Market-making in on-demand markets

I’ve written a post on LinkedIn about the need for market-making in on-demand markets. I argue that for a market to be on-demand for one side, you require the other side to be able to provide liquidity. This liquidity comes at a cost and the side needs to get compensated for it. Driver incentive schemes at Ola/Uber and two-part electricity tariffs are examples of such incentives.

An excerpt:

In a platform business (or “two sided market”, or a market where the owner of the marketplace is not a participant), however, the owner of the market cannot provide liquidity himself since he is not a participant. Thus, in order to maintain it “on demand”, he should be able to incentivise a set of participants who are willing to provide liquidity in the market. And in return for such liquidity provided, these providers need to be paid a fee in exchange for the liquidity thus provided.

Read the whole thing! 🙂

Airline delays in India

So DNA put out a news report proclaiming “Air India, IndiGo flyers worst hit by flight delays in January: DGCA“. The way the headline has been written, it appears as if Air India and Indigo are equally bad in terms of delayed flights. And an innumerate reader or journalist would actually believe that number, since the article states that 96,000 people were inconvenienced by Air India’s delays, and 75,000 odd by Indigo’s delays – both are of the same order of magnitude.

However, by comparing raw numbers thus, an important point that this news report misses out is that Indigo flies twice as many passengers as Air India. For the same period as the above data (January 2015), DGCA data (it’s all in this one big clunky PDF) shows that while about 11.65 lakh passengers flew Air India, about 22.76 lakh passengers flew Indigo – almost twice the number. So on a percentage basis, Indigo is only half as bad as Air India.

airlinedelays

The graph above shows the number of passengers delayed as a proportion of the number of passengers flown, and this indicates that Indigo is in clear second place as an offender (joined by tiny AirAsia). Yet, to bracket it with Air India (by not taking proportions) indicates sheer innumeracy on the part of the journalist (unnamed in the article)!

I’m not surprised by the numbers, though. The thing with Indigo (and AirAsia) is that the business model depends upon quick turnaround of planes, and thus there is little slack between flights. In winters, morning flights (especially from North India) get delayed because of fog and the lack of slack means the delays cascade leading to massive delays. Hence there is good reason to not fly Indigo in winter (and for Indigo to build slack into its winter schedules). Interestingly, the passenger load factor (number of passengers carried as a function of capacity) for Indigo is 85%, which is interestingly lower than Jet Airways (a so-called “full service carrier”)’ s 87%. And newly launched full service Vistara operated at only 45% in January!

We are in for interesting times in the Indian aviation industry.

Pricing and waiting in line

Every time I have to stand in what seems like an exceedingly long line for something, I wonder if they’ve got the pricing wrong. If they had their economics straight, I reason, they would raise prices to an extent where the market just “clears”, and there is no need for a line.

In this context, this piece by Tyler Cowen comes in handy, where he talks about the various advantages of lines and waiting in line. Apart from some superfluous stuff such as lines making us more patient and waiting in line being less painful now thanks to smartphones, Cowen makes some very interesting points. For example,

Higher prices also skew the customer mix toward wealthier and thus older people, who exert less influence over the purchasing decisions of their peers. They are less likely to text about a concert, put it on their Facebook pages or talk up its reputation to dozens of friends at parties. The younger buyers are usually the ones who make places trendy, thus many sellers use lower prices, with lines if need be, to lure in those individuals and cultivate their loyalties.

The above passage illustrates why it is sometimes necessary to keep prices lower than the “clearing price” and let lines form (as long as you have an orderly way of dealing with the lines). Essentially, by raising price until a point where the market just clears, you are optimising the revenues for that particular day or point in time.

However, if you are a “going concern”, as all businesses are normally assumed to be, you don’t optimise for revenues or profits on a particular day. What you optimise for is long-term sustainable profit, and you do what it takes in order to maximise that.

As Cowen says above, by keeping prices lower than clearing price, you draw crowds that are likely to talk about the experience of your offering, thus giving you free advertising. As the “flash sales” conducted by Xiaomi (where phones sold out in a few seconds after sales opened) show, lines can end up being reported in the press which creates free publicity for you – leading to greater future sales.

Then there is (Cowen touches upon this) the signalling effect of the line itself – that so many people are waiting in line for something signals that there’s something inherently worthy about the product, and results in increasing demand (and more people in the line!). The line is an act of discovery – you may not go to a food card if you didn’t see the line in front of it.

There is also the issue of price elasticity – beyond certain levels, prices can be extremely elastic, in that if you raise prices at a particular margin, demand drops significantly (this has to do with “price barriers” in people’s minds – possibly a behavioural issue). So it becomes impossible for you to set the price at the precise level where your establishment just fills up. So you have a choice between not filling up your capacity or getting people to stand in line. And the latter is more profitable.

The lesson from this is that you should think long term when you are analysing pricing decisions, and not optimise for maximising instantaneous profits! Read the full piece by Cowen. It’s well worth it.

The Peer Pressure of Finishing An Exam Early

Today is the final exam of my course at IIMB. It’s a two part exam – students have been given the problems today and they have to describe on paper how they are going to approach the problem. Tomorrow I’ll send them relevant data and then they need to build an Excel model and solve the problem.

The point of this blog post, however, is to do with the peer pressure of finishing an exam early. Today’s exam is taking place in two rooms, with the students having been divided equally between the rooms. I’m writing this two and a half hours into a four hour exam, and so far about a dozen students have handed in their papers. The interesting thing is that eleven of these are from one room, and one from the other.

This makes me wonder if there is some kind of “peer pressure” in terms of finishing an exam. When you hand in your paper early, you signal one of two things – either that you have really aced the exam or that you really have no clue. By looking at the people who have walked out so far and their academic reputations, it is possible for the remaining students to know whether the people who have left have aced the exam or given up.

So the question is if there is some kind of gamesmanship involved in finishing an exam early. Let’s say a stud walks out of a 4-hour exam in an hour. Does he walk out early in part to let his peers know that it was a bloody easy exam and that they should be doing better than they already are? And does this in part put pressure on the other studs to “preserve their reputations” in some manner by also finishing early? And does this imply that they might hurry up and not do a good enough job of the exam, leading to suboptimal performance and better grades (let’s assume a relative grading system) for the person who originally walked out?

Or do you think walkouts are independent? That two students walking out i close succession to each other were independent events that I’m reading into too much? I wish I had actually tabulated the timings at which papers had been handed in, and maybe perhaps correlated them with the actual performance in an exam (to analyse how early finishing affects performance). As it stands, though,I should work on the data available.

I’m writing this blog post siting in room 1 (posting later since Wi-Fi has been switched off here for purpose of the exam). After I started writing, one of the studs sitting in room 1 walked out. Almost in quick succession one other stud in this room followed him. This is the room where one guy had walked out really early, and he’s also one of the studs of the class.

This suggests that there is some kind of correlation. A sort of relationship. That one person walking out puts pressure on others to also walk out. And can result in some good “relative grading”!

I’ll end with an anecdote from my days as a student here, almost exactly 9 years back. It was an objective final exam, with multiple choice questions only. And in that series of exams it had been some sort of a competition as to who would walk out early.

So it was the last exam, and this one guy decided to “show off” by walking out within five minutes. Unfortunately one other guy had decided to turn up late for the exam. The institute rules state that nobody is allowed into an exam after at least one student has walked out. So the second guy was not allowed to take the exam.

As it turned out, he got a better grade than the guy who had walked out within five minutes!

What sets Uber apart from other marketplaces

While at the gym this evening I was thinking of marketplaces.  To give some context, the reason I went there was that there were too many thoughts running around my head, so I needed to focus on something mindless or something that required so much concentration that I could only hold one other thought in my mind at that point in time. In fact, when you go “under the bar” (do a back squat),  even that one thought will vanish – you need all your physical and mental energy to complete the squat.

Anyway so I was thinking of marketplaces, and marvelling at the kind of impact companies like Uber and Ola have had. They have been an absolute gamechanger in their business in that it has completely changed the way that people and cabs get matched to each other. This was a matching that had been extremely inefficient in the past, but with these apps, they have become better by an order of magnitude. And it is this order of magnitude that sets apart Uber/Ola from other marketplace businesses.

And as I was moving between weights, I had another thought – the trick with Uber/Ola as a marketplace is that it is near impossible to do “side deals”. The ultimate nightmare for a platform/marketplace provider is to let the two sides “discover each other” and conduct further deals “offline”. This can be the bane of services such as dating services, where once you go on your first date (as recommended by OkCupid or Tinder), the dating service need not know anything about your relationship after that! They’ve “lost” you. In fact, talking to someone from the industry recently, I learnt that they do dating rather than marriage since in the former there is the hope of “repeat happy customers”.

It is similar with a service such as Airbnb, where once you’ve located a B&B you like, you can cut airbnb out of the deal from the next time on. Of course availability and stuff matter, but given how much in advance you book, a quick call to check availability is a small cost vis-a-vis the benefit of cutting out the middle man.

The beauty of Uber/Ola, however, is that it is impossible to do deals offline. Yes, after a ride, the driver and the passenger have each other’s numbers. But the next time the passenger wants a ride, the probability that the same driver is in the vicinity and free to give a ride is infinitesimal. So the passenger has to go to the app again. Moreover, it is the app that takes care of the pricing (using GPS, etc.) – something that is impossible to estimate if you try to cut out the app.

So when people say that they are building the “Uber for <some other service>”, in most cases it is not exactly the case – not all marketplace transactions are like Uber. For to be like Uber, you need to be an instant matching mechanism that changes the way the industry fundamentally operates; and you need a mechanism that keeps deals “online” by force.

Chew on it!

Recommendations and rating systems

This is something that came out of my IIMB class this morning. We were discussing building recommendation systems, using the whisky database (check related blog posts here and here). One of the techniques of recommendation we were discussing was the “market basket analysis“, where you recommend products to people based on combinations of products that other people have been buying.

This is when one of the students popped up with the observation that market basket analysis done without “ratings” can be self-fulfilling! It was an extremely profound observation, so I made a mental note to blog about this. And I’ve told you earlier that this IIMB class that I’m teaching is good!

So the concept is that if a lot of people have been buying A and B together, then you start recommending B to buyers of A. Let us say that there are a number of people who are buying A and C, but not B, but based on our analysis that people buy A and B together, we recommend B to them. Let’s assume that they’ve taken our recommendation and bought B, which means that these people are now seen to have bought both B and C together.

Now, in case we don’t collect their feedback on B, we have no clue that they didn’t like B (let’s assume that for whatever reason buyers of C don’t like B), but in the next iteration, we see that buyers of C have been buying B, and so we start recommending B to other C buyers. And so a bad idea (recommending B to buyers of C, thanks to A) can spiral and put the confidence of our recommendation system in tatters.

Hence, it is useful to collect feedback (in the form of ratings) to items that we recommend to customers, so that these “recommended purchases” don’t end up distorting our larger data set!

Of course what I’m saying here is not definitive, and needs more work, but it is an interesting idea nevertheless and worth being communicated. There can be some easy workarounds – like not taking into account recommended products while doing the market basket analysis, or trying to find negative lists and so on.

Nevertheless, I thought this is an interesting concept and hence worth sharing.

Rating systems need to be designed carefully

Different people use the same rating scale in different ways. Hence, nuance is required while aggregating ratings taking decisions based on them

During the recent Times Lit Fest in Bangalore, I was talking to some acquaintances regarding the recent Uber rape case (where a car driver hired though the Uber app in Delhi allegedly raped a woman). We were talking about what Uber can potentially do to prevent bad behaviour from drivers (which results in loss of reputation, and consequently business, for Uber), when one of them mentioned that the driver accused of rape had an earlier complaint against him within the Uber system, but because the complainant in that case had given him “three stars”, Uber had not pulled him up.

Now, Uber has a system of rating both drivers and passengers after each ride – you are prompted to give the rating as soon as the ride is done, and you are unable to proceed to your next booking unless you’ve rated the previous ride. What this ensures is that there is no selection bias in rating – typically you leave a rating only when the product/service has been exceptionally good or bad, leading to skewed ratings. Uber’s prompts imply that there is no opportunity for such bias and ratings are usually fair.

Except for one problem – different people have different norms for rating. For example, i believe that there is nothing “exceptional” that an Uber driver can do for me, and hence my default rating for all “satisfactory” rides is a 5, with lower scores being used progressively for different levels of infractions. For another user, for example, the default might be 1, with 2 to 5 being used for various levels of good service. Yet another user might use only half the provided scale, with 3 being “pathetic”, for example. I once worked for a firm where annual employee ratings came out on a similar five-point scale. Over the years so much “rating inflation” had happened that back when I worked there anything marginally lower than 4 on 5 was enough to get you sacked.

What this means is that arithmetically averaging ratings across raters, and devising policies based on particular levels of ratings is clearly wrong. For example, when in the earlier case (as mentioned by my acquaintance) a user rated the offending driver a 3, Uber should not have looked at the rating in isolation, but in relation to other ratings given by that particular user (assuming she had used the service before).

It is a similar case with any other rating system – a rating looked at in isolation tells you nothing. What you need to do is to look at it in relation to other ratings by the user. It is also not enough to look at a rating in relation to just the “average” rating given by a user – variance also matters. Consider, for example, two users. Ramu uses 3 for average service, 4 for exceptional and 2 for pathetic. Shamu also uses 3 for average, but he instead uses the “full scale”, using 5 for exceptional service and 1 for pathetic. Now, if a particular product/service is rated 1 by both Ramu and Shamu, it means different things – in Shamu’s case it is “simply pathetic”, for that is both the lowest score he has given in the past and the lowest he can give. In Ramu’s case, on the other hand, a rating of 1 can only be described as “exceptionally pathetic”, for his variance is low and hence he almost never rates someone below 2!

Thus, while a rating system is a necessity in ensuring good service in a two-sided market, it needs to be designed and implemented in a careful manner. Lack of nuance in designing a rating system can result in undermining the system and rendering it effectively useless!

On using slides for a lecture

Last evening I attended Pratap Bhanu Mehta’s New India Foundation lecture on the role of religion in politics in Modern India. It was a rather complex topic, with a lot of philosophical underpinnings, and as it happened, I soon lost track, and consequently interest, and so I ended up writing this blogpost as I sat in the audience.

Later, I was wondering what PBM could have done to communicate better and make his lecture more easily understood. Let’s assume here that PBM is an academic and so the last thing he would want to do is to “dumb down” the lecture (also my hypothesis is that he perhaps upped the academic quotient since the venue for the lecture was IISc but that’s just speculation and we’ll keep it aside). There is a certain way he is comfortable speaking in, weaving academic arguments, and let us assume that he is best talking that way, and it is not ideal to change that.

Imposing the above and any other reasonable constraints on what PBM would not change about the lecture, the question remains as to how he could have made it easier for the audience to follow him. And thinking about it in hindsight, the answer is rather obvious – visual aids such as slides (or even a blackboard).

The problem with the lecture was that given its complexity there were several threads of thought that the audience member had to keep track of as PBM built his argument. The flipside of this is that if you happened to miss a line of what PBM said, one important thread in the web would get lost, after which it would be extremely difficult to follow the rest of the lecture (This is perhaps what happened to me because of which I started blogging). To put it another way, the lecture as it happened required a high degree of concentration as well as maintenance of a reasonably sized cache in the minds of each audience member, which meant that the mental energy required to follow was really high.

In an unrelated conversation after the lecture, someone was talking about how the ancient Greeks reacted to the invention of writing with horror, saying that the human mind was perfectly capable of storing and transmitting information, and that writing would lead to a diminishing of human mental powers. As it turned out, writing helped free up memory space in human minds and that allowed for more complex thinking and a lot of subsequent scientific development. Of course cultures such as India’s which continued to insist on learning the scriptures by rote lost out a bit because considerable mental capacity continued to be used as a means of storage rather than for processing power.

So the idea is that when you have a complex talk that involves a complex web of thought, considerable mental energies of the audience goes into just maintaining a cache of all that you’ve said, and the arguments that you’ve constructed. And on top of that they need to continue to listen to you with concentration as you continue to weave the web. The rate of dropoff can be rather high. So the least you can do to help the audience ingest your lecture better is to help free up their cache, and putting out all the arguments spoken thus far on a screen, which means that their mental power can go into ingesting and digesting your new information rather than simply maintaining the cache. And that will improve your throughput!

So to generalise, use of visual aids (slides are preferable to blackboards since you don’t waste time writing, but if there isn’t much to be written blackboards will do, too, since slides might constrain) is a necessary condition to ensure high throughput when your talk involves a rather complex web of argument. It simply makes it easier for the audience to follow you and you can communicate better!

Of course if you are of the persuasion that there is a certain way you communicate which you’re unwilling to change and it is the audience which needs to make an effort to catch your pearls of wisdom, none of the above applies to you.

Sociology and economics

A few years back I was interviewing a sociology graduate for a scholarship and loudly exclaimed that it was absurd that she had a masters in sociology while not knowing much economics – she had mentioned that her courses in sociology (bachelors and masters) had no “papers” (the word used by students of certain prominent Indian universities when they mean “courses”. The choice of words possibly indicates their priorities) in economics.

It is a result of my prior – everything I know and have learnt about sociology and social behaviour is from the realm of economics and game theory (iterated prisoners’ dilemma and derivatives). I’ve learnt it from reading blogs (Marginal Revolution, Econlog, etc.) and from pop economics books written by authors such as Steven Levitt and Malcolm Gladwell. So every time I think of a sociology problem I can’t think of any method apart from economic reasoning to attack it.

However, it turns out that the use of economic reasoning for sociological analysis is rather recent, and started only with the work of Chicago economist Gary Becker, who wrote a series on love and marriage. Becker’s wife had died, and he was a single father, when he wrote his series of papers on this topic. This is supposed to be one of the first steps in the “creep” of economics into (now) related disciplines such as sociology and political science. This has been uncharitably called by Becker’s critics as “economic imperialism“.

So my exclamation that a masters program in sociology not including a course in economic reasoning being absurd would be valid only in very recent times, when syllabuses would have been updated to keep track of any such above “imperialism” and “creep”. Given the glacial pace at which Indian universities move, however, I think my remark might have actually come across as absurd!

PS: Read this excellent Lunch with FT interview of Gary Becker by Tim Harford.

ADHD and appreciating art

So a week back I finished reading my third fiction book in four months – “the Rosie project”, a book about a professor of genetics who has Asperger’s syndrome and his effort to find a wife. I got this recommendation via Twitter and procured the kindle sample, and having really liked it went on to read and like the book.

This is not a book review. Essentially in this post I try to analyse why I don’t really read too much fiction. About why in the last ten years I read not more than two or three books of fiction before finally starting on and finishing Neal Stephenson’s cryptonomicon. And then read the same authors 3000 page eight part baroque cycle.

So I’m not a great fan of movies. There are many movies which look interesting thanks to which I DVR them and start watching them but am just unable to sustain interest in them thanks which I end up not watching them. And these movies end up unwatched.

On the other hand there at movies that generate such deep interest that I can’t take my eyes off them and I finish seeing them in one sitting. Thinking  about them these movies have really taut plots, without any fluff, and this allows me to sustain my interest and watch them.

Three years back I was diagnosed with attention deficit hyperactivity disorder (ADHD). This so-called disorder means that it’s really hard for me to hold my interest on anything that I’m doing. That it means that I’m perennially distracted. That I’m not able to be in the present and am always daydreaming. Because of which I under perform and am occasionally not able to function etc.

Now thinking about it, thanks to my ADHD I’m a great judge of movies and books and lectures and any other media that need to hold your attention to succeed. Because I’m forever distracted it’s very difficult to hold my interest in anything. So my interest can be held iff the “plot” (this applies to movies books articles lectures and all such) is tight and without too much extra fittings. When the plot is not taut there is a higher chance that I get distracted, and since I have ADHD I stop making an effort to concentrate and focus and capture essence and it’s all lost.

Its interesting to note that movies that I like instinctively are those that are generally highly rated. The converse is also true – movies that fail  to hold my attention by not having a taut enough plot are those that are generally not highly rated. Of course you could argue that I’m a sucker for public approval but the correlation is remarkable.

So my ADHD means that I’m unable to enjoy a movie or a book or a lecture or an article unless it’s really well written/ Spoken/performed. In that sense my lack of tolerance for something that’s not up to par – by having redundancies and inanities and thus having too many “extra fittings” – means that I’m unable to consume any content that is even marginally under par. Or that I have very high standards for grabbing my attention towards anything which means that I consume little but whatever i consume is of high quality!

So the reason I gave up on fiction itself is a function of reading a lot of bad fiction. Stuff that was badly written but what I forced myself to read because of the “I’ve started so I’ve finished” principle. And the trouble it’s caused my has meant that I’ve decided not to read fiction at all!

In terms of non fiction I’ve been much more discerning in the first place in terms of stuff I’ve started reading. And the generous peppering of “fundaes” in most non fiction books means that my interest has been sustained and I’ve managed to read a fair bit!

I’ve written this blog post sitting at a lecture written by a rather popular academic. It’s a promising lecture but the first few minutes were not crisp or competing enough – which means that my interest hasn’t been sustained and so I’ve switched off!

The lecturer’s reputation precedes him so my opinion may not match popular opinion about the lecture ( expressed publicly) this time. But I believe that my ADHD has made be a great judge of whether something has been communicated well!