important activity is not necessary the first activity of the day or not even the activity that takes most time, but the activity that are considered most important in some respect. Mandatory activities are of course top priority and thus put restriction on the decision process Bowman and Ben-Akiva, 2001 (Arentze and Timmermans, 1998). Recker, 1994 formulates the complete activity-travel problem as an optimization problem, where the objective is to minimize transport costs under a large set of temporal and spatial constraints. The spatial and temporal constraints or time-space prisms is also the main feature of the model PCATS, prism-constrained activitytravel simulator (Kitamura and Fujii, 1998). PCATS is given structure by dividing the day in open and blocked periods, where blocked periods are such periods where a specific activity are performed with probability one. In open periods on the contrary, individuals have real opportunities to chose among different activities and trips Gärling et al., 1998. make a distinction between routine and nonroutine activities, where a routine activity are considered fixed in time and space and not a result of short term decision making. A similar concept is also core stops used by (Hanson and Huff, 1988).. 4.3. Interaction between ICT and the activity-travel pattern. We have not yet discussed in witch stage of the decision process that ICT issues are dealt with. It is quite tempting to condition different aspects of the activity-travel pattern on ICT variables, i.e. use the ICT variables as regressors. Ben-akiva et al., 1996, for instance, regards ”IT options accessibility” as a mobility and life style choice, on the same decision level as employment, housing and auto ownership. They assume that decisions on that level has a time horizon of years. This is probably a troublesome assumption. Is decisions about ICT made on such long term basis as years? And isn’t there a feedback loop from lower level decisions? The view of Golob, 2001 is that choices of ICT and activity patterns are ”intertwined” and that joint modelling is necessary to capture the mutual causality. An approach in this fashion is a system of equations where several contact variables are used both as dependent and independent variables presented in Mohktarian and Meenakshisundaram, 1999. The findings are that there are in some cases relations in more than one direction.. VTI notat 48A-2003. 9.

(10) 4.4. Example model. One comprehensive activity-based models is developed by a research team headed by Chandra Bhat (Misra, 1999) (Bhat and Misra, 2001) (Bhat and Singh, 2000) (Bhat, 1996a) (Bhat, 1998). The model suggests a specific form of decision order for the activity-travel pattern. The model departs from the differences in restrictions between workers, that are are obliged to attend at their working place during a specified time interval, and non workers. If lunch time is included, the worker has three periods under which he or she can perform non fixed activities, before work, during lunch and after work. Non workers on the other side have no coupling constraint of similar importance and their choice of activity travel pattern are thus more complex (Bhat and Singh, 2000). The model is divided in three parts that are connected to each other in a specific order. Similar principles are used both workers and non-workers. An important difference though is that the non worker model are estimated for the whole day, while the worker model are employed separately to each of the three non-fixed periods.. Pattern level attributes • Stay home all day or not • Number of stops in the day (if individual does not stay home all day) • Number of stops of each activity type • Assignment of stops to tours and sequencing of all activity stops. Tour level attributes • Home-stay duration before tour • Mode choice for tour. Stop level attributes • Travel time to current stop • Activity duration • Destination. Figure 4.1 Structure for activity-travel models In the model an activity outside the home is called a stop and the sequence of stops (activities) between two in-home activities are called a tour. On the highest level the pattern level attributes are determined. Here models are used to determine the number of stops, assign stops to tours and sequencing the different activities. Misra, 1999 argues that from an activity-based perspective it is natural that the activity pattern has a superior position in the model. There can also be a model for the choice to stay home all day or not. On the middle level 10. VTI notat 48A-2003.

(11) we find models for tour level attributes, home stay duration before tour and mode choice. On the bottom level stop level attributes are modelled. The group of stop level attributes includes travel time to stops, activity duration and destination.. 5. Data sources. The data sources primarily intended to be used in the project are a Swedish survey about travel and communications behaviour and a Norwegian time use study. In the following section the two data sources will be described briefly.. 5.1. KOM. The collection of diary data for both travelling and other means of communication has been pointed out as a requirement for the understanding of interaction between physical and non-physical forms of communication (Hjorthol, 2001). According to Golob, 2001, information about type, purpose, duration, location and send or receive mode are required, additional to corresponding activity-travel data. He proposes travel surveys to be extended with questions about non physical communication. This is the type of data in KOM, a Swedish diary database with information about transports as well as non physical communication such as phone, fax, e-mail and so on. The purpose of the survey is to describe the travel and communication patterns in different demographic segments and analyse the relationship between ICT usage and transports (SIKA, 2001). Since a very important aspect of the activity-travel pattern is the presence of fixed activities, the number of workers and non workers in the sample is crucial. In KOM 59 percent of the interview object 2000 were either full time or part time workers or run companies. Thus 41 percent were classified as non-workers. In the 1999 KOM data 60 percent is workers according to the same definition and 40 percent non-workers. Table 5.1 Sample size Workers Non workers Total. VTI notat 48A-2003. 1999 516 347 863. 2000 Total 968 1487 673 1020 1641 2504. 11.

(12) 5.1.1. First trip. As the diagram shows the distribution of time for first trip of the day has two peaks. The most common is that the first trip is performed between in the morning, while the other peak is in the afternoon.. 14,00%. 12,00%. 10,00%. 8,00%. 6,00%. 4,00%. 2,00%. 23:00-23:59. 22:00-22:59. 21:00-21:59. 20:00-20:59. 19:00-19:59. 18:00-18:59. 17:00-17:59. 16:00-16:59. 15:00-15:59. 14:00-14:59. 13:00-13:59. 12:00-12:59. 11:00-11:59. 10:00-10:59. 09:00-09:59. 08:00-08:59. 07:00-07:59. 06:00-06:59. 05:00-05:59. 04:00-04:59. 03:00-03:59. 02:00-02:59. 01:00-01:59. 00:00-00:59. 0,00%. Figure 5.1 Homestay duration. 5.1.2. Distance and duration. About 40 percent of the total number of trips performed lasts between 1 and 10 minutes. The shares of trips decreases as the travel time increases. A comparison with the distribution of travelled distances reveals a compelling similarity.. 45,00%. 40,00%. 35,00%. 30,00%. 25,00%. 20,00%. 15,00%. 10,00%. 5,00%. 0,00% 1-10 min. 11-20 min. 21-30 min. 31-60 min. 1- 2 h. 2- 4 h. Figure 5.2 Travel time. 12. VTI notat 48A-2003.

(13) 50%. 45%. 40%. 35%. 30%. 25%. 20%. 15%. 10%. 5%. 0% - 1,9 km. 2- 4,9 km. 5- 9,9 km. 10-19,9 km. 20-49,9 km. 50-99,9 km. 100 km -. Figure 5.3 Distance distribution. 5.1.3. Number of trips. In KOM there is a distinction between main trips and part trips. A main trips starts or ends at home, work, school or temporary place of living, e.g. a hotel. Every main trip consists of one ore more part trips, which are defined as a movement between places, where a specific activity are performed. A large share of the respondents in the sample conducts two to four part trips every day and very few more than ten.. 30%. 25%. 20%. 15%. 10%. 5%. 0% 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Figure 5.4 Number of part trips. VTI notat 48A-2003. 13.

(14) 5.2. Norwegian time use data. The Norwegian time use data are collected 1970, 1980, 1990 and 2000 and contains detailed information about how much time that are devoted to activities during two days. The most recent data, i.e. 2000, contains information about ICT usage.. Table 5.2 Time spent on trips. Do not use PC Use PC but not Internet Use PC and Internet Total. Minutes a day Working Leisure Household trips trips work trips 11.3 27.5 20.8 18.5 30.4 17.1 22.5 37.1 17.5 18.0 32.6 18.5. Without proving anything about the eventual causality between ICT usage and time spent on working trips, the table above shows that people that are more intensive ICT users tend to spend more time commuting and on trips related to leisure. For trips related to household work there is no clear tendency. Here we have only studied the time spent on travels with different purposes. This is just to show what type of analyses that can actually be made with this data. There is nothing that restricts our time use studies to travel time however. The Norwegian time use survey material contains a very rich data that allows us to study time usage on different activities on a very detailed level. The Norwegian time use data can be used to compare time usage and allocation over the years 1970-2000 and try to identify possible structural shifts for the last period, i.e. 2000, eventually caused by ICT usage. One first attempt to look at this potential relationship is presented in the table below. Obviously time used on travels have increased during the period.. 40 35. Minutes/day. 30 25. Work related travel. 20. Travel related to household work Lesiure related travel. 15 10 5 0 1970. 1980. 1990. 2000. Figure 5.5 Development of travel time (ages 16-74). 14. VTI notat 48A-2003.

(15) 6. Econometric analysis of the temporal dimension. Time can as we have seen be treated as a constraint on the possible activities, but has a number of alternate conceptualizations. It can also be incorporated in the models by ordering the activities temporarily or by measuring the duration (Kurani and LeeGosselin, 1997). Since the same data generation process can produce binary, count and duration data, different but related econometric methods can be used (Alt et al., 2000). One first option is to conceptualize timing of an activity as a discrete choice from a set of points in time, an approach used in Bowman and Ben-Akiva, 2001. If the measurement period is an interval instead of a point one might instead count how many times something has happened during that period. In this case a count model such as a Poisson or negative binomial can be used. An alternative to count models is ordered discrete models, where the discrete number of stops are assumed to be a result of an underlying continuous propensity to make trips. Bhat and Singh, 2000 argues that the ordered response model is better than a count model such as Poisson when the variation is number of stops is limited. A third possibility is to measure the time that elapses between the occurrences and use this data to estimate a duration model. Duration can be viewed as way to optimize utility by fine scaled time allocation (Kitamura and Supernak, 1997) (Supernak, 1992). Duration models are interpreted in terms of probability for the activity to end, given that it has already last a certain amount of time. This conditional probability is called the hazard. In many instances there is a priori reasons to assume certain relations between the hazard and time. If one does not take the hazard rate in consideration, the risk is that the choice of survival distribution are not consistent with reasonable assumptions about the hazard (Kiefer, 1988) (Bhat, 2000) (Hensher and Mannering, 1994) (Lee, 1992). In some contexts it is reasonable to assume that the relationship between time spent and probability to quit is positive and in others a negative relation seem more appropriate. There is also examples where this probability does not depend on the time spent. Such processes, e.g. Poisson, are called memoryless (Greene, 1997). If one cannot justify the choice of any specific distribution, nonparametric distributions can be an option (Bhat, 1996b). In Bhat, 1996a a hazard based model is used to model the duration of activities performed during the home commute. Home stay durations for non-workers are estimated by Misra, 1999, using a hazard based model. Although hazard based models are a suitable choice for duration data, more general approaches can also be found in the literature. In a model of the period after evening commute, a system of linear equations with normal distributed errors are used to estimate home stay duration and activity duration (Bhat, 1998). In the nonworker model in (Misra, 1999) a log-normal model is used for travel time duration of different activities. Duration models deals with the amount of continous time allocated to activities. A more aggregate approach is used within time use research. Here the total amount of time used for specific purposes are measured without attention to whether this time is consecutive or a sum over several periods. Time use analysis, with roots in economics, treats time as a resource witch are allocated efficiently under different set of time and budget restrictions (Becker, 1965) (Hallberg, 2002). The attention to different activities in the activity based research field shows in many instances a VTI notat 48A-2003. 15.

(16) close resemblance to the studies of the allocation of time. A survey about time use research in a context of activity based transportation models can be found in Bhat and Koppelman, 1999. From the perspective of travel modelling transports are considered just one type of activity that the individual allocates a certain amount of time to. Fujii et al., 1999 uses a structural equations model, i.e. a system of linear regression equations to analyse time allocation between different activities as well as the number of trips and time used for travelling. Even though it is a growing interest in the transport research community about time use studies, this approach seem to be under utilized in this context (Pas and Harvey, 1997). There are although examples of transport related time use studies (Fujii et al., 1999) (Gangrade et al., 2000).. 7. Conclusion. The review of earlier studies about implications of ICT on the activity-travel pattern, reveals that this is a very complex issue, not at least methodologically. That the research suffers from lack of theory seem to be a widespread opinion. Many authors emphasizes that the research lacks theoretical foundations. The reviewed activitybased paradigm, although somewhat immature are based on sound assumptions and some comprehensive models have been presented. To use the activity-based approach in research about ICT and the activity-travel behaviour seems therefore to be a fruitful approach for future research in this project. What remains to be decided before work in this project proceeds, is which particular aspects of the activity-travel pattern to cover. As we have seen there are many dimensions that can be used to define the scope of the study. With the data material available through KOM and the Norwegian time use survey different kinds of micro studies are possible. To use these data sources should result in quite quick progress, and the conclusions could at a later stage be used to formulate how surveys should be designed to capture important aspect of this issue.. 16. VTI notat 48A-2003.

(17) Bibliography Alt, J. E., King, G., and Signorino, C. S. (2000). Aggregation among binary, count and duration models: estimating the same quantities from different levels of data. http://gking.harvard.edu/preprints.shtml, 000925. Arentze, T. A. and Timmermans, H. J. (1998). Albatrossa learning based transport oriented simulation system. http://www.infra.kth.se/tlenet/meet5/papers/Timmermans2.pdf. Becker, G. S. (1965). A theory of the allocation of time. The economic journal, 75:493–517. Ben-akiva, M., Bowman, J. L., and Gopinath, D. (1996). Travel demand model system for the information era. Transportation, 23:241–266. Ben-Akiva, M. and Lerman, S. R. (1985). Discrete choice analysis- theory and application to travel demand. MIT Press. Bhat, C. and Misra, R. (2001). Comprehensive activity travel pattern modeling system for nonworkers with empirical focus on organization of activity episodes. Transportation research record, (1777):16–24. Bhat, C. and Singh, S. K. (2000). A comprehensive daily activity-travel generation model system for workers. Transportation Research, 34A:1–22. Bhat, C. R. (1996a). A generalized multiple durations proportional hazard model with an application to activity behaviour during the evening work-to-home commute. Transportation Research, 30B(6):465–480. Bhat, C. R. (1996b). A hazard-based duration model of shopping activity with nonparametric baseline specification and nonparametric control for unobserved heterogeneity. Transportation Research, 30B(3):189–207. Bhat, C. R. (1998). A model of post home-arrival activity participation behavior. Transportation research, 32B(6):387–400. Bhat, C. R. (2000). Duration modelling, chapter 6, pages 91–111. In (Hensher and Button, 2000). Bhat, C. R. and Koppelman, F. S. (1993). A conceptual framework of individual activity program generation. Transportation research, 27A:433–446. Bhat, C. R. and Koppelman, F. S. (1999). A retrospective and prospective survey of time-use research. Transportation, 26:119–139. Börjesson, M. (2003). Communication technology and travel demand models. TRITA-INFRA 03-051, Royal institute of technology, departement of infrastructure. Bowman, J. and Ben-Akiva, M. (2001). Activity-based disaggregate travel demand models system with activity schedules. Transportation research, 35 A:1–28. VTI notat 48A-2003. 17.

(18) Bowman, J. L. and Ben-Akiva, M. (1997). Activity-based forecasting. In (act, ). Carlestam, G. and Sollbe, B., editors (1991). Om tingens vidd och tinges ordningtexter av Torsten Hägerstrand. Byggforskningsr˚adet. In Swedish. de Palma, A. and Fontan, C. (2001). Trip timing and chaining, chapter 21, pages 351–355. In (Hensher, 2001). Dignon, S. (2001). Location patterns: case study of IT firms in stockholm. Master of science thesis 01-160, Royal institute of technology, Stockholm, Sweden. Engström, M. G. and Johanson, R. (1996). It-utvecklingens effekter p framtida res- och transportstrukturer. Rapport 4515, Naturvrdsverket. in swedish. Eriksson, M. (1996). Selektion till arbetsmarknadsutbildning. Ume˚a economic studies 410, Ume˚a university. In swedish. Eriksson, M. (1997). To choose or not to choose. Ume˚a economic studies 443, Ume˚a university. Fujii, S., Kitamura, R., and Kishizawa, K. (1999). An analysis if individuals joint activity engagement using a model system of activity-travel behavior and time use. Transportation Research Record, 1676:11–19. Gangrade, S., Kasturirangan, K., and Pendyala, R. M. (2000). Coast-to-coast comparison of time use and activity patterns. Transportation research record, 1718:34– 42. Gärling, T. (1997). Psykologisk teoretisk referensram för hush˚allens beslutsfattande avseende resor. Rapport 1997:48, KFB. In swedish. Gärling, T., Gillholm, R., Romanus, J., and Selart, M. (1997). Interdependent activity and travel choices: behavioural principles of integration of choice outcomes, chapter 7, pages 135–149. Pergamon. Gärling, T., t. Kaln, Romanus, J., Selart, M., and Vilhelmson, B. (1998). Computer simulation of household activity scheduling. Environment and planning, 30A:665– 679. Glaeser, E. L. (2000). The future of urban research: nonmarket interactions. Brookings-Wharton papers on urban affairs, pages 101–138. Golob, T. F. (2001). Travelbehaviour.com: Activity approaches to modelling the effects on information technology on personal travel behaviour, chapter 6, pages 145–183. In (Hensher, 2001). Gould, J. and Golob, T. F. (1997). Shopping without travel or travel without shopping? an investigation of electronic home shopping. Transport reviews, 17(4):355– 376. Greene, W. H. (1997). Econometric analysis. Prentice Hall, 3:rd edition. Hallberg, D. (2002). Essays on household behavior and time-use. Economic studies 63, Department of economics, Uppsala university. 18. VTI notat 48A-2003.

(19) Handy, S. L. and Mokhtarian, P. L. (1996). Forecasting telecommuting. Transportation, 23(2):163–190. Hanson, S. and Huff, J. O. (1988). Systematic variability in repetitious travel. Transportation, 155(1-2):111–135. Hensher, D., editor (2001). Travel behaviour research, the leading edge. International association for travel behaviour research. Hensher, D. A. and Button, K. J., editors (2000). Handbook of transport modelling. Pergamon. Hensher, D. A. and Golob, J. (2002). Telecommunications-travel interaction: workshop report, chapter 9, pages 209–219. In (Mahmassani, 2002). Hensher, D. A. and Mannering, F. L. (1994). Hazard-based duration models and their application to transport analysis. Transportation Reviews, 14(1):63–82. Hjorthol, R. (2001). Samspill mellom mobilitet og informasjons- og kommunikasjonsteknologi- en litteraturstudie. TØI Rapport 576, TØI. In norweigan. Hjorthol, R. J. (2002). The relation between daily travel and use of the home computer. Transportation Research, 36A:437–452. Kiefer, N. M. (1988). Economic duration data and hazard functions. Journal of Economic Literature, XXVI:646–679. Kitamura, R., Chen, C., and M.Pendyala, R. (1997). Generation of synthetic daily activity-travel patterns. Transportation research record, 1607:154–162. Kitamura, R. and Fujii, S. (1998). Two computational process models of activity -travel behavior, pages 251–279. Kitamura, R. and Supernak, J. (1997). Temporal utility profiels of activities and travel: some empirical evidence, chapter 14, pages 339–350. In (Stopher and Lee-Gosselin, 1997). Kurani, K. S. and Lee-Gosselin, M. E. (1997). Synthesis of past activity analysis applications. In (act, ). Lee, E. T. (1992). Statistical methods for survival data analysis. John Wiley Sons. Mahmassani, H. S., editor (2002). In perpetual motion Travel behavior and research opportunities and application challanges. Pergamon. McNally, M. G. (2000). The activity-based approach, chapter 4, pages 53–69. In (Hensher and Button, 2000). Misra, R. (1999). Toward a comprehensive representation and analysis framework for non-worker activity-travel pattern modeling. PhD thesis, The University of Texas at Austin. VTI notat 48A-2003. 19.

(20) Mohktarian, P. L. and Meenakshisundaram, R. (1999). Beyond tele-substitution: disaggregate longitudinal structural equations modeling of communications impact. Transportation research, 7C:33–52. Mokhtarian, P. L. and Salomon, I. (2002). Emerging travel patterns: Do telecommunications make a difference, chapter 7, pages 143–182. In (Mahmassani, 2002). Pas, E. I. and Harvey, A. S. (1997). Time use research and travel demand modelling, chapter 13, pages 315–338. In (Stopher and Lee-Gosselin, 1997). Ramjerdi, F. (1999). Transport and land use in the information society. Report TRITA-IP FR 99-53, Royal institute of technology, Stockholm, Sweden. Rapp, B. and Rapp, B. (1999). Flexibla organisationslsningar- om flexibla arbetsformer och flexibla kontor. KFB-rapport 1999:30, Kommunikationsforskningsberedningen. in swedish. Recker, W. W. (1994). The household activity pattern problem: general formulation and solution. Transportation research, 29B(1):61–77. Recker, W. W., McNally, M. G., and Root, G. S. (1986a). A model of complex travel behavior: part I- theoretical development. Transportation research, 20A(4):307–318. Recker, W. W., McNally, M. G., and Root, G. S. (1986b). A model of complex travel behavior: part II- an operational model. Transportation research, 20A(4):319–330. Salomon, I. (1998). Technological change and social forecasting: the case of telecommunication as a travel substitute. Transportation research, 6 C:17–45. SIKA (1998). It-utvecklingen och transporterna 2. Rapport 4, SIKA. In swedish. SIKA (2001). Kommunikationsmönster hos befolkningen, resultat fr˚an SIKA:s kommunikationsundersökningar. Rapport 6, SIKA. In swedish. Sk˚amedal, J. (1999). Arbete p˚a distans och arbetsformens p˚averkan p˚a resor och resmönster. Linköping studies in science and technology 752, Department of computer and information science, Linköpings universitet, Linköping. In swedish. Sk˚amedal, J. H. B. (2001). How telecommuting affect travel- recent findings from the literature! Technical report, Department of computer science, economic information systems, Linköpings universitet. Paper presented at Cities of Tomorrow, 23-24 August 2001, Gteborg, Sweden. Stopher, P. and Lee-Gosselin, M., editors (1997). Understanding travel behaviour in an era of change. Pergamon. Supernak, J. (1992). Temporal utility profiles and travel: uncertainty and decision making. Transportation research B, 26:61–76. Svensson, T. and Haraldsson, M. (2002). Konsekvenser av dagligvaruhandelns strukturomvandling. Rapport 485, VTI, Linkping, Sweden. In swedish. 20. VTI notat 48A-2003.

(21) Timmermans, H., Arentze, T., and Joh, C. (2000). Modeling learning and evolutionary adaptation processes in activity settings-theory and numerical simulations. Transportation research record, 1718:27–33. Viswanathan, K., Goulias, K., and Kim, T. (2001). On the relationshipbetween travel behavior and information and communications technology (ict): what do the travel diaries show? In Sucharow, L. and Brebbia, C., editors, Urban transport VII, Urban transport and the environment in the 21st century. Zumkeller, D. (2002). Transportation and telecommunication: first comprehensive surveys and simulation approaches, chapter 8, pages 183–207. In (Mahmassani, 2002).. VTI notat 48A-2003. 21.

(22)

No results found