I want to do it in 3 different versions for different lengths, details, and complexity.
Tweet format
It explores why people trust a well connected person's recommendation when they know seller invests disproportionately in those Influencers (139 characters including space)
Elevator Pitch format
This paper looks at how sellers form a reputation for themselves and also tries to think about why people believe in others' endorsements. I focus on exploring how different types of learning (tastes and skills) affect investment behaviour by the seller. I find that overall, having a more connected network is beneficial for the buyers (as they enjoy higher levels of investment) but may result in lower system surplus.
Essay format
In short, my thesis explores reputation dynamics when a seller can invest differently in different buyers. Some examples of such behaviour include management consulting (where the consultant can distribute their effort to different projects that they handle), restaurant chefs (an example we'll delve into in a moment), and investment portfolio managers (again, splitting their effort between different portfolios.)
Once you go to page 6 of my thesis, you'll realize that it gets a bit mathematical. For those who have done some game theory then the math shouldn't look too unfamiliar but it can definitely be confusing. This paper essentially describes and explores the following situation:
Imagine that you're deciding where you're going to be eating dinner. You have 2 choices. One restaurant that has received favourable reviews from many reviewers and food critic. The other restaurant has one (and only one) positive review on Yelp. If the price is the same, the cuisine is the same, and booking is not an issue, most people (myself included) will choose the restaurants that has received many positive reviews.
But restaurants know this fact. So let's think one step further. Put yourself in the shoe of the chef that has to cook hundreds of meals for a service. And you're facing two tickets. The server has informed you that one of the ticket likely belongs to one of those food reviewers (youtuber, influential blogger... you name them), or anyone with a wide audience. If you have to choose between investing more in one diner's dinner, it would probably be the acclaimed food reviewer's.
And here's the puzzle, as customers, we know that this is the case. We know that the food critic (or anyone that can communicate to a wide audience) probably received a better service and better food that we will. In other examples of the same phenomenon, we know that shoes or brands that received celebrity endorsement will not live up to our expectation because of our understanding of endorsement contracts.
So why is it that we still buy those things? Why do we still see value in those endorsement that we know results from investments by the seller that we're not likely to receive ourselves?
What I focus on is 2 different learning processes. Seller learns how to serve a good quality product better by serving good quality products (a classic case of learning by doing.) The first type, which I termed taste learning, is the kind of learning you get by serving person A that you can only use to serve person A better. When you learn about a specific person's taste, it's hard to apply that same kind of learning to other people. The second type, which I termed skill learning, is the kind of learning you get from serving person A well that you can use to serve person B better. In other words, you learn the important skills in serving something.
Now, let's add another layer of abstraction and think about how these two effects interact with one another. Let's say you serve person A a good meal and as a result, learns their taste, say they want a bit more pepper in their meals, and you also learn a skill, how to put more pepper in the meal without breaking the overall taste. Now, the skill you learnt is tied to the taste profile for each buyers. This is to say, that you can't really use this new skill if you don't know the new diner's taste profile! Which means that without knowing a person's taste (which you only get by serving them well), you can't fully use the skill learning you got by serving another person well. Some skill learning obviously doesn't fall into this category and thus you can fully utilize those skill without knowing the other person's taste.
Finally, let's add the reputation aspect to this model. I use a very simple framework here. There are 2 buyers and one of the buyers (Influencer) can communicate their quality realization to the other buyer (Consumer.) Which means that the Consumer has access to 2 quality realizations: their own and the Influencer's. The Influencer on the other hand only observes their own realized quality. The expected quality that each buyer comes up with is the seller reputation.
I then explore equilibrium behaviour (which means given everyone's actions, people don't have an incentive to move away from what they're doing) in this model, specifically on how the seller invests in these two buyers. What I find is not very surprising: that a limited network means that there's an imbalance in the level of investment between the Influencer and the Consumer. This tends to diminish as the importance of taste learning increases (so seller can't take advantage of the expectation of skill learning.) Further, as complementarity between the two learning effect increases, you have less incentives to invest in both (since another person's success is now more important.)
Finally, a more connected network (where both people observes each other's realization) is better off for the buyers since they get a high level of investment. However, overall surplus, that is once we take seller costs into account, is higher ex-ante in limited observation network.
Hopefully in the future I can extend this to a more complex network and come up with empirical specifications to test this idea, but it was definitely a fascinating topic to explore.