Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 52
4 Summary and Conclusions SUMMARY OF RESULTS This study has examined mode choice and average private vehicle occupancies for the ~ourney-to-worx through the analysis of both cross-sectional data for the largest metropolitan areas in the US and through a time-series analysis of peak period cordon count data for New York City. Static cross-sectional models of mode share and occupancy were estimated for 1990 for the primary sub-regional commute flow markets served by transit, namely the suburb to central city commute and the within central city commute. Models of changes in mode shares and occupancies between 1980 and 1990 were also estimated for the total .. . . . .. . .. . . ~ . .. commute market. In each case, separate models were estimated for the market share of single occupant vehicle, and the market share of public transit. The time series models for New York City were estimated for the period 1978-1994. The results of the analysis indicate that development patterns, vehicle availability, and price and service levels are all important determinants of mode shares and occupancies. Specifically, the models indicate that transit service levels can have a significant impact on SOV shares, and that downtown parking prices are also a powerful deterrent to driving alone. The time series models of SOV share for New York City also indicate that in this highly transit-competitive market, SOV share is sensitive to automobile operating costs in the form of tolls and gasoline prices. CONCLUSIONS The modeling results presented in Chapter 3 would seem to confirm the hypotheses developed in Chapter 1. The dispersion of residential and workplace locations, combined with smaller household sizes, have reduced the opportunities to carpool, causing occupancies to decline. At the same time, the ever increasing availability of private vehicles has caused SOV shares to increase. We have cautioned throughout this report that the relationships of interest in a study such as this are inherently complex, and the ability to assign causality definitively is therefore limited. Notwithstanding this caveat, however, it is also an objective of this study to be of practical value to policymakers by informing

OCR for page 52
our understanding of at least the most important factors in the variations and trends that we have observed. In this vein, we draw the following conclusions: 1. Public policies a/one are not the culprit The modeling results confirm our assertion in Chapter ~ that much of the trends in market shares and occupancies that have been observed in the last twenty years can be explained by factors that are beyond the direct control of transportation policymakers. Few would argue, for example, that the decline in household sizes that appears to have made carpooling less attractive is the direct result of transportation-related policy, given that the causes of this decline stem primarily from changes in family structure. Likewise it has been argued that national public policies that legislate (for example) tax incentives for home ownership have provided the impetus for the large migration of population to the suburbs described in Chapter i. But at the same time we also know that rising incomes also make home ownership possible, and given that home ownership is also possible in the central city, more pervasive lifestyle choices are certainly a key factor in the desire of many for suburban . . wing. Also, employers move to the suburbs because of economic factors - there are more people in the metropolitan areas now, they are increasingly living in the suburbs, and it is often cheaper and more competitive for employers to locate there 2. But public policies do appear to matter On the other hand, there is some evidence from our results to indicate that variables within the control of policymakers may have been an important influence in the trends that we have studied, and that they could be an even more important influence on these trends in the future. For example, the models provide evidence that HOV facilities actually do seem to encourage ridesharing, rather than just offer a convenient faster alternative for the few carpools that 53

OCR for page 52
already exist. Our results imply that, other factors being equal, building more HOV facilities will encourage additional ridesharing. The results also indicate that the transit service level can help reduce the amount of single-occupant vehicle commuting. Of course, this may reflect the obvious supply-side effect that places with a lot of transit have a lot of transit riders. But on the other hand, it may also indicate that increasing service levels in a given metropolitan area could potentially divert additional trips from single occupant vehicles. Not surprisingly, transit fares were not found to be a significant determinant of market shares in the cross-sectional models. This is consistent with the fact that although fare changes may incite political controversy, the evidence is that shares are relatively insensitive to fare levels. In addition, the time series results for New York City imply that in a transit- competitive environment, prices do matter. The mode} results indicate that at least 50% of the variation in the observed SOV share can be explained by changes in relative prices, accounting for both of the cost of the private vehicle mode as well as the cost of competing transit service. The static 1990 models also inclicate that parking prices are an important factor in explaining transit market share. Taken together, these results have some interesting implications for the efficacy of some specific policies, which we summarize below: . . Focus on transit service levels, not fares. In Chapter 4 of our Project H- 4A report, we reviewed the literature on transit demand and provided some generalizations based on what this understanding implies for transit policy. We specifically recommended that transit managers focus their attention on service levels as a more important determinant of ridership than fares. The results from the current study reaffirm this assertion, and confirm the evidence that patronage is more sensitive to service than it is to fare levels. Parking pricing policies deserve continued" attention. While the data for parking prices used for this study are necessarily both incomplete and deficient in quality (we could not get a historical time-series), the results nonetheless support the contention that parking policy is a very important factor in the mode shares we currently observe. The causality of the relationship between HOV lanes and ridesharing may of course be more complex. It may be true that places with the worst congestion and the highest density are most likely to have HOV lanes, perhaps built in response to these factors. These factors may in turn also make carpooling rates higher in these places irrespective of the amount of HOV facilities. 54

OCR for page 52
· Road pricing may be effective in certain areas. We can make a few generalizations from the results of this study. First, while the elasticities of trios and SOV share with respect to tolls and fuel vices indicated bv the ~ ~ 1 J · . . . . . . . ~ . . _ _ . . Ah. time series models are relatively low, they are statistically significant and thus at least in a New York-like environment of good quality alternatives, drivers do respond to price signals. The results of the cross-sectional models further confirm that parking prices can affect mode shares across a wide array of metropolitan areas. To the extent that parking prices represent an important (and in some places, large) part of the total cost of commuting by private vehicle, these results provide further evidence for the potential efficacy of road pricing policies. 3. And the outlook for the future is not all bad While the introductory discussion in Chapter ~ may paint a somewhat dismal future of commuting in America, with everything seemingly trending inevitably toward an all-auto society, there is already evidence that some of these underlying trends may be slowing, or may not continue into the future. This is particularly good news, considering that we have argued that it is in large part these underlying socioeconomic and demographic forces that are ultimately compelling the shift away from high occupancy modes and the consequent rise in vehicle trips and congestion. While the data of Chapter ~ show that for some time household sizes have been declining, again in response to underlying demographic changes, a reexamination of Table ~ shows that total workers per household has in fact remained more or less constant, actually increasing slightly in 1990. In fact, as Pisarski has observed, 1990 has signaled the end of the "worker boom". Figure 5 shows that growth in the labor force has already slowed dramatically from its rate in the 1970s, and it is expected to grow even more slowly in the future. This is an important result, given that the VMT model presented in Table 22 indicates that employment is one of the primary deterrrunants of total vehicle travel. Figure 6 shows that average household sizes may well have stabilized at around 2.6 persons per household for the foreseeable future. Given this factor's importance in providing opportunities for carpooling, and therefore in determining vehicle occupancies, this leveling off is also encouraging. * A more thorough examination of future economic demographic and social trends and their implications for the mode share and use of public transit in particulars can be found in TCRP Report H-4B, Transit Markets of the Future The Challenge of Change (cited earlier). 55

OCR for page 52
Figure 5. Percent Change in Labor Force and Total Population, 1960-2000 25% ~5 3 20% a ._ Q , 15% o 10% A.... . ·. .. / / .. .. .. ... . . . . . . :. ...... ~ ...... 5% l l 1960 1970 1980 Source: US Bureau of the Census, 1996. i .., - \ \ \ .. <, . ................... \ \ ........... :.:.:.:.:: :-: ::::: :0 1 990 2000 Figure 6. Average Household Size 1970-2010 3.2 3.1 - _ 3.0 2.9 2.8 - _ 2.7 2.6 2.5 , 1 , , \ - - - - 1970 1980 1990 2000 2010 Source: US Bureau of the Census, 1996. Labor Force Population ................. ; There is also evidence that the seemingly all-important factor of vehicle availability may be reaching a measure of saturation as well. We have shown in Table 2 that vehicles per capita and vehicles per household have both increased markedly in the last two decades, and we also know that the growth in total vehicles far exceeded the growth in population during the 1980s. On the other 56

OCR for page 52
hand, the number of vehicles grew more slowly than the number of workers during this same period, and in fact the average number of vehicles per worker actually declined slightly from i.34 in 1980 to 1.32 in 1990.i The fact that this ratio is greater than one also indicates that on average, every worker who wants to commute by pnv ate vehicle now has a vehicle available to do so. Perhaps most compelling, though, is the fact that results of the most recent wave of the NPTS indicate that the total number of vehicles in the US now exceeds the number of licensed drivers. Table 24 illustrates this trend, a clear indication of the saturation of vehicle availability. Table 24. Vehicles Per Licensed Driver 1977-1990 . ~. ~ . ~ ~ ~ ~.~ ~. ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . ~ ~ ~ it. ~ ~ ~ ,~ i: ~ - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Year ~ ~ ~ ~ ~ ~ Vehicles~per~ ~ ~ . ~ ~ ~ ~ ~ ~ . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . ~ ~ ~ ~ . . ~ ~ ~ ~ ~ ~ ~ . .~ ~ ~ ~ ~ .~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . ~ ~ . . .. ~ ~ ~ ~ ~ ~ ~ - ~ ~ - ; - ~ ~ . . ~ ~ ~ ~ ~ ~ . ~ ~ ~ ~ ~ . ~ ~ ~ __ ~ __ ~ ~ ~ I, _ ~ ~ ~ ~ ~ . . . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~.~ ~ ~ ~ ~ ~ ~ ~ ~ . ~ ~ ~ .~ ~ At.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 1977 0.94 1983 0.98 1990 1 .01 l Source: US Department of Transportation, Nationwide Personal Transportation Survey. RECOMMENDATIONS FOR FURTHER RESEARCH While we have stressed the inherent complexity of any of analysis of this nature, the evermore urgent problems of congestion, air pollution, and decaying infrastructure that plague our cities today require that policymakers have some guide to understanding the potential drivers of mode share and occupancy levels that have been contributing to these problems. As such, research of these issues, despite their complexity, remains very important. It is with this in mind, that we offer the following recommendations for further research: · Analysis of the 1995 NPTS data. While the NPTS data are best used for national-level analysis, the latest recent wave of this survey will be very important in examining the trends in vehicle ownership and use that we have descnbed. A brief review of summary results from the 1995 NPTS suggests that the interpretations in this report remain valid today, and are consistent with our conclusions about the future of the trends examined (the results show that vehicle availability is saturating and that occupancies may have stabilized, for example). More detailed examination of these tiara may provide further evidence of the stabilization of these trends. 57

OCR for page 52
. . Continued enhancement of the Census Journey-to-Work data. Although current fiscal realities may make expansion of the Census unlikely, perhaps some consideration could be given to a rethinking of Census priorities to emphasize higher quality transportation-related data. The inclusion in 1990 of questions on departure time to work were very helpful, and information questions on time returning from work and on work trip distance would also be very useful. In adclition, the question of geographic definitions could be addressed more systematically to allow easier comparisons of the data over time. While we have explained that changes in geographic definitions are often logical and necessary, future Census data could perhaps be presented using both the previous and updated definitions (in a manner akin to double-basing indices for the first few years after introducing the new base). Finally, the metropolitan area portions of other complimentary sources such as the NPTS and American Housing Survey could be strengthened and coordinated to help amplify data collected with the existing Journey-to-Work questions. Examination of parking costs over time. We were not able to include parking costs in the time-series analysis, nor were we able to examine changes in parking costs from lL980 to 1990. Given the potential importance of parking costs as a determinant of commuting mode choice, this type of analysis could benefit greatly from the development of time series data on parking costs in the major metropolitan areas. To this end, the federal government could recommend to metropolitan planning organizations a standardized method of collecting parking price data, that could perhaps be incorporated as part of the planning guidelines. More thorough examination of the elects of transit service levels. As we describe in Chapter 2, there is an inherent identification problem that makes modeling the relationship between transit service levels and transit demand problematic. Given that we have in our analysis found evidence that transit service levels may be important in mode choice decisions, but were unable to identify the independent effects of transit service levels in several market segments, additional analyses, perhaps using a more sophisticated simultaneous equations framework, could be very useful. Further, examination of the effect of transit capacity, particularly during the peak period when high load factors can cause marked changes in service quality, might provide additional important insights. The assembly of consistent data across many metropolitan areas would greatly facilitate such a study. Examination of household travel behavior surveys conducted by MPOs. This potential source of cross-sectional data was not explored in this study, 58

OCR for page 52
but these surveys can provide detailed work and non-work travel information. Since a number of cities have done these surveys relatively recently, and MPOs have begun to perform them as longitudinal studies, the source might eventually provide an important additional resource for analyses such as those described in this report. As described in Chapter i, the constraints of this study prevented us from performing a more detailed and thorough analysis of trends in mode shares at Suburb-regional levels of geography. Such an analysis might involve comparing shares for specific commuting flows such as the suburb to central city market for specific metropolitan areas over time with identical geographies. This would require the aggregation of the geographic areas from place level data in order to compute the mode shares, a considerably more intensive effort than was possible in this study. Nevertheless, this more detailed analysis might provide even more insight into these important issues. Finally, the extension of these types of analyses from central cities to areas such as suburban employment centers might further our understanding of the impacts of the reorientation of commuting patterns. Pisarski, A., op cit., p. 33. 59

OCR for page 52