1Center for Urban Transportation Research, University of South Florida

Do Variable-Pricing Strategies Influence Activity-Travel Patterns of Carsharing Users? A Case Study

WORKSHOP 196: Using Experimental and Behavioral Economics to Improve the Understanding of Road, Transit, and Parking Pricing

January 13, 2013 l TRB 92nd Annual Meeting, Washington D.C

Sisinnio Concas

2

What is Carsharing?

• Short-term auto rental option• Hourly flat rate• Annual membership fee• No gasoline costs • Unlimited mileage• No parking costs

3

Program Implementation

• Launched at USF main Campus in July, 2009• Four hybrid vehicles

– 3 Toyota Prius ( later changed to 2 Nissan Cubes & 1 Ford Focus) at $7.50/hour

– 1 Ford Escape at $9.00.hour– $30.00 overnight– $70.00 daily

4

Program Participation39

995

4002

6

4005

7

4008

7

4011

8

4014

8

4017

9

4021

0

4023

8

4026

9

4029

9

4033

0

4036

0

4039

1

4042

2

4045

2

4048

3

4051

3

4054

4

4057

5

4060

3

4063

4

4066

4

4069

5

0

5

10

15

20

25

0

20

40

60

80

100

120

140

160

180

200

Nonrenewals Active Members Net Members (Cumulative)

5

Monthly Rentals

Sep-09Nov

-09Jan

-10Mar-

10May

-10Jul-10Sep

-10Nov

-10Jan

-11Mar-

11May

-11Jul-11Sep

-110

20406080

100120140160180

0

500

1000

1500

2000

2500

Rentals by Month

RentalsCumul Rentals

Num

ber

of R

enta

ls

Cum

ulat

ive

Num

ber

of R

enta

ls

6

Member Profile

Do not own a car/truck

Do not want to depend on others for transportation

Save money

Protect the environment

Did not bring my car/truck to campus

0% 20%

40%

60%

80%

Why did you join WeCar?(n=34)

Shopping/errands

Visit friends/relatives

Visit public place: historical site/museum/park/library

Social event

Family personal business/obligations

Attend business meeting/trip

Buy services: video rentals/dry cleaner/post office/car service/bank

Coffee/ice cream/snacks

Pet care: walk the dog/vet visits

Day care

Go to gym/exercise/play sports

Where do you go with WeCar?(n=34)

7

Usage Patterns

Sun Mon Tue Wed Thu Fri Sat0

50100150200250300350400450 Rentals by Day of Week

0

200

400

600

800

1000

1200Rentals by Time Period

8

Participant Selection

• Random selection of 30 Carsharing members

• 15 assigned to treatment (pricing)• 15 to control (no pricing)• Length of Experiment: from February 10 to

April 29, 2011• All given cell phone with TRAC-IT app

9

Interactive Data Collection

TRAC-IT GIS Data StorageDatabase Server

TRAC-ITMobile

ApplicationCell Phone

TRAC-ITSurvey

Participant

Web or Desktop ApplicationsData Viewing/

Analysis

TRAC-ITParticipant

LEGEND

UDP

JDBC

SMTPHTTP(S)

Application Data

Location DataLocation Data

Application Data

TRAC-IT Database ToolkitAutomated/Manual

Batch Data Processing

TRAC-IT Web Application

Web Services

TRAC-IT File

StorageWeb Server

TRAC-ITData Analyst

TRAC-IT Emails

Email Server

POP3/IMAP/Exchange

FTP

TRAC-ITAdministrator /Data Analyst

SOAP

Toolkit Monitors Web Application Application &

Location Data ProcessedData

Personalized Emails

KML, GPX, ICS files

PersonalizedEmails

KML, GPX, ICS files

User Feedback

File URL

TCP

Adaptive Location Data Buffering

10

Communicating with Users

TRAC-IT GIS Database

ServerData Storage Tier

GIS Web or Desktop

ApplicationsClient Tier

TRAC-ITParticipant

Raw GPS Data

Personalized Email withlink to Data

TRAC-IT Toolkit Software

TRAC-IT Web Server

Hosts files for download

Formatted GPS Data(KML, GPX, ICS, etc.)

1)

2)

3)

4) User downloads KML file from server

“This is my trip to campus, via the Bull Runner, to pick up the WeCar. I then drove the WeCar to the CVS on Fowler to pick up medication and then drove to the grocery store on Bears Ave. After shopping, I dropped the groceries off at home and then drove back to campus to return the WeCar. I then took the Bull Runner back home.”

11

Data Collection

12

Trips by Mode and Purpose

Mode Count Percent

Car 666 40.8%

Bicycle 155 9.5%

Bus 132 8.1%

Scooter 18 1.1%

Walking 410 25.1%

Carsharing 31 1.9%

Not Reported 221 13.5%

Total 1,633 100.0%

84; 5% 417; 26%

303; 19%120; 7%

93; 6%110; 7%

211; 13%

28; 2%

267; 16%Dining OutReturn HomeSchool / Religious ActivityShopping / ErrandsSocial / RecreationalTransport Someone Work or Work-RelatedOtherNot Reported

13

Communicating Pricing Changes

• Incentives to encourage off-peak travel• 50-percent reduction in hourly rates• Rate changes communicated by email to

treatment:– Scheduling constraints accounted for– Discount applied on actual rental starting time

14

Hourly Pricing Schedule

Vehicle Rate ($/hour) Time period

Honda Civic $3.75 (50 % reduction) 9:00 a.m. to 11:00 a.m. and 1:00 p.m. to 3:00 p.m. --> Monday through Friday (note that the rate goes back to the base rate between 11 a.m. and 1 p.m.)

Nissan Cube $3.75 (50 % reduction) 9:00 a.m. to 11:00 a.m. and 1:00 p.m. to 3:00 p.m. --> Monday through Friday

Ford Escape $4.50 (50 % reduction)

9:00 a.m. to 11:00 a.m. and 1:00 p.m. to 3:00 p.m. --> Monday through Friday

Implementation Schedule Starting Date Details

Pricing Strategy A March 1 to March 21, 2011 USF to communicate schedule to carsharing on 02/21/2011

Pricing Strategy B March 28 to April 28, 2011 USF to communicate schedule to carsharing on 03/21/2011

15

Before/After Trip Distribution

0:00

0:59

1:59

2:59

3:59

4:59

5:59

6:59

7:59

8:59

9:59

10:59

11:59

12:59

13:59

14:59

15:59

16:59

17:59

18:59

19:59

20:59

21:59

22:59

0.00

0.50

1.00

1.50

2.00

2.50

3.00

No Pricing PricingTime of Day

Den

sity

50 P

erce

nt D

isco

unt

50 p

erce

nt D

isco

unt

16

Spatial Behavior

Trip Length (miles) SDE (square miles) User Type Average Carsharing

TripNon

CarsharingAverage Carsharing

TripNon-Carsharing

TripCarsharing 2.6 8.0 1.7 0.5 0.2 0.5Non-Carsharing 4.2 - 4.2 7.8 - 7.8

Activities Mean Center

SDC

Minor Axis

Major Axis

SDE

𝑥

Y

X

Y

X

17

Rental Cost and Rental Time

Before Variable Pricing After Variable Pricing

18

Discussion (1)

• The research examined changes in travel behavior through pricing mechanisms based on time-of-day and day-of-week hourly rate discounts.

• The study found that overall carsharing users are price sensitive

• Changes in rental rates of the order of 20 to 50 percent have a significantly large impact on daily rentals.

• The analysis shows that carsharing users’ activity space contracts while using carsharing as a mode of transport.

19

Discussion (2)

• User interaction and real-time data monitoring contributed to continued participation throughout the study.

• Participants responded positively to this interaction and expressed a sense of confidence in the GPS tracking tool.

• Further work is needed to test if the pricing response would be the same by penalizing carsharing travel during peak-hour travel (i.e., increasing hourly rates).