Popular in Professional

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

Education

The Role of Education in Reducing Racial Inequality

University of Southern California Pedro A. Noguera

Law

Diversity and Social Mobility in the UK legal profession

Law Society I. Stephanie Boyce

Internet Law

The Law of AI for Good

University of San Diego Orly Lobel

Pharmacy

Cytotoxic Activity and Anti-Migratory Effect of Berberine-Phytantriol Liquid Crystalline on Non-Small-Cell Lung Cancer

University of Technology Sydney Kamal Dua

Human Rights Law

Transforming Restorative Justice for Transitional Settings

University of Nottingham Kerry Clamp

Education

The Impact of Ride-Hailing Services on Congestion: Evidence from Indian Cities

View more insights in Professional - Business - Information Systems - Operational Research

View more figures from Indian School of Business

View the page of Deepa Mani

Read the Study

Deepa Mani discusses how the absence of ride-hailing services affects congestion levels in three major cities in India, a market where most ride-hailing drivers participate full time.

Image courtesy of interviewee. September 11, 2024

Generating Workspace

Export as PDF

This video discusses the rapid growth of the ride-hailing market, particularly in India, highlighting its expansion to approximately $160 billion by 2023. It emphasizes the need to study the socioeconomic impacts of ride-hailing services, which have provided benefits such as flexible work opportunities and increased labor pool diversity. However, there are concerns about the negative impacts on congestion, especially as ride-hailing platforms operate as commercial taxi services rather than private ride-sharing.

The study seeks to address two key gaps in the literature: the limited causal evidence linking ride-hailing to congestion and the contrasting findings of previous studies, with some indicating that ride-hailing reduces congestion while others suggest it exacerbates it. The study exploits recent driver strikes to analyze congestion in major Indian cities like Delhi, Mumbai, and Bangalore, using various empirical methods.

Findings reveal that congestion decreased significantly during strike periods, especially in highly congested areas, indicating that ride-hailing services contribute to traffic congestion. This video study's results show that the congestion costs associated with ride-hailing in India may be much higher than those found in Western studies, suggesting that unique economic and regulatory factors in India lead to greater deadheading – instances of cabs driving around empty.

The authors also identify other mechanisms influencing congestion, such as drivers taking shorter routes to avoid traffic and a shift of commuters to public transport in the absence of ride-hailing services. The study concludes that managing ride-hailing supply and improving public transport are vital in minimizing congestion, recommending further welfare analysis for comprehensive policy implications.

Language:

Export as PDF

Export as PDF
Here are five key points from the video:

1. **Growth of Ride Hailing Market**: The ride hailing market is rapidly expanding, with a valuation of approximately $160 billion in 2023. This market has potential for socioeconomic benefits, including flexible work opportunities and the inclusion of women and marginalized groups in the labor force.

2. **Unique Convideo of India**: India is highlighted as a crucial area for studying the impacts of ride hailing services due to its relatively under-investigated market, rapid growth, and increasing user base, with significant participation from platforms like Uber and Ola.

3. **Empirical Analysis of Congestion**: The study explores the causal relationship between ride hailing services and congestion, addressing gaps in literature. The findings reveal that during driver strikes, travel times decreased significantly, suggesting that ride hailing services contribute to congestion, particularly in busy urban areas.

4. **Operational Differences and Regulation Impacts**: Ride hailing platforms in India operate as commercial taxi services rather than private ride-sharing outlets. This structural difference leads to issues such as increased deadheading (drivers operating without passengers), which exacerbates congestion. The need for effective regulation to manage driver supply and improve service efficiency is emphasized.

5. **Policy Recommendations**: The study suggests several policy recommendations to mitigate congestion, including enhanced public transport services, improved pricing strategies for ride hailing to manage supply during peak times, and a critical examination of existing regulations to balance the benefits and costs associated with ride hailing operations.

Export as PDF
Here are some key quotes from the video discussing the ride-hailing market and its implications in India:

1. **Market Size and Growth**:
- "This was rooted in the observation that the ride hailing market is a rapidly expanding and growing one. As of 2023, this was at about $160 billion."

2. **Socioeconomic Benefits**:
- "A large stream of literature has also documented various socioeconomic benefits of these services, from flexible work to an expansion of the labor pool to include women and other groups that face barriers in traditional work."

3. **Research Focus**:
- "We wanted to examine this phenomenon in India, where the growth of the market is significantly higher than that worldwide."

4. **Convideo of the Study**:
- "Both given the salience of the market, its relevance and the fact that it was under investigated set the convideo for this study."

5. **Impact on Congestion**:
- "Ride sharing platforms have been implicated overall in worsening congestion... and that's reflected in several policies such as congestion pricing plans... that have been passed in cities like Seattle and Chicago."

6. **Causal Relationships**:
- "There is limited evidence, especially causal, of the relationship between ride hailing and congestion."

7. **Examination of Strikes**:
- "The strike allowed us to exogenously estimate the impact of the absence of these services on congestion."

8. **Findings on Travel Times**:
- "We find a robust finding, which is that travel times were consistently lower during the periods of the strike."

9. **Congestion Costs**:
- "We found the effect sizes, congestion costs of about $2 million per day."

10. **Implications for Policy**:
- "Management of supply is critical to the observed impacts on congestion... it becomes important to explore, given the benefits of the current set of regulations, what the costs are."

11. **Recommendations**:
- "Improving incentives to make public transport more effective. This could be by adding routes, it could be by reducing fares, and improving the coverage of public transport."

These quotes reflect the video's discussion about the ride-hailing market's growth in India, its socioeconomic implications, the concerns regarding congestion, and the necessity for further research and policy recommendations.

Export as PDF
Here are some articles that share similarities with the video, focusing on topics related to ride-hailing markets, their impacts, and related socioeconomic aspects, particularly in the convideo of India and other growing economies:

1. **"The Impact of Ride-Hailing Services on Urban Congestion: Evidence from Indian Cities"**
- This article examines how ride-hailing services like Uber and Ola have affected traffic congestion in major Indian cities, exploring both positive and negative consequences.

2. **"Dynamics of the Indian Ride-Hailing Market: Opportunities and Challenges"**
- This piece discusses the rapid growth of ride-hailing services in India, highlighting the technological innovations being utilized and the socioeconomic benefits for different demographic groups.

3. **"Evaluating the Socioeconomic Effects of Ride-Sharing in Developing Economies"**
- An analysis that focuses on the socioeconomic benefits of ride-sharing services such as job creation and mobility access, particularly how these platforms empower underrepresented groups.

4. **"Ride-Hailing Services in Urban India: A Double-Edged Sword?"**
- This article investigates the dual impact of ride-hailing services on urban infrastructure, looking at how they can both alleviate and exacerbate traffic problems in rapidly urbanizing areas.

5. **"Public Transport and Ride-Hailing: Competition or Complement?"**
- This review explores the relationship between public transport systems and ride-hailing services, discussing how they can work together or compete, especially in cities with developing transportation infrastructure.

6. **"Understanding Congestion Pricing and Its Impact on Ride-Hailing Operations"**
- This article discusses various congestion pricing measures implemented in urban areas and how they affect ride-hailing platforms and traffic patterns.

7. **"Drivers of Change: The Role of Regulation in Shaping India's Ride-Hailing Landscape"**
- An exploration of how regulatory frameworks in India influence ride-hailing operations and their broader socioeconomic implications.

8. **"From Private Rides to Public Taxis: The Evolution of Ride-Hailing in India"**
- This piece discusses the transformation of ride-hailing platforms in India from private-sharing models to accepted commercial taxi operations and its implications for congestion and urban mobility.

These articles could provide additional insights and convideoual understanding relevant to the discussion in your video.

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.195 --> 00:02.285
This was rooted in the observation

00:02.435 --> 00:04.805
that the ride hailing market is a rapidly

00:05.075 --> 00:06.485
expanding and growing one.

00:06.945 --> 00:11.845
Uh, you know, as of 2023, this was at about $160 billion.

00:12.705 --> 00:16.045
Um, and the size of the market also spoke

00:16.045 --> 00:17.285
to the promise of the market.

00:17.785 --> 00:19.965
Um, you know, ride ha platforms.

00:20.265 --> 00:24.525
Uh, the, you know, the, the intent to leverage a whole host

00:24.545 --> 00:27.365
of tech enabled innovations that we outlined in the paper

00:27.905 --> 00:30.285
to push towards sustainable mobility, right?

00:30.285 --> 00:32.685
These could be interventions like dynamic pricing,

00:33.125 --> 00:35.365
efficient matching, pooling of rides, all of that.

00:36.225 --> 00:38.205
And a large stream of

00:38.845 --> 00:42.965
literature has also documented various socioeconomic

00:43.365 --> 00:46.885
benefits of these services, from flexible work

00:47.145 --> 00:49.645
to an expansion of the labor pool to include women

00:49.705 --> 00:52.005
and other groups that face barriers in

00:52.005 --> 00:53.525
traditional, work.

00:54.545 --> 00:57.245
Uh, and, you know, reduction of driver, moral hazard.

00:57.245 --> 00:59.125
There's a whole bunch of socioeconomic benefits

00:59.125 --> 01:00.165
that have been documented,

01:01.105 --> 01:05.445
and we wanted to examine this phenomenon in India

01:05.695 --> 01:09.045
where the growth of the market is significantly higher than

01:09.045 --> 01:10.045
that worldwide.

01:10.305 --> 01:13.085
Uh, for instance, in, for Uber,

01:13.295 --> 01:15.925
India is the third largest market by volume globally,

01:16.145 --> 01:18.685
and one of the fastest growing markets, right?

01:18.745 --> 01:21.005
It has over a million users on its platform.

01:21.555 --> 01:24.485
It's clocked to revenue of whatever, nearly, three 50

01:24.505 --> 01:26.765
to $400 million in 2023.

01:27.545 --> 01:29.805
And therefore, you know, it was, we,

01:29.805 --> 01:31.165
we thought it would be useful

01:31.225 --> 01:34.005
to study these impacts in the Indian context,

01:34.135 --> 01:35.645
which is under investigated.

01:36.025 --> 01:39.765
Um, so both given the salience of the market, its relevance

01:40.025 --> 01:42.925
and the fact that it was under investigated set

01:42.925 --> 01:44.005
the context for this study.

01:44.385 --> 01:49.045
The transportation infrastructure is not terribly mature.

01:49.945 --> 01:54.085
Um, and, you know, for instance, incentives to use public

01:54.725 --> 01:57.165
transport, present trade-offs in terms of quality

01:57.305 --> 01:58.925
of service, et cetera.

01:59.585 --> 02:03.085
Um, and also, you know,

02:03.275 --> 02:07.525
what the ride hailing platforms have also done over time is

02:07.615 --> 02:09.725
adopt to the idiosyncrasies of the market.

02:10.505 --> 02:12.765
Uh, for instance, they don't operate

02:12.765 --> 02:14.565
as private ride sharing platforms.

02:14.565 --> 02:17.165
They operate as commercial taxi operators here.

02:17.785 --> 02:21.125
And therefore, that brings into play, you know,

02:21.125 --> 02:23.325
while it provides benefits such as safety

02:23.905 --> 02:26.325
and benefits to, the supply,

02:26.425 --> 02:27.925
the labor supply, et cetera.

02:28.305 --> 02:29.925
Uh, however, there are costs

02:29.995 --> 02:31.925
that it imposes, which is what we study.

02:32.145 --> 02:35.765
Uh, in the paper ride sharing platforms have been implicated

02:36.035 --> 02:37.765
overall in worsening congestion.

02:38.065 --> 02:39.645
Uh, and, you know, and,

02:39.645 --> 02:43.085
and that's reflected in several policies such as congestion,

02:43.085 --> 02:44.925
pricing plans, et cetera,

02:44.925 --> 02:47.605
that have been passed in cities like Seattle and Chicago

02:47.825 --> 02:49.565
and, New York City, et cetera.

02:49.575 --> 02:51.125
There were two gaps that we sought

02:51.145 --> 02:52.405
to address in the literature.

02:52.425 --> 02:55.245
One, there is limited evidence, especially causal

02:55.245 --> 02:57.285
of the relationship between ride hailing

02:57.285 --> 02:58.725
and congestion. This is

02:58.725 --> 03:00.715
Driven by essentially paucity of data

03:00.975 --> 03:02.475
and a whole bunch of empirical

03:02.795 --> 03:03.995
identification challenges, right?

03:04.015 --> 03:07.995
So that's one. Uh, the second is that, you know, there is,

03:08.005 --> 03:09.795
there are these countervailing impacts

03:09.825 --> 03:11.155
that the literature talks about.

03:11.415 --> 03:14.515
One class of studies says these services reduce congestion

03:14.585 --> 03:18.195
because they allow for demand pooling, the, and,

03:18.195 --> 03:20.595
and reduce the pressure on transportation infrastructure.

03:20.825 --> 03:23.755
They act as a compliment to public transport,

03:23.975 --> 03:25.995
and they reduce private ownership of vehicles.

03:26.455 --> 03:28.835
Uh, however, there's a whole bunch of recent studies,

03:28.845 --> 03:31.635
especially ones that try to, you know, pre,

03:31.635 --> 03:33.395
present causal, evidence,

03:33.395 --> 03:35.315
which show a different viewpoint,

03:36.015 --> 03:40.035
and they stock about deadheading substituting,

03:40.035 --> 03:42.515
public transport rather than complementing it

03:42.515 --> 03:44.875
and inducing trips, which I think is relevant to India

03:44.875 --> 03:46.875
as well, that would otherwise never have been made.

03:47.155 --> 03:48.395
Remember I said that, you know, the transportation

03:48.395 --> 03:51.275
infrastructure is not terribly mature, and

03:51.275 --> 03:53.675
therefore, perhaps these trips would not have been made, and

03:53.675 --> 03:55.795
therefore there was this theoretical ambiguity

03:55.825 --> 03:58.915
regarding the impact of right haing on congestion,

03:59.165 --> 04:02.035
which made it an interesting empirical question.

04:02.295 --> 04:04.955
And that's what we, you know, that's the gap,

04:04.955 --> 04:06.115
that we sought to address.

04:06.215 --> 04:08.555
And like I already mentioned, in addition

04:08.555 --> 04:11.595
to a reliable measure, of the causal relationship,

04:11.895 --> 04:12.995
the Indian context

04:13.535 --> 04:16.715
and ascendant economies in general are under investigated

04:16.775 --> 04:17.795
or understudied, right?

04:17.795 --> 04:20.315
They've all, all empirical studies have almost exclusively

04:20.315 --> 04:21.475
focused on the United States.

04:22.015 --> 04:23.035
Um, and, and

04:23.035 --> 04:26.155
therefore, it, it, you know, we thought it was,

04:26.215 --> 04:28.635
it would be useful a, a, a useful contribution

04:28.635 --> 04:31.275
to study this in the context of, India.

04:31.495 --> 04:33.395
The data was kind of serendipitous

04:33.395 --> 04:35.795
because, you know, there was, there was a lot

04:35.795 --> 04:40.035
of distress from so Uber, they were offering a certain,

04:40.035 --> 04:41.555
compensation to their drivers.

04:41.705 --> 04:43.715
They were taking off a certain commission fee,

04:43.715 --> 04:45.635
which they amped up, you know, over time.

04:46.015 --> 04:49.555
And the distress from reduced earnings, let both Ola

04:49.555 --> 04:51.595
and Uber drivers, which together, you know,

04:51.595 --> 04:55.595
control about 70% of the ride daily market in India,

04:55.615 --> 04:57.675
to go on a strike in New Delhi

04:57.735 --> 05:01.195
and Bangalore, somewhere in 2017,

05:01.215 --> 05:02.875
in the early part of 2017.

05:03.615 --> 05:05.835
And then the last reported strike,

05:05.835 --> 05:06.835
there was a ripple effect.

05:06.835 --> 05:10.155
And the last reported strike took place in Mumbai in

05:10.155 --> 05:11.355
October, 2018.

05:12.175 --> 05:15.595
So now for three different time intervals in these three

05:15.605 --> 05:18.595
major Indian cities spread over a period of a year

05:18.615 --> 05:22.955
and a half, you know, the strike allowed us to,

05:23.535 --> 05:27.475
allowed us to exogenously,

05:28.035 --> 05:32.075
estimate the impact of, you know, the absence

05:32.075 --> 05:33.795
of these services on congestion.

05:34.015 --> 05:37.955
So in terms of methods, what we did was, you know, it,

05:38.055 --> 05:40.395
for each city, we used something different.

05:40.695 --> 05:43.035
Uh, we used, difference in differences approach

05:43.175 --> 05:46.595
by comparing the change in congestion levels

05:47.135 --> 05:51.555
in the affected cities in Mumbai, to a controlled city,

05:51.785 --> 05:55.235
Pune, which is very similar to Mumbai, that did not,

05:55.375 --> 05:56.635
you know, that that was not

05:56.635 --> 05:58.045
subject to the strike, right?

05:58.185 --> 06:01.085
And for Bangalore, we looked at the period of the strike

06:01.145 --> 06:02.925
and compared it with traffic

06:03.145 --> 06:04.965
for similar dates in the following year.

06:05.625 --> 06:08.965
And for Delhi, we used a regression discontinuity design,

06:08.965 --> 06:11.845
like a simple pre, pre-post, comparison,

06:11.895 --> 06:14.925
which compared the changes in congestion during the strike,

06:15.545 --> 06:16.965
to the pre-strike period.

06:17.265 --> 06:21.245
So all, and, and we estimated both distance

06:21.345 --> 06:25.685
and time, using high frequency data for a set

06:25.685 --> 06:29.365
of origin destination pairs, from Google Maps.

06:29.665 --> 06:31.645
So this allowed us to, and,

06:31.645 --> 06:34.925
and that was in essence a measure of congestion, right?

06:34.985 --> 06:37.725
So the travel time that it took between these,

06:37.745 --> 06:40.605
origin destination pairs was what we used.

06:41.025 --> 06:44.205
Um, and we, we chose these endpoints

06:44.425 --> 06:46.805
to get coverage across these cities

06:47.065 --> 06:49.645
and also adequately measure variation in traffic.

06:49.645 --> 06:50.805
That's what we tried to do.

06:51.825 --> 06:53.685
Um, and although they were not random,

06:53.825 --> 06:57.405
we manually chose the, chose these endpoints, they,

06:57.405 --> 06:59.725
they were still, most of them were arbitrarily chosen.

07:00.025 --> 07:01.845
So, you know, that that was something.

07:02.505 --> 07:04.645
And, and yeah,

07:04.645 --> 07:07.925
and we used a variety of empirical models to estimate how,

07:08.345 --> 07:10.525
travel times changed.

07:10.865 --> 07:13.245
Uh, during the strike, after the strike,

07:13.245 --> 07:15.645
before the strike, to estimate,

07:15.645 --> 07:20.365
congestion in all the three cities, we find,

07:20.985 --> 07:23.085
a robust finding, which is

07:23.085 --> 07:25.485
that travel times were consistently lower

07:25.505 --> 07:27.085
during the periods of the strike.

07:28.105 --> 07:32.885
Uh, the, what we also did was we divided these

07:32.945 --> 07:37.725
origin destination pairs into siles of congestion based on,

07:37.725 --> 07:39.685
you know, pre-strike traffic data.

07:40.305 --> 07:43.405
Uh, we normalized that using, free flow traffic.

07:43.785 --> 07:46.205
Um, so that traffic, so that, you know,

07:46.275 --> 07:49.045
this was comparable across origin destination points,

07:49.425 --> 07:51.365
but we found that the reductions were highest

07:51.545 --> 07:54.085
for the most congested regions during peak hours.

07:55.305 --> 07:58.605
And, and likely

07:58.605 --> 08:00.765
because, you know, these, these places were likely

08:00.785 --> 08:02.045
to witness maximum right?

08:02.045 --> 08:04.405
Hailing activity on a usual day, right?

08:04.705 --> 08:07.485
And, essentially emphasizing that, right?

08:07.485 --> 08:09.765
Hailing services were contributing significantly

08:09.945 --> 08:12.605
to congestion in these, cities studied.

08:12.805 --> 08:16.205
I can go into the magnitude of these effects, if you like,

08:16.345 --> 08:19.325
but that was the niche, that was the direction

08:19.325 --> 08:23.005
of the finding a substantive reduction in, in speeds.

08:23.005 --> 08:25.725
So there are studies which have looked at the US

08:26.425 --> 08:31.125
and have also found, a reduction in traffic speeds,

08:31.425 --> 08:32.685
right,

08:33.035 --> 08:37.085
because of right hailing services, roughly

08:37.225 --> 08:39.925
to the magnitude of about two and a half percent or so.

08:40.145 --> 08:43.885
And, you know, they translate these slowdowns to estimation

08:43.965 --> 08:47.445
of congestion costs of about $50 million annually,

08:47.445 --> 08:51.405
somewhere in the range of 35 to $50 million annually, back

08:51.405 --> 08:52.645
of envelope calculations.

08:53.385 --> 08:55.845
Uh, they further find that the,

08:56.095 --> 08:58.685
these costs are similar in magnitude

08:58.705 --> 09:00.805
to the consumer surplus generated

09:01.065 --> 09:03.245
by the right hailing platforms, and

09:03.245 --> 09:06.245
therefore they conclude that, you know, restricting

09:06.245 --> 09:08.885
or taxing this activity is unlikely

09:08.905 --> 09:12.765
to generate large welfare gains through reduced congestion.

09:12.765 --> 09:13.765
Right? So that's

09:13.765 --> 09:16.405
what is the broad conclusion about the Western,

09:16.785 --> 09:19.645
at least I must say more specifically about, you know,

09:19.885 --> 09:22.125
c certain cities in the us This city

09:22.125 --> 09:23.685
that I'm talking about is Austin.

09:24.945 --> 09:26.245
For us, it was interesting

09:26.405 --> 09:29.925
'cause the FX sizes were much larger than those reported in,

09:29.985 --> 09:32.285
you know, in the studies that were based in,

09:32.285 --> 09:33.485
in the United States, right?

09:33.545 --> 09:35.765
We found the effect sizes, congestion costs

09:35.765 --> 09:37.725
of about $2 million per day.

09:38.545 --> 09:43.165
And, which, which leads us to think about, you know,

09:43.185 --> 09:44.565
how is India different, right?

09:44.785 --> 09:46.645
Uh, or how are these economies different?

09:47.225 --> 09:49.085
And there are a few things that came out.

09:49.305 --> 09:52.925
One is the high costs of participation,

09:53.115 --> 09:54.885
because of regulatory barriers

09:55.595 --> 09:59.125
have obviously induced more of these drivers

09:59.305 --> 10:02.405
to participate on a full-time basis on these platforms.

10:02.405 --> 10:03.645
These are not private ride sharing,

10:03.665 --> 10:06.245
but they're registered as commercial taxi operators,

10:06.415 --> 10:08.805
which means that's like a full-time day, you know,

10:08.925 --> 10:10.245
a full daytime job, correct?

10:10.745 --> 10:15.045
Um, and therefore there was a set of costs

10:15.075 --> 10:16.845
that was being, imposed,

10:16.935 --> 10:20.325
which is you're inducing drivers to operate full-time, and

10:20.325 --> 10:22.445
therefore you're increasing deadheading miles, right?

10:22.465 --> 10:25.125
The, the, the time that these cabs are driving around,

10:25.125 --> 10:27.605
vacant on streets, et cetera,

10:27.605 --> 10:30.085
they're not going off when traffic is, you know,

10:30.185 --> 10:31.325
is, is in the trough.

10:31.325 --> 10:34.085
They're, they're staying on the roads and,

10:34.475 --> 10:37.245
therefore, you know, it, in, in some sense,

10:37.715 --> 10:40.045
exploring the role of these regulations

10:40.305 --> 10:43.205
and the associated costs in moderating the supply

10:43.205 --> 10:47.005
of drivers, became important, right?

10:47.185 --> 10:48.765
And therefore, this led us,

10:48.765 --> 10:51.125
so deadheading was clearly one we, you know, we wanted

10:51.125 --> 10:54.005
to implicate, but we also were wanted to understand

10:54.005 --> 10:55.565
what were some of the other mechanisms.

10:56.065 --> 10:58.045
And there were two other mechanisms

10:58.235 --> 11:00.365
that we found in addition to deadheading, we find

11:00.365 --> 11:03.205
that one is deadheading, the second is recommendation

11:03.205 --> 11:04.565
of shorter routes, right?

11:04.785 --> 11:09.285
So when the, the, the, when these, platforms are off,

11:09.785 --> 11:12.845
you know, Google was recommending shorter routes

11:13.105 --> 11:14.525
to people navigating

11:14.525 --> 11:16.725
between the original destination pe, right?

11:17.225 --> 11:19.605
Um, and therefore it was important to account

11:19.605 --> 11:21.965
for these costs in terms of additional distance

11:21.965 --> 11:24.685
that people may be, choosing to travel

11:24.745 --> 11:27.565
to avoid congested streets in the center when the

11:27.765 --> 11:28.845
platforms are operational, right?

11:28.845 --> 11:31.685
That was the second. And the third thing was we looked at

11:31.685 --> 11:33.605
ridership data from the Delhi Metro,

11:34.265 --> 11:37.445
and we found that commuters in Delhi when the platforms were

11:37.465 --> 11:40.485
off were shifting to public transport, in the absence

11:40.485 --> 11:41.725
of the ride hailing services.

11:42.385 --> 11:45.165
So there was some substitution here that was going on

11:45.745 --> 11:48.645
in terms of public transport and ride hailing platforms.

11:48.705 --> 11:50.965
We don't do a welfare analysis in our study,

11:51.025 --> 11:52.965
and I think it's important for policy makers

11:53.045 --> 11:56.885
to perform a more, comprehensive welfare analysis

11:56.905 --> 11:59.485
to compare the costs with what are the benefits that,

11:59.585 --> 12:03.645
you know, these platforms occurred to, to the city.

12:03.865 --> 12:05.645
And there, we, we don't do that,

12:05.665 --> 12:07.765
and that's definitely a topic for future research.

12:08.465 --> 12:11.285
Uh, but what we do find, you know,

12:11.285 --> 12:12.965
but there are certain clear implications

12:12.965 --> 12:15.005
that follow even from the findings, which is

12:15.715 --> 12:18.645
clearly management of supply is critical

12:18.705 --> 12:20.885
to the observed impacts on congestion.

12:21.465 --> 12:24.685
And given that, it, you know, we don't operate

12:24.745 --> 12:28.405
as private ride sharing platforms in, in India, you operate

12:28.405 --> 12:30.725
as commercial taxi operators.

12:31.355 --> 12:33.685
It's important to explore the role

12:33.745 --> 12:37.245
of these ride hailing regulations in India that allow

12:37.265 --> 12:38.645
for dynamic supply, right?

12:38.645 --> 12:39.725
Because if you're operating

12:39.725 --> 12:42.365
as a private ride sharing platform, you get off

12:42.365 --> 12:44.645
during the troughs, you come on the road during peaks

12:44.645 --> 12:46.725
and maybe pooling of demand, all

12:46.725 --> 12:48.885
of this happens far more efficiently,

12:49.105 --> 12:51.245
and that's not happening in this part of the world.

12:51.625 --> 12:55.405
So therefore, it becomes important to explore,

12:55.455 --> 12:57.885
given the benefits of the current set of regulations,

12:57.885 --> 12:58.885
what the costs are.

12:59.505 --> 13:02.125
The second is platform self-regulation, you know,

13:02.125 --> 13:03.845
through search pricing, et cetera,

13:03.865 --> 13:06.285
to ease out supply from congested areas.

13:06.445 --> 13:09.485
I mean, we, we have this whole analysis of traffic

13:09.505 --> 13:10.725
around the airport area

13:10.935 --> 13:13.525
where clearly deadheading can be eliminated

13:13.665 --> 13:15.765
and, you know, you can get these folks off the road

13:15.825 --> 13:16.885
at that point in time.

13:17.585 --> 13:19.485
Uh, the third is improving incentives

13:19.745 --> 13:21.605
to make public transport more effective.

13:21.605 --> 13:23.485
This could be by adding routes, it could be

13:23.505 --> 13:25.285
by reducing fares,

13:26.225 --> 13:30.085
and, you know, improving the coverage

13:30.225 --> 13:32.485
of public transport, not just within cities,

13:32.585 --> 13:33.845
but also across cities.

13:34.505 --> 13:38.565
So I think all of those are policy recommendations

13:38.675 --> 13:42.285
that would follow from this paper, with a caveat that,

13:42.785 --> 13:44.085
you know, one has

13:44.285 --> 13:46.925
to look at a more comprehensive welfare analysis

13:46.925 --> 13:48.285
before one concludes on any of these.

00:00 - This was rooted in the observation that the ride hailing market is a rapidly expanding and growing one. you know, as of 2023, this was at about $160 billion. and the size of the market also spoke to the promise of the market.

00:17 - you know, ride ha platforms. the, you know, the, the intent to leverage a whole host of tech enabled innovations that we outlined in the paper to push towards sustainable mobility, right? These could be interventions like dynamic pricing, efficient matching, pooling of rides, all of that. And a large stream of literature has also documented various socioeconomic benefits of these services, from flexible work to an expansion of the labor pool to include women

00:49 - and other groups that face barriers in traditional, work. and, you know, reduction of driver, moral hazard. There's a whole bunch of socioeconomic benefits that have been documented, and we wanted to examine this phenomenon in India where the growth of the market is significantly higher than that worldwide. for instance, in, for Uber, India is the third largest market by volume globally,

01:16 - and one of the fastest growing markets, right? It has over a million users on its platform. It's clocked to revenue of whatever, nearly, three 50 to $400 million in 2023. And therefore, you know, it was, we, we thought it would be useful to study these impacts in the Indian context, which is under investigated. so both given the salience of the market, its relevance and the fact that it was under investigated set

01:42 - the context for this study. The transportation infrastructure is not terribly mature. and, you know, for instance, incentives to use public transport, present trade-offs in terms of quality of service, et cetera. and also, you know, what the ride hailing platforms have also done over time is adopt to the idiosyncrasies of the market. for instance, they don't operate as private ride sharing platforms.

02:14 - They operate as commercial taxi operators here. And therefore, that brings into play, you know, while it provides benefits such as safety and benefits to, the supply, the labor supply, et cetera. however, there are costs that it imposes, which is what we study. in the paper ride sharing platforms have been implicated overall in worsening congestion. and, you know, and,

02:39 - and that's reflected in several policies such as congestion, pricing plans, et cetera, that have been passed in cities like Seattle and Chicago and, New York City, et cetera. There were two gaps that we sought to address in the literature. One, there is limited evidence, especially causal of the relationship between ride hailing and congestion. This is Driven by essentially paucity of data

03:00 - and a whole bunch of empirical identification challenges, right? So that's one. the second is that, you know, there is, there are these countervailing impacts that the literature talks about. One class of studies says these services reduce congestion because they allow for demand pooling, the, and, and reduce the pressure on transportation infrastructure. They act as a compliment to public transport, and they reduce private ownership of vehicles.

03:26 - however, there's a whole bunch of recent studies, especially ones that try to, you know, pre, present causal, evidence, which show a different viewpoint, and they stock about deadheading substituting, public transport rather than complementing it and inducing trips, which I think is relevant to India as well, that would otherwise never have been made. Remember I said that, you know, the transportation infrastructure is not terribly mature, and

03:51 - therefore, perhaps these trips would not have been made, and therefore there was this theoretical ambiguity regarding the impact of right haing on congestion, which made it an interesting empirical question. And that's what we, you know, that's the gap, that we sought to address. And like I already mentioned, in addition to a reliable measure, of the causal relationship, the Indian context and ascendant economies in general are under investigated

04:16 - or understudied, right? They've all, all empirical studies have almost exclusively focused on the United States. and, and therefore, it, it, you know, we thought it was, it would be useful a, a, a useful contribution to study this in the context of, India. The data was kind of serendipitous because, you know, there was, there was a lot of distress from so Uber, they were offering a certain,

04:40 - compensation to their drivers. They were taking off a certain commission fee, which they amped up, you know, over time. And the distress from reduced earnings, let both Ola and Uber drivers, which together, you know, control about 70% of the ride daily market in India, to go on a strike in New Delhi and Bangalore, somewhere in 2017, in the early part of 2017. And then the last reported strike,

05:05 - there was a ripple effect. And the last reported strike took place in Mumbai in October, 2018. So now for three different time intervals in these three major Indian cities spread over a period of a year and a half, you know, the strike allowed us to, allowed us to exogenously, estimate the impact of, you know, the absence of these services on congestion. So in terms of methods, what we did was, you know, it,

05:38 - for each city, we used something different. we used, difference in differences approach by comparing the change in congestion levels in the affected cities in Mumbai, to a controlled city, Pune, which is very similar to Mumbai, that did not, you know, that that was not subject to the strike, right? And for Bangalore, we looked at the period of the strike and compared it with traffic for similar dates in the following year.

06:05 - And for Delhi, we used a regression discontinuity design, like a simple pre, pre-post, comparison, which compared the changes in congestion during the strike, to the pre-strike period. So all, and, and we estimated both distance and time, using high frequency data for a set of origin destination pairs, from Google Maps. So this allowed us to, and, and that was in essence a measure of congestion, right? So the travel time that it took between these,

06:37 - origin destination pairs was what we used. and we, we chose these endpoints to get coverage across these cities and also adequately measure variation in traffic. That's what we tried to do. and although they were not random, we manually chose the, chose these endpoints, they, they were still, most of them were arbitrarily chosen. So, you know, that that was something. And, and yeah,

07:04 - and we used a variety of empirical models to estimate how, travel times changed. during the strike, after the strike, before the strike, to estimate, congestion in all the three cities, we find, a robust finding, which is that travel times were consistently lower during the periods of the strike. the, what we also did was we divided these origin destination pairs into siles of congestion based on,

07:37 - you know, pre-strike traffic data. we normalized that using, free flow traffic. so that traffic, so that, you know, this was comparable across origin destination points, but we found that the reductions were highest for the most congested regions during peak hours. And, and likely because, you know, these, these places were likely to witness maximum right? Hailing activity on a usual day, right?

08:04 - And, essentially emphasizing that, right? Hailing services were contributing significantly to congestion in these, cities studied. I can go into the magnitude of these effects, if you like, but that was the niche, that was the direction of the finding a substantive reduction in, in speeds. So there are studies which have looked at the US and have also found, a reduction in traffic speeds, right, because of right hailing services, roughly

08:37 - to the magnitude of about two and a half percent or so. And, you know, they translate these slowdowns to estimation of congestion costs of about $50 million annually, somewhere in the range of 35 to $50 million annually, back of envelope calculations. they further find that the, these costs are similar in magnitude to the consumer surplus generated by the right hailing platforms, and therefore they conclude that, you know, restricting

09:06 - or taxing this activity is unlikely to generate large welfare gains through reduced congestion. Right? So that's what is the broad conclusion about the Western, at least I must say more specifically about, you know, c certain cities in the us This city that I'm talking about is Austin. For us, it was interesting 'cause the FX sizes were much larger than those reported in, you know, in the studies that were based in,

09:32 - in the United States, right? We found the effect sizes, congestion costs of about $2 million per day. And, which, which leads us to think about, you know, how is India different, right? or how are these economies different? And there are a few things that came out. One is the high costs of participation, because of regulatory barriers have obviously induced more of these drivers

09:59 - to participate on a full-time basis on these platforms. These are not private ride sharing, but they're registered as commercial taxi operators, which means that's like a full-time day, you know, a full daytime job, correct? and therefore there was a set of costs that was being, imposed, which is you're inducing drivers to operate full-time, and therefore you're increasing deadheading miles, right? The, the, the time that these cabs are driving around,

10:25 - vacant on streets, et cetera, they're not going off when traffic is, you know, is, is in the trough. They're, they're staying on the roads and, therefore, you know, it, in, in some sense, exploring the role of these regulations and the associated costs in moderating the supply of drivers, became important, right? And therefore, this led us, so deadheading was clearly one we, you know, we wanted

10:51 - to implicate, but we also were wanted to understand what were some of the other mechanisms. And there were two other mechanisms that we found in addition to deadheading, we find that one is deadheading, the second is recommendation of shorter routes, right? So when the, the, the, when these, platforms are off, you know, Google was recommending shorter routes to people navigating between the original destination pe, right?

11:17 - and therefore it was important to account for these costs in terms of additional distance that people may be, choosing to travel to avoid congested streets in the center when the platforms are operational, right? That was the second. And the third thing was we looked at ridership data from the Delhi Metro, and we found that commuters in Delhi when the platforms were off were shifting to public transport, in the absence of the ride hailing services.

11:42 - So there was some substitution here that was going on in terms of public transport and ride hailing platforms. We don't do a welfare analysis in our study, and I think it's important for policy makers to perform a more, comprehensive welfare analysis to compare the costs with what are the benefits that, you know, these platforms occurred to, to the city. And there, we, we don't do that, and that's definitely a topic for future research. but what we do find, you know,

12:11 - but there are certain clear implications that follow even from the findings, which is clearly management of supply is critical to the observed impacts on congestion. And given that, it, you know, we don't operate as private ride sharing platforms in, in India, you operate as commercial taxi operators. It's important to explore the role of these ride hailing regulations in India that allow for dynamic supply, right?

12:38 - Because if you're operating as a private ride sharing platform, you get off during the troughs, you come on the road during peaks and maybe pooling of demand, all of this happens far more efficiently, and that's not happening in this part of the world. So therefore, it becomes important to explore, given the benefits of the current set of regulations, what the costs are. The second is platform self-regulation, you know,

13:02 - through search pricing, et cetera, to ease out supply from congested areas. I mean, we, we have this whole analysis of traffic around the airport area where clearly deadheading can be eliminated and, you know, you can get these folks off the road at that point in time. the third is improving incentives to make public transport more effective. This could be by adding routes, it could be

13:23 - by reducing fares, and, you know, improving the coverage of public transport, not just within cities, but also across cities. So I think all of those are policy recommendations that would follow from this paper, with a caveat that, you know, one has to look at a more comprehensive welfare analysis before one concludes on any of these.

Faculti AI can make mistakes