People are terribly illogical: irrational behavior in public procurement

Shopping at a regular store is simple and straightforward. Conventionally, you simply remove the kettle from the shelf and go to the cashier to pay. There are less obvious options when you buy, for example, a refrigerator, a dishwasher and an automatic feeder for your beloved parrot. In such a situation, I recommend catching a senior manager on the trading floor and asking for an individual discount, motivating competitors with good offers. Even in large retail chains, this way you can slightly reduce the price using such a mini-version of the tender.



The most impressive clashes of competitors occur on large trading floors, when the quantity of goods can be measured not in pieces, but in whole freight cars.



We, as the largest electronic trading operator, have accumulated a lot of interesting statistics on this topic, which I want to share. Today I’ll talk about how strange the participants in various auctions begin to behave when the number of applicants for a particular lot increases significantly.



Spoiler: do you think from this the price drops to a minimum? Do not wait!

Why simplified theories do not work well

At first glance, it might seem that the competitive behavior of a market participant must obey quite strict and rational rules. From these premises classical theories of games and competition proceed. They usually have a rather simplified model at their base: all participants act strictly within their budget, and the choice is limited to two products. It is usually assumed that the number of competitors is known in advance, and each participant makes his proposal in turn, waiting for proposals from other participants. Many may have the impression that in the end everyone should behave like rational algorithms for automated trading.



You can forget about it. It does not work in reality. There are a lot of factors. For example, the budget can float significantly due to affordable lending, and not just two products are bought, but much more. At the same time, participants have a wide variety of alternative and related products. With information about the position, quality and volume of goods, competitors are much more interested. Procurement law prohibits the disclosure of information on proposals of participants to both the customer and, until a certain point, to other participants. Of course, participants can communicate with each other, entering into cartel conspiracies, which the Federal Antimonopoly Service is actively fighting against, and offering kickbacks to the customer, which law enforcement agencies are fighting with, but ... But!



We are used to perceiving companies as a kind of soulless mechanistic entity and, as Vladimir Ilyich would say, a capitalist bastard who is extremely rational in extracting the maximum possible profit. In fact, companies are run by ordinary people who are just as characteristic of a complete set of cognitive biases. As a result, the market resembles a chaotic system, which is often driven by emotional and psychological factors rather than strict logic.

How much is wine

People and even successful calm-headed entrepreneurs with iron nerves tend to give transactions subjective value, which often leads to irrational decisions. Richard Thaler, a Nobel laureate in economics, describes the following example in his book Behavioral Economics: Together with psychologist Eldar Shafir of Princeton, they conducted a survey among participants in the annual newsletter of the wine auction. The question was asked in the survey: “Let's say you bought a box of Bordeaux at $ 20 per bottle. Now this wine is sold at auction for about $ 75 per bottle. You have decided to drink one bottle of wine. Which of the statements below best describes your attitude to the cost of the bottle you’ve drunk? ”



The results were quite interesting:

• 0 dollars. I already paid for this bottle (30%).
• 20 dollars. So much I paid for it (18%).
• $ 20 plus interest (7%).
• 75 dollars. So much I could help out for it now (20%).
• 55 dollars. I am going to drink wine, which costs $ 75, but I paid only 20 dollars for it, so it turns out that I will save money by drinking this bottle (25%).

In this case, the best option from the point of view of the economic theory of rational behavior was the option of $ 75, since the cost of an alternative opportunity is to sell wine for $ 75 instead of drinking it. However, more than half decided that drinking a bottle of wine worth $ 75 is the equivalent of drinking it for free or saving money on a purchase.



Richard Thaler created a whole theory of behavioral economics, which proves that people are mostly guided by emotions and inner convictions, and not by logic when making decisions in the market. The emotional component is formed by the previously spent money and time to enter the market, market experience, incentives in the form of support from partners and colleagues, as well as the concept of fair prices in the market and the time needed to make decisions.

N effect

We regularly analyze the conducted bidding and noticed an interesting feature. The theory of rational behavior suggests an increase in aggressiveness with more competitors. Everything is clear: since there are more applicants for a tasty contract, it makes sense to invest much more in order to achieve success. Real graphics turned out to be much more interesting. When the participants became more than a certain psychological threshold, their activity in the auction began to decline sharply, and each individual player quickly lost interest, passing to passive tactics.



We hypothesized that this is a manifestation of the so-called “N-effect” described by behavioral scientists Stephen Garcia and Absalom Thor in 2009. They revealed a very strange anomaly in the results of ACT (American College Testing). The highest average scores were obtained in states where traditionally low population density. Students get to the exam venue for a long time, and fewer people participate in the test itself. Researchers have tried to standardize statistics on average income, the percentage of national minorities and a bunch of other factors. Nevertheless, a sample of sparsely populated states consistently yielded better results. Garcia and Thor called it the "N-effect." The more people perform the same task, the less they "try." When a small number of participants participate in the competition, each of them tries to prove himself as best as possible against the background of the rest. In fact, this phenomenon is very similar to the “witness effect”, when in a huge crowd no one dares to take responsibility for helping the victim. If the number of rivals is too large, then we lose the meaning of competition and personal interest.



Initially, in the framework of school education, this effect was explained by the fact that smaller classes interact more effectively with the teacher. However, Stephen Garcia defined this effect as a consequence of a more intense rivalry between students.

Our observations

We took statistics on electronic auctions held on the Unified Electronic Trading Platform under the 44- over the last couple of years and tried to play around with models. We arbitrarily divided all bids into three groups depending on the outcome:

1. The original contract price has declined (66%).
2. Price has not changed (28%).
3. The price increased at the stage of conclusion or execution (6%).

We were interested in identifying from a variety of factors those that had the greatest influence on the final result. For this, we applied the method of principal components, highlighting the factors with the greatest weight. As a result, two factors turned out to be the most significant: the size of the bid security and the number of bidders. In order to more strictly assess the influence of variables on the final price, we built a regression model with the dependent variable log (PriceDrop) - the logarithm of the difference between the initial and final prices of the trades found earlier by influence factors. The model was built for those trades that ended in lower prices:

1. With an increase in application security by 1%, the average price decreases by 0.52%.
2. With an increase in the initial price of the contract by 1% - by 0.38%.
3. Each submitted application is reflected in a decrease in prices by an average of 0.09%.
4. Each eligible application is 0.03%.
5. With an increase in contract enforcement by 1%, the average price decreases by 0.06%.

Immediately the thought arises that the ideal option for the buyer would be a huge crowd of suppliers who are fighting to their death for his contract. And here comes the very N-effect. Suppliers really gnaw out the contract until the last while there are three or four of them, but it’s worth approaching the critical threshold of six to nine for different groups of goods / work / services, as all competition begins to decline. A further increase in the number of participants only reduces the fall in the final price. We tried to get feedback from participants in such "mass" tenders, and one of the main complaints was the lack of time to navigate the chaos of continuous price offers, which took 10 minutes.



Observations, of course, are empirical, but nevertheless fit very well on the theory of Garcia and Thor. As a result, paradoxically, too many competitors sometimes only reduce competition.



The study was conducted by the specialists of EETP JSC together with the senior lecturer of the State University of Management Svetlana Suyazova (Aksyuk).