were found during the first 10 years after the capacity increase. This means that if a segment's capacity is increased by 10 percent, traffic on that segment will increase by 3 to 4 percent during the first 10 years. After approximately 20 years, the elasticities were found to increase to 0.4 to 0.7.

The authors noted an important limitation of the study of individual segments: only traffic levels on the improved segments were studied. They stated “ ... clearly, any additional traffic on the improved segment must also use other links on the roadway network as well. Further, a large proportion of the additional traffic may have diverted from other routes. These complement-substitute relationships between different links in a road network imply that if a change to one link has a substantial traffic impact on that link, other links are likely to be significantly affected as well” (Hansen et al. 1993, 3,4).

Elasticities such as those developed by Ruiter et al. and Hansen et al. provide a simple means for summarizing study findings regarding the effects of increases in highway capacity on travel. Hansen et al. measured changes in capacity in terms of lane miles, and Ruiter et al. measured changes in capacity in terms of vehicle miles of capacity.¹ One advantage of using the changes in lane miles or vehicle miles of capacity to represent improvements in capacity is that these measures can often be compiled for a given area directly from highway system inventories maintained by state departments of transportation.

However, these measures also have a number of limitations that greatly affect the transferability of elasticities to other areas or capacity improvements. Most important, the amount of time savings produced by a given change in lane miles or vehicle miles of capacity is highly variable, depending on such factors as preexisting levels of congestion and bottlenecks. Consider, for example, a congested bridge that is a traffic bottleneck during peak periods. Widening the bridge could provide large peak-period time savings with a small increase in lane miles or vehicle miles of capacity. Conversely, adding lanes to a facility that is not currently congested will have a small effect on travel time, even though the addition may represent a significant increase in lane miles. In addition, the construction of new limited-access highways (with much higher free-flow speeds than the facilities they replace) will have travel time effects that are not well represented by the change in capacity. Finally, the use of elasticities must take into