جایگزینی بین کار در خانه و خارج از خانه : نقش فناوری اطلاعات و ارتباطات و هزینه های رفت و آمد

This paper analyzes the trade-off between working at home and out-of-home, ICT and commuting time. To this end, we develop a microeconomic demand system, which explicitly incorporates both time and income constraints. Commuting time is considered as the price to be paid for working out-of-home and a decrease in earnings as the price for working at home. For the latter, we find that working at home leads to a (marginally significant) reduction of the wage rate of about 19%, but this gap largely disappears when ICT is used for at home work. To examine the relation between out-of-home and at home work empirically, we estimate a translog indirect utility function, from which we are able to estimate substitution and price elasticities between working at home and out-of-home for The Netherlands. The results show that changes in ICT and commuting time display rather weak substitution effects on working out-of-home and at home, respectively. Moreover, individual characteristics – especially age and education – seem to be more important for the choice between working at home and out-of-home than ICT availability or commuting time.

Since the mid 1970s, the phenomenon of teleworking has witnessed a great deal of attention in the literature (see for overviews, inter alia, Nilles, 1988, Salomon, 1986, Salomon, 2000 and Mokhtarian, 1990). Telework in itself is especially interesting, because since the industrial revolution individuals increasingly organized themselves in firms and institutions outside the location of the household (Mokyr, 1999). This initial shift in working location caused an increase in specialization and was mainly driven by the high fixed costs of complementary capital. Thereafter, with the general decline of the industrial sector and the rise of the service, government and non-commercial sector, clustering of workers outside the household became more and more beneficial due to the benefits of spillovers between workers themselves. However, with both the adoption of information and communication technology (ICT) and the increase in flexibility of the labor force, workers nowadays are less constrained to work continuously together on the same location. Because commuting costs – temporal, monetary, and even emotional – do not have to be incurred and because working life can be better combined with family life, teleworking seems to be an attractive alternative to working out-of-home. However, although figures of teleworking are scarce and international comparisons difficult to make, the general impression is that the number of teleworkers is still relatively low. Most figures show that within Europe Finland, Sweden and The Netherlands are the countries with relatively the largest numbers of teleworkers.1 Usually, these teleworking figures show large variations between studies, which are partly caused by the definitions used. In this study, we adopt the definition from the Dutch Ministry of Transport, Public Works and Water Management: A teleworker is an individual who works partly at home (or somewhere else than at work) and who uses for that purpose information and communication technology.2 Note that this definition is fundamentally different than that of telecommuting, which we interpret as the actual substitution of the commuting trip (cf. Salomon, 2000). Estimates of teleworking for The Netherlands commonly range from 3.3% (CBS, 2001) to 6% (Steyaert and de Haan, 2001). However, actual percentages of people working at home will not exceed 3–4% of the labor force at any time (see, e.g., De Graaff and Rietveld, 2004). Therefore, combined with the large increase in the diffusion and its corresponding fast decrease in relative prices of ICT applications during the late 1990s, one would expect a large growth in teleworking. Nevertheless, from the datasets we use (SCP, 1996 and SCP, 2001) we measure a 18% increase in the hours per week worked at home between 1995 and 2000, which is substantial but not dramatic. Thus, despite an increasing flexible labor force and an increasing availability of (cheaper) means to communicate, working at home still remains a marginal issue. If a substantial part of the labor force starts teleworking, then profound consequences for living, travel and labor behavior are expected to take place in the long run (Toffler, 1980). In the short run, it is assumed that if individuals telework more they commute less, at least during peak hours. However, the empirical information available does yet not convincingly indicate such a relationship (see De Graaff, 2004). If anything, individuals usually tend on aggregate to travel more per day than less (see, e.g., Van Wee et al., 2002), however this may also be attributed to an increase in income, which seems positively related to traveling. To complicate matters further, income also appears to be positively related with ICT use and teleworking (Vilhelmson and Thulin, 2001). Thus, to relate ICT, travel, and the location of work temporal and monetary constraints need to be incorporated explicitly, to control both for income and for substitution effects between activities. Moreover, such a model of activity participation does not only need to look into the decision between working at home and out-of-home, but also between labor supply and non-labor supply (e.g., leisure) in general, if the preference for non-labor supply is correlated with the decision to work at home. Traditional activity demand analysis (Damm and Lerman, 1981 and Kitamura, 1984) often focuses solely on time constraints. In this paper, however, we choose to model activity demand – in our case labor supply and leisure time – in a microeconomically rigorous framework, by applying microeconomic theory in terms of rational (economic) behavior and utility maximization subject to constraints imposed by both monetary and time budget constraints (see for overviews dealing with both types of constraints, inter alia, Juster and Stafford, 1991, Kraan, 1996 and Bhat and Koppelman, 1999). For our purpose, a more microeconomically oriented approach offers two salient advantages above other activity demand methodologies. First, income effects play a pivotal role in well rooted microeconomic labor supply models, a feature which is strongly supported by empirical research in both teleworking (Vilhelmson and Thulin, 2001 and De Graaff and Rietveld, 2004) and especially in economic labor supply theory (Gronau, 1977 and Deaton and Muellbauer, 1980). Second, the model specification is firmly rooted in an economics structural behavioral framework, which is lacking in a more traditional activity based approach (such as Golob and McNally, 1997, Kuppam and Pendyala, 2001 and Zhang et al., 2005).3 The latter advantage enables us to quantify the relations between the various activities in terms of elasticities. On the other hand, a disadvantage of such a microeconomic approach is the strong emphasis on strict assumptions one has to make (Bhat and Koppelman, 1999). Empirical (economic) estimates of elasticities between teleworking and working out-of-home are rather scarce (see, e.g., Yen, 2000). Therefore, we aim to fill this gap by providing these elasticities for The Netherlands in the late 1990s and by that means look specifically into the question whether working at home and working out-of-home are substitutes or complements, and to what degree. Especially the influence of ICT availability and commuting time on working at home is examined, both in a relative and in an absolute sense. The remainder of this paper is organized as follows. Section 2 provides an exposition of the trade-off individuals make when considering teleworking. In Section 3 we continue with a microeconomic labor supply model dealing with the trade-off between working at home and working out-of-home. Subsequently, we use ICT availability to construct an estimation procedure which is able to construct an implicit price for working at home relative to working out-of-home within the same job. We then use our model of time allocation together with the corresponding estimated prices for working at home, out-of-home and leisure to construct a system of demand equations based on a translog indirect utility function in Section 4. This enables us to estimate the relevant relations between working at home, out-of-home and leisure in Section 5. The last section concludes.

This paper focuses on the relation between commuting time, ICT and working at home. To do so, we develop a theoretical framework where preferences for working at home and out-of-home enter the utility function directly. To construct a full system of demand equations, we first derive an implicit price for working at home, which is relative to working out-of-home. It turns out that for workers at home wages are 19% lower than for workers out-of-home – however, only marginally significant. Moreover, this gap diminishes with 80% when workers have access to an internet connection. With the use of the theoretical labor supply model and the wage equation, we specify a demand system based on the translog utility function, taking working at home, working out-of-home and leisure time directly into account. We account for group-specific differences in preferences for activities by incorporating individual characteristics in each demand equation. With this demand system, we are empirically able to find price and substitution elasticities between working at home and out-of-home. The main results are that (i) working at home and out-of-home act as (slightly imperfect) substitutes, conditional on individual characteristics, and (ii) changes in commuting time and ICT availability have small absolute effects on working at home and out-of-home, respectively. Therefore, working at home and out-of-home seems to be more determined by individual characteristics than by (changes in) commuting time and ICT availability. Age and education, and to a lesser extent sex and job characteristics turn out to be important. Although the results are intuitive (at least for the case of The Netherlands), some further research is yet warranted. The fact that some workers are physically prevented to work at home – because of the nature of their job – can still be accounted for by sector dummies. However, more explicit modeling of the behavior of firms is called for in order to explain why some firms within the same sector allow their employees to work at home and others do not. Another barrier to working at home is that some workers who are able and willing to work at home do not have the required workspace available at home. Therefore, further research is needed to incorporate residential constraints. Moreover, the implicit prices of working at home need more variation. Possibly, some workers may even experience positive productivity effects from working at home (e.g., in the short run). Also, prices are now assumed proportional to the wage rate and the activity involved (working at home and out-of-home). Using a fixed and variable cost component may introduce a much more complex and realistic model. However, for such extended analyses additional information on ICT availability and wages is required. An additional avenue of future research is a direct modeling of maintenance activities, instead of treating them as rest categories. However, as long as maintenance activities can be regarded as derived demand, and given that they are sufficiently accounted for by the incorporation of individual characteristics – such as having children, type of household, etc. – in the demand equations, there is no reason to believe that our present estimations are biased. Related to that issue, the inclusion of trade-offs between working partners within two-earner households might shed some new light on the trade-off between working at home and out-of-home. Finally, we would like to stress that by treating commuting time exogenous we constructed a short-run model. Namely, in the long-run it may well be that workers decide to move residence further from work when increasing their at home labor supply.