**Travel-Time Impacts of Mayor
Giuliani’s
Midtown-Manhattan Crosswalk Barriers**

** **

**Charles Komanoff
8 January 1998
**

**The Barriers Cost Pedestrians 8-10 Times as Much Time
as They Save Motorists**

Barring pedestrians from half of the crosswalks on five avenues (6^{th},
5^{th}, Madison, Park and Lexington) __lengthens crosstown pedestrian
travel times across the affected area (from 6 ^{th} Avenue to Lexington)
by more than a minute and a half (101 seconds, or 16%).__

In comparison, even granting Mayor Giuliani’s (unsupported) claim of
a 20% improvement in vehicle travel times, __the same trip in a motor
vehicle is shortened by only about a minute (64 seconds) in a crosstown
vehicular trip covering the affected area__.

**Thus, the typical pedestrian loses 1 1/2 times as much time as the
typical motorist gains.**

Moreover, crosstown pedestrian traffic vastly outnumbers vehicular traffic — by around 6 or 7 to 1. Thus, the aggregate lengthening of pedestrian travel times exceeds the shortening of vehicular travel times by around 10 to 1. Even if the barriers improved crosstown vehicular speeds by 30%, the total time costs to pedestrians would still outweigh the time savings for motorists by almost 8 to 1.

And whereas the increases in pedestrian travel times reflect physically-lengthened journeys (and associated street-crossing waiting times) and thus would be permanent, the improvements in vehicular travel times are likely to prove transitory as new vehicle trips by opportunistic drivers "use up" the reduction in vehicular congestion. Finally, the barriers force pedestrians to make new street crossings through streams of turning traffic, for a considerable increase in risk for them.

**The bottom line on the barriers: pedestrians lose 8-10 times as much
time as motorists gain, and are exposed to greater danger; and the pedestrian
losses are permanent while the motorist gains are probably transitory.**

**Background**

On Dec. 23, 1997, the New York City Police Department initiated a pilot
program intended to improve crosstown motor vehicle speeds in midtown,
which are notoriously slow. Possible effects on pedestrians, as the Police
Department acknowledges, were not considered. At five avenues crossing
49^{th} and 50^{th} Streets from 6^{th} Avenue
to Lexington Avenue, 3-foot high barriers and an intensive police presence
bar pedestrians from avenue crosswalks so that these can be used exclusively
by turning vehicles.

For example, to cross 5^{th} Avenue, a southbound traffic artery,
pedestrians must use the crosswalk on the north side of the street. To
cross northbound Madison or 6^{th} Avenues, pedestrians must use
the south crosswalk. Thus, half of crosstown pedestrian trips that previously
involved crossing only avenues now require zigzagging across two additional
streets.

The pilot program was originally designed to run until approximately Jan. 5, but Mayor Giuliani has extended it until Jan. 12 or 15.

The program has been wildly unpopular with pedestrians and has been roundly ridiculed in the press. The mayor has staunchly defended it while attacking critics as "anti-car." On Dec. 30, the mayor said the program has improved traffic flow 20-30% and will potentially save lives by permitting emergency vehicles to travel faster.

This paper is intended to remedy the Police Department's failure to
consider the effect on *pedestrian* travel times.

**A Typical Crosstown Pedestrian Trip: Before**

Consider a pedestrian trip along 50^{th} Street from the west
side of 6^{th} Avenue to the east side of Lexington Avenue — a
distance of 2355 feet. (The particular trip length or configuration is
unimportant; the typical trip chosen here is used to derive pedestrian
and vehicular before-and-after travel time *ratios*, which will vary
little if at all for other crosstown trips.) At an assumed walking pace
of 3 mph (4.4 feet/second), the time actually spent walking is 535 seconds.

However, five avenues must be crossed, involving possible waiting time. A typical traffic-signal cycle lasts approximately 85 seconds, of which 52 seconds is green/walk for north-south travel and 33 seconds is green/walk for east-west travel. We estimate that pedestrians typically wait an average of 21 seconds to cross an avenue (see Appendix). This adds 105 seconds (5 x 21) to the journey, for a total of 640 seconds (535 + 105), or 10 minutes and 40 seconds.

The "effective" speed for the pedestrian is 2.51 mph (calculated by dividing 2355 feet by 640 seconds, and converting from feet per second to miles per hour by simultaneously dividing by 5280 and multiplying by 3600).

**A Typical Crosstown Pedestrian Trip: After**

Now consider the same trip with the pedestrian having to detour to the
non-barriered avenue crosswalks. Its duration is increased in two ways:
__greater distances__ reflecting the detours across the side streets;
and __new waiting times__ incurred in crossing those streets.

First, the pedestrian must make six street crossings. Assume the pedestrian’s
journey would ordinarily proceed eastward along the north side of the street
(i.e., from the northwest corner of 50^{th} and 6^{th}
to the northeast corner of 50^{th} and Lexington). The pedestrian
must now cross:

• from the north to the south side of 50^{th} Street, along
the west side of 6^{th} Avenue;

• from the south to the north side of 50^{th} Street, along
the west side of 5^{th} Avenue;

• from the north to the south side of 50^{th} Street, along
the west side of Madison Ave.;

• from the south to the north side of 50^{th} Street, along
the west side of Park Avenue;

• from the north to the south side of 50^{th} Street, along
the Park Avenue median;

• from the south to the north side of 50^{th} Street, along
the west side of Lexington.

Each crossing is approximately 30 feet from curb to curb. Thus, the six crossings add 180 feet to the trip, converting a 2355-foot journey to a 2535-foot journey. At a walking pace of 3 mph, it takes 576 seconds to cover the distance, 41 more seconds than before.

Second, each street crossing involves possible waiting time. We estimate that pedestrians typically wait an average of 10 seconds to cross a street (as opposed to 21 seconds of average waiting time for avenue crossings; see Appendix). The six street crossings thus add 60 seconds (6 x 10) to the journey.

Total travel time is thus the sum of: 576 seconds walking time; 105 seconds waiting time to cross avenues (as before); and 60 seconds waiting time to cross streets. Total travel time is 741 seconds, or 12 minutes and 21 seconds.

Although the pedestrian has walked 2,535 feet, the true (or "net"
or "intended") distance is the previous 2355 feet. Dividing this
by 741 seconds (and converting from feet per second to miles per hour)
gives the new travel speed of 2.17 mph. The speeds are incidental, however;
what matters is travel __time__, which has increased by 101 seconds.

**Summary of Results for Crosstown Pedestrian Travel**

Old travel time for typical trip: 640 seconds

New travel time for typical trip: 741 seconds

Increase in travel duration: 101 seconds, or 16%

**Vehicle Travel Times and A Comparison with Pedestrian
Impacts**

The mayor has claimed that his pedestrian barriers have improved crosstown
vehicular speeds by 20-30%. Although no evidence has been provided to support
this claim, we accept it *arguendo* and use it here for purposes of
evaluation.

Midtown vehicle speeds were reported earlier this month as averaging
6.2 mph. However, crosstown speeds tend to be even slower. To estimate
actual crosstown vehicle speeds on 49^{th} and 50^{th}
Streets, we consulted the "Speed Books" maintained by traffic
expert and engineer Sam Schwartz (the *Daily News*’ "Gridlock
Sam"), which sample and record vehicle travel times and speeds at
hourly intervals on each city block. Based on the most recent measurements,
taken on 12 hourly intervals on four different days in 1995 and 1996, crosstown
vehicle speeds on the two streets average 4.2 mph. A 20% improvement would
produce a gain of 0.84 mph, to a speed of 5.04 mph. (These speeds reflect
waiting time, including at intersections.)

The relevant parameter, of course, is __time__, not speed. At the
old speed of 4.2 mph, a vehicle takes 382 seconds to cover the crosstown
distance of 2355 feet (2355 divided by 4.2, times 3600 divided by 5280).
At the putative 20%-improved new vehicular speed of 5.04 mph, the same
trip in a motor vehicle takes 319 seconds, __a saving of 64 seconds__
(rather than 63, which reflects rounding error).

__In comparison, the same trip by a pedestrian has now been lengthened
by 101 seconds.__ Moreover, crosstown-proceeding pedestrians far outnumber
crosstown-proceeding vehicles. Official figures on relative traffic volumes
are not readily available. However, on Jan. 6, 1998, personnel from Transportation
Alternatives counted pedestrians and motor vehicles per hour entering 5^{th}
Avenue, eastbound, as follows:

From East 48^{th} Street: 3732 pedestrians, 386 vehicles — a
ratio of 9.7 to 1.

From East 50^{th} Street: 3618 pedestrians, 352 vehicles — a
ratio of 10.3 to 1.

Accordingly, we assume a 10-to-1 ratio of pedestrians to vehicles for crosstown travel.

Granting an average vehicle occupancy factor of 1.5, __the ratio of
crosstown pedestrians to crosstown motorists is 6.7 to 1__.

**Thus, for each vehicle occupant saving 64 seconds, 6.7 pedestrians
are losing 101 seconds each, for an aggregate loss of 675 seconds — slightly
over 10 times as much.**

(Parallel calculations for an assumed 30% improvement in vehicular speeds yield a ratio of pedestrian losses to motorist gains of 8 to 1, calculated by substituting average motorist gains of 88 seconds for the 64 seconds used above.)

It is true that some vehicular traffic — freight and, especially, buses — has a high economic value. (The mayor’s claim that his experiment may save lives by expediting emergency vehicles is purely hypothetical and is unverifiable without empirical data on lives lost due to traffic delays.) But much pedestrian traffic has a high economic value as well — consider the shopper approaching Tiffany’s or the attorney walking to a meeting with a client or another attorney. Moreover, pedestrians are less well equipped than motorists to withstand delays. Not only do pedestrians lack shelter from cold, rain, etc., as well as the greater diversions and comforts available to motorists; the very essence of walking is its directness in proceeding from Point A to Point B — a quality uniquely abridged by the barriers.

Finally, police officials assert that the barriers improve pedestrian
safety by separating walkers from turning cars. This claim plainly inverts
reality since, as shown here, the barriers roughly double the number of
intersection crossings pedestrians need to make. For every *avenue*
crossing supposedly made safer by keeping turning cars separate (an option
previously available, in any event to pedestrians, who could always choose
the long way around if they felt safer that way), pedestrians must now
perform a new *street *crossing with vehicles turning from avenues
onto the side street.

The Police Department does not appear to have noticed that it could
improve pedestrian safety at every NYC crosswalk by visibly and stringently
enforcing the pedestrian’s legal right-of-way over turning vehicles. The
advent of the new year would be a propitious time for the mayor to move
boldly to reduce pedestrian endangerment by motor vehicles, perhaps beginning
with release of the annual but long-delayed *Traffic Fatalities in New
York City* reports for 1995 and 1996. An estimated 226 pedestrians were
killed by motor vehicles in New York City in 1996; the public needs and
deserves official statistics as a basis for remedial action.

**Appendix: Estimating Pedestrian Waiting Times for Street
and Avenue Crossings**

Typical midtown traffic-signal cycles are 50-54 seconds of green (walk + flashing don’t walk) for north-south travel and 30-36 seconds of red (steady don’t walk). We assume 52 seconds of green for north-south and 33 seconds of green for east-west, with cycle totals of 85 seconds.

A pedestrian crossing an avenue and approaching the curb observes either a walk signal (including the flashing don’t walk) or a don’t walk signal. In 52 out of 85 crossings, the signal will be don’t walk. Average waiting time in this circumstance is 26 seconds (one-half of the don’t-walk duration of 52 seconds).

During the 33 out of 85 crossings in which the signal reads walk when the pedestrian approaches, the pedestrian will walk unless the walk signal is near the end of the cycle. We assume the pedestrian will walk so long as there are 8 or more seconds remaining in the cycle. Thus, in 25 out of 85 crossings there is no waiting time. In the remaining 8 out of 85 crossings, the pedestrian must wait an average of 4 seconds for the signal to turn to red (steady don’t walk), plus 52 seconds for the red cycle to conclude.

Arithmetically, typical waiting time for crossing avenues is thus the sum of:

• 52/85 (probability of red) x 26 seconds (average waiting time); plus

• 25/85 (probability of enough green to proceed) x 0 (no waiting time); plus

• 8/85 (probability of green but "too late to cross") x 56 (4 + 52) seconds waiting time.

__The result is: 21 seconds average avenue-crossing waiting time.__

Waiting times are less for __street__ crossings, since north-south
movement has more green (walk) than red (don’t walk). The calculation follows
the same format:

• 33/85 (probability of red) x 16.5 seconds (average waiting time); plus

• 44/85 (probability of enough green to proceed) x 0 (no waiting time); plus

• 8/85 (probability of green but "too late to cross") x 37 (4 + 33) seconds waiting time.

__The result is: 10 seconds average street-crossing waiting time.__

Note that we have not considered the possibility that pedestrians may be able to reduce waiting times by choosing street and avenue crossings in accordance with traffic signals. This could attenuate somewhat the worsening of travel times calculated here. On the other hand, we have probably understated waiting times by assuming that the walk and don’t walk signals are randomly distributed. In fact, consecutive street and avenue crossings, of which there are five pairs for a crosstown walk across the affected area, are linked; one who has just crossed a street, taking an estimated 7 seconds to do so, then faces 26 seconds of waiting time before obtaining a walk signal to cross the avenue, rather than the 21 seconds we derived based on random signals.

At least as importantly, we have not considered that increased pedestrian congestion on crosswalks and sidewalks would reduce walking pace from the assumed 3 mph. Avenue crosswalks will be doubly congested since half have been taken out of service. And street crosswalks and sidewalks will also be more congested, since the same number of pedestrians are now having to cover greater distances for the same journeys.