The Functioning of Self-Governing Sharing Economies.

This study collects field data on an unregulated sharing economy to observe the extent to which these markets function without regulation. In the field experimental part, which we preregister, a sample of local markets is randomized into a treatment and control group. After baseline data have been collected, members of the treatment group receive a positive supply shock and we test whether this shock has a transitory or more permanent effect.

The field experimental intervention is part of our study in which we mainly collect field data.

Our study starts with a one month period in which we collect for each participating free library baseline data about the number of books in their library, the (second-hand) value of each book and the weekly number of transactions. Second-hand price data in this period are collected from the website of a large Dutch online retailer (Bol.com). On this website, the general public can place all types of second-hand goods for sale. It functions as a platform, as the buyer receives the second-hand goods directly from the seller. We determine the average price of each book by looking at all offerings listed under the book's unique ISBN number or author and title. Books that are registered without an ISBN code are dropped from the sample, unless an author and title are specified. The average price of each second-hand book is determined by taking the average of all price quotes.

This information is used to stratify the libraries into a treatment and control group in week 5. We stratify along the following two dimensions: number of books and the population in the 4digit zipcode in which the library is located. Four subgroups or strata are constructed based on the population and number of books being in the lower or upper 50 percent. We then randomly assign in week 5 of the study half of the libraries within each stratum to the treatment group and the other half to the control group.

Owners of libraries in the treatment group are then send by mail an identical package of 3 books. The three selected book-titles are all novel and best-selling titles. The sales price of this package in stores is about 50 euro. At the time of sending the estimated second-hand price of the package is around 47 euro, around 15 per book. This is higher than the average second-hand book price in the collection of books as reported by the library owners in the first week of the study (which was around 9.5 euro). Hence, when these books appear in the libraries, they will in most libraries be among the most valuable books. The treatment hence positively shocks the quality of local libraries as measured by second-hand market value.

Owners are asked to put these books in their library in the sixth week of the study. Library owners in the control group receive the same package of books but they receive them in August or September 2019 such that in the intermediate period, these libraries serve as the control group, showing how transactions evolve without the positive supply shock mentioned above.

Of course, the value of the package of books is likely to change over time. These titles are just on the market when the libraries in the treatment group receive them but considerably longer when the libraries in the control group receive them. To account for this difference in value (which we expect to decrease over time), the rating of the individual book value is repeated in every week of the study, using the current second-hand market value as listed on Bol.com. We not only do this for the package of three books but for all books present in the libraries. In this way, we correct for the common value depreciation of books. (Even if no transactions take place in a library, the value of its collection is likely to decrease.) The books that are not listed on Bol.com will be searched for individually on Boekwinkeltjes.nl, a website specialized in second-hand book sales.

We have three hypotheses regarding the field experimental intervention.
1) The three books (our treatment) disappear faster from the library compared to books in the control group.
2) The three books (our treatment) are brought back to the library less often compared to books in the control group.
3) In the period post treatment, books that newly enter libraries in the treatment group are taken out of the library faster than books that newly enter libraries in the control group. This measures the multiplier effect from our intervention. In the test of this third hypothesis, we exclude from the analysis the three books from our treatment package that were sent to the libraries in the treatment group.

2019-06-10

2019-09-15

The primary outcome variable of our field experimental intervention is the turnover rate of our experimental units compared to the control units.

The secondary outcome variable of our field experimental intervention is the multiplier effect that our experimental units have on new non-experimental units in the treatment group compared to new control units.

For the field experimental part of this study, we randomize a number of local markets into a treatment and control group. After baseline data have been collected, members of the treatment group immediately receive a positive supply shock while members of the control group receive this shock later in the study (a few weeks before the end).

The study started in May and lasts for almost six months. Each owner submits on a weekly basis the stock of books in their library and indicates of each book whether it was added or taken out by themselves or by a passenger.

The study starts in May 2019. During the first five weeks of the study, no treatment variation takes place. This covers the baseline period in which owners weekly register the stock of books in their library. After this period, in June 2019, the treatment variation is implemented. Hence, the months June, July and August 2019 are our experimental period. The libraries in the control group will receive the treatment in August or September 2019. 30 libraries are allocated to the treatment group and 28 libraries to the control group.

The treatment consists of a package of books that each owner receives to be placed in their library. The package of books is exactly the same for the treatment and the control group, only the timing of delivery differed. Owners in the treatment group received the package of books one month after the start of the study, while owners in the control group will receive the package of books in August or September 2019. The value of the package is set at 50 euro (around 47 euro second-hand price, around 15 per book) such that it was higher than the average second-hand book price as reported in the first week of the study (which was around 9.50 euro).

Owners of libraries in the treatment group (nor in control group) are not informed about the experimental variation in book arrivals. Owners of libraries in the control group are informed about the timing of their delivery. All owners of the libraries (control and treatment group) continued the study in the experimental phase as before, so they submitted on a weekly basis the stock of books.

In april-august 2020, a survey was sent out to the library owners and flyers were placed in the libraries to invite library users to complete the survey. The survey contained questions about why owners started their library, and why users visited the library and how often. Additionally, both versions of the survey contained a social norms measure (Krupka and Weber) and a standard public goods game. The social norms measure consisted of 7 items that which owners and users evalueted on a 1 (= very socially inappropriate) to 4 (= very socially appropriate) scale. The PGG consisted of investing 10 euros in the common project (all contributions*1.6 and divided over 4 participants) or keep it for themselves.

30 local markets were allocated to the treatment group and 28 to the control. Randomization was stratified based on four-digit postal codes of the local market and the number of observations per local market as reported in the first week of the study. Randomization was done in the office by a computer using Stata-code.

Local market.

No

58 local markets.

58 local markets, per local market a time series of 24 observations will be collected.

28 local markets in the control group, 30 local markets in the treatment group