Who did button my waistcoat, a human or a sufficiently intelligent algorithm, (SIAL)? At Manchester and Smurfit business schools we explore the economics of the sharing economy from a game theory perspective. We have shared observations and experiences of online transactions. They represent data patterns that are embedded in the clicking and scrolling behaviour of shoppers in cyberspace. Additionally, we would also like to understand, why it is that online rational consumers – we call them, the ‘onsumers’ – appear to bid against themselves during online transactions. In other words, a final price, the END price, is more frequently higher than the opening bid, the buy-it-now or BIN price for an online transaction.
Is it a behavioural conundrum? Is it rational? Our research continues as we test the hypothesis:
BIN price < END price
Sufficiently Intelligent Algorithms
Online purchases are guided by search algorithms. Price is an important signal. As the ‘onsumer’ arrives at the landing page of the online site he or she clicks and scrolls up and down the screen up to a tipping point. In that quantum moment in time the clicking behaviour of the ‘onsumer’ has created a pattern of online behaviour, a pattern that has been captured by the algorithm. The pattern is a valuable tradeable asset of information.
The scrolling and clicking behaviour diverts from the original purchase and has allowed the algorithm to add in menu and scroll costs that are not discounted by the online shopper. In reality you are likely to be bidding against yourself as you search beyond the BIN price on offer, as you continue to click and compare, discuss with others, price compare the BIN offer in related sync’d online sites. As you deliberate and prevaricate, the sufficiently intelligent algorithm captures all this behavioural information in a moment in time, ∆.
Your before and after pattern of behaviour can be represented as a first order difference equation, the early building blocks of an intelligent algorithm:
Bt+1 = Bt + ∆
Rational ‘onsumers’ may choose to discount transaction costs to zero but the sufficiently intelligent algorithm does not. There is a revenue purse to be shared amongst the online digital advertisers. Data patterns are a tradeable asset. They are embedded at ∆ into your online price, influencing your online behaviour in a subliminal bidding war against yourself: a final price, the END price, is higher than the opening bid, the buy-it-now or BIN price.
If we can identify the sources of the online transaction costs (TC) that are incurred by a rational consumer, then we might find an explanation for the divergence between BIN and END prices. If a rational ‘onsumer’ knows that TC > 0 then he or she will make every effort to minimise the transaction costs. If so, why is it that BIN < END? Is it unique to specific online transactions such as hotel or holiday bookings?
Non-Observable (Online) TC
Whatever the heuristics involved we believe that there are additional non-observable online TC complementing the data recovery by the SIAL. There are at least two nonstandard transaction costs at ∆ in your online purchases that require some attention:
- a rational ‘onsumer’ booking a holiday may be part of a group of friends but the algorithm is sufficiently intelligent to pick up an individual’s click as correlated with a group set of clicks leading to a purchase – call this a herding effect; and
- rational ‘onsumers’ are inclined to repeat purchases online from the same airline or hotel or from the same online platform with stored auto-fill information on address and credit card details and again the algorithm is sufficiently intelligent to know this – call this a hard commitment or no-switching
You Are Your Own Opponent
Allowing SIALs to influence your behaviour deprives you of a sovereign right as a rational consumer to choose freely. In bidding against yourself you have become your own irrepressible opponent[1]. Do we as rational ‘onsumers’ discount these non-observable costs at zero? Yes, we do because buying online generates latent or hidden transaction costs (TC). It is a different experience to the transaction of buying an apple at the fruit market. Consequently, an online purchase is always perceived to be cheaper than in the (offline) store in the shopping mall. This may well be the case for many products and services purchased online. But do you necessarily end up buying the product or service at a lower online price? The jury is still out. The pricing is less transparent than at your local bookstore (if it is still in existence) where buying a book presents a degree of transparency on the price paid. It too has opportunity costs but they are observable – for example, car-park ticket or bus fare to the book shop or time spent browsing in the store.
Prognosis
Anecdotally, paying others to wait in the queue line[2] is evidence that the transaction costs are not discounted to zero. In fact, they are positive; and if they are observably positive then a rational consumer would be prepared to pay up to the discounted value of their opportunity cost in order to be in the queue to secure an iPhone or a bargain priced 4K Sony TV. With online transactions, there may be a herding effect and a non-switching effect identified in the data patterns. The SIAL can translate them at ∆ into non-observable waiting time costs and queue costs. Unknowingly, the rational ‘onsumer’ discounts them to zero and BIN < END price. Machines ‘think’ because humans have stopped thinking. Memory as well as patterns of behaviour have been outsourced to smart devices. It is not inconceivable that an unwitting thief in the digital age of the Turing thinking machine may have buttoned my waistcoat. Honestly, I can’t remember!
[1] Check out an early idea at http://www.patrickmcnutt.com/blog/game-vision-you-are-your-own-irrepressible-opponent/
[2] https://www.theatlantic.com/business/archive/2014/07/the-growing-market-for-getting-paid-to-wait-in-line/375083/