Monday, January 15, 2018

Uber and traffic congestion - two sides of the same coin?

I have blogged earlier pointing to the negative externalities of transport aggregators by way of increasing traffic congestion, drivers without social protections, and crowding in large number of drivers into a business whose commercial viability (especially in developing country markets) is still not established. 

But I am surprised that the case about traffic congestion is still contested.

Consider this. Cities suffer from traffic congestion. Public policy response to address them include congestion pricing, license plate auctions, high vehicle taxes, prohibitive parking fees and so on. All of these seek to make vehicle ownership and use expensive. In other words, reduce the number of vehicles on the road. There is nothing profound here - take vehicles off the road to reduce congestion! 

Now what does Uber do. It brings those idle vehicles into the road. In other words, more vehicles into the road. Exactly the opposite of what public policy tries to do to reduce congestion. 

So here is an innovation, which in the guise of enhancing efficiency (better use of investments already made), artificially increases vehicles on to roads, which are already choking with traffic, with limited prospect of any commensurate or reasonable accompanying infrastructure expansion. 

Note that the efficiency improvement argument too is biased - it takes into account the private benefit for the vehicle owner while overlooking the social costs from all the negative externalities. In fact, I will argue that far from promoting innovations that allow more optimal use of purchased vehicles, public policy should have made the vehicle ownership itself prohibitive. Having missed the first window, the least that one can do is to ensure that vehicle use is curtailed. Aggregators do exactly the opposite!

In simple terms, the likes of Uber increase our private convenience (we can easily summon a car at any time, not own a vehicle and so on) but at the far higher social cost of greater traffic congestion. 

The real efficiency improvement would be to encourage all existing taxis (individual and operators) to move into aggregator platforms (and individual taxis can even offer their own specific rates) and there is competition among the various platforms.


Karthik Dinne said...

Some studies find "decrease in congestion" due to Uber. They explore three possible mechanisms driving the result.

1. Uber's surge price is essentially a "congestion tax". Surge prices disincentivize people to travel during peak traffic times, thereby reducing congestion.
2. Higher utilization due to sharing (1.8 per car in Uber v. 1.1 for private car)
3. Due to better technology in Uber, it's easier for Uber drivers to find customers, as opposed to private taxis. It reduces wandering of taxi drivers in search of passengers, which in turn reduces congestion.

There are other studies too, finding "increase in congestion". But I find the 1st mechanism of "Uber's surge price acting as congestion tax" interesting. Maybe it has more impact than what we think of.

Gulzar Natarajan said...

Karthik. Thanks.

1. The point I am making is that just getting Uber into the roads (more vehicles on road) is itself contributing to congestion. Then managing the larger pool of vehicles with price signals is only salvaging the situation which is already more congested than would have been the case without Uber (with only private cars and taxis).

2. On your Sl No 3, it is precisely such efficiency gains that can be reaped by having existing taxis move into aggregator platforms through some business model.

The central point is that Uber/Ola etc are innovations that bring in more vehicles into already congested roads. We need to control them. Alternatively, every private car (personal cars, Uber, taxi etc) ride inflicts a social cost. Further, given the traffic levels in our cities, the incremental cost is even more.

Karthik Dinne said...


The paper compares "congestion before Uber" v. "congestion after Uber put cars on road". It finds that congestion decreases after Uber puts cars on roads, as compared to the case without Uber.

From the paper -

"....after entering an urban area, ride-sharing services such as Uber significantly decrease traffic congestion time, congestion costs, and excessive fuel consumption. "

"...difference-in-differences framework to examine the difference in congestion before and after Uber entry across multiple areas"

Intuitively, it seems that Uber puts more cars on road and hence it should lead to congestion. But what these results suggest is that the other effects are significant enough to counter the effect of "putting more uber cars on road. The countering effects being the removal of personal cars from the road, disincentivizing people to use roads during peak time, better utilization of car seating capacity, and so on.

Gulzar Natarajan said...

Karthik, thanks.

You will find exactly similar points to quote in exactly the opposite direction in the paper linked here (as well as the graphic)

Btw, this is what I am increasingly finding out in such complex problems - for every evidence data point in one direction, there is another in the opposite direction!

But that is not the point I am making.

Consider this. We already have too many vehicles in our roads - maybe 2-3 times the number in peak times in the larger cities (and growing very fast). The primary objective of public policy (or any market innovation) should be to get a significant number of vehicles off the road by inducing people to use public transport (assuming complementary investments in public transport) or shift their travels to off-peak times. Btw in most cities, congestion has become so excruciating that increasingly off-peak times are only when we are all asleep!

For Uber to be doing this, it would have to induce people who were using private cars to either shift their travel times to off-peak hours or use pooled Uber. And this would have to be done in very large numbers to at the least off-set the natural rate of increase in private cars entering the roads each year. We also know about the lack of success with dynamic toll-pricing and HOV lines in changing travel behaviours (by shifting travel times and using pooled vehicles) and thereby lowering congestion in several cities.

Besides, it would also have to ensure two things - one, as I mentioned in the linked post (and the paper referenced), that the middle-class who have recently gravitated from cars and motor-cycles to the new Metro rail will have to stay on with the rail; and two, those who (not rich enough to own a car, but not poor too) otherwise would not have used cars or would have used the auto (say) or a bike would now not shift to the greater personal comfort of a Uber. We can safely say that both these will happen in large numbers.

I could actually list out several more factors (that drive private vehicle use and reluctance to ) that Uber will have to offset. Not to mention evidence from places like London where private cars have actually decreased since 2012, while Uber vehicles have more than made up the decline, leaving London ever more congested in every passing year since (also due to increased delivery vehicles).

The counterfactual (in my judgement) is orders of magnitude off, for me to be even taking the paper you mentioned seriously...

The point is this - we need public policy and market innovation that gets vehicles off the road. I don't believe you can do that by getting more vehicles on to the road.