Payday loans like elastic - Reading

This paper reports the results of a large-scale field experiment designed to test the hypothesis that group membership can increase participation and prosocial lending for an online crowdlending community, Kiva. The experiment uses variations on a simple email manipulation to encourage Kiva members to join a lending team, testing which types of team recommendation emails are most likely to get members to join teams as well as the subsequent impact on lending.
We find that emails do increase the likelihood that a lender joins a team, and that joining a team increases lending in a short window (1 wk) following our intervention.

The impact on lending is large relative to median lender lifetime loans. We also find that lenders are more likely to join teams recommended based on location similarity rather than team status. Our results suggest team recommendation can be an effective behavioral mechanism to increase prosocial lending.,To increase member engagement, some online communities have created group structures.
For example, in 2008, Kiva instituted a lending teams program, a system through which lenders can create teams or join existing teams of other lenders. Once a team is created, it appears on Kiva’s team leaderboard, which sorts teams by the total loan amounts designated to them by their team members.

Since 2008, more than 38,957 Kiva teams have been created based on lender group affiliations such as organizations, geographic location, religious affiliation, or sports interests. Of note, many of the highly ranked teams are identity based, such as the “Atheists” and the “Kiva Christians.” Each team has a dedicated forum where team members can coordinate their lending activities, ask and answer questions, and set goals for the team.,In our study, we use a lender’s likelihood of joining a team to recommend teams based on both homophily and status.
Homophily refers to the tendency to associate with similar others (38, 39). As such, we recommend teams to lenders based on their similarity to the existing members of those teams.
In our study, we use two different measures of homophily: location similarity and loan history similarity. The former is based on the number of lenders in a team who share the same location as the target lender, whereas the latter is based on how often the lenders have lent to the same borrowers.

In addition to homophily, we recommend teams based on status (40), using the top three teams on the Kiva leaderboard as the high-status teams.,To study the causal effects of team recommendations on the likelihood of joining a team and increasing contributions, we use a group of 69,802 lenders who have made at least two loans in the past 6 mo but have never joined a team. We then randomly assign each lender to one of eight experimental conditions with equal probability.
Pairwise Kolmogorov–Smirnov tests based on observable characteristics verify that our randomization works (Table S2).,Sample and population comparison. The number of lenders and median number of loans of all public users, those who are selected as participants, those whose data are used in our analyses, and those who joined at least one team during our experiment.,Although lenders in the control condition were not contacted during the experiment, for each treatment, we sent one of five email messages.

Each email consists of three parts. Part 1 is common to all treatments and the placebo,“Hi [FirstName], Since you’re such an awesome Kiva lender, we wanted to let you know about a fun feature of the Kiva experience: Kiva Lending Teams!payday loans pomona ca
Lending Teams are self-organized groups around shared interests—location, alumni orgs, social causes, you name it. You can connect with other lenders, discover loans you might be interested in, and track your collective impact.”,“Based on your past lending, people who have made similar loans enjoy being a part of these teams: [TEAMS].”,The simplest recommendation strategy is to recommend teams that are ranked highly on the team leaderboard. Kiva provides several leaderboards that rank teams based on either the total loan amount attributed to the team or the number of team members, in the most recent month or all time.

For the experiment, we use the default leaderboard that lenders see when they visit the Kiva Team page, the all-time total amount lent.,We also construct a recommender system based on the loan history of a lender. This is motivated by the homophily conjecture that lenders who lend to similar borrowers share similar interests and are thus more likely to join the same teams.,• Their pages and loans are set to public in their account settings.,• They have made at least two nonpromotion loans in the past 6 mo.,We then assign each user to one of the treatments, the placebo, or the control condition using stratified randomization.
The stratified random assignment is based on the total loan amount by each lender before the experiment. We want to ensure that the most active Kiva lenders are not all concentrated into one treatment, so we rank the lenders based their total loan amounts, taking the top eight lenders and randomly assigning them to different conditions. We then repeat this for each group of eight lenders, proceeding down the ranked list.

Between assigning lenders to conditions and running the experiment, 43 users joined a team and were dropped from our sample.
This yields a final sample of 69,802 users. The size of the sample and population is summarized with a Venn diagram in Fig.
S2.,Before running the experiment, we run pairwise Kolmogorov–Smirnov tests of the equality of distributions based on the user statistics to verify that our randomization produces balanced treatments across observable characteristics. The results of these tests show that the number of loans, average amount per loan, balance, average loan terms for fundraising or repayment, and autolending settings do not differ significantly at the 10% level between any treatments. Thus, the Kolmogorov–Smirnov tests do not reject the hypothesis that these values are drawn from the same distribution.

We summarize the lending and location statistics of each treatment in Table S2.,We next explore which types of teams lenders are most likely to join by examining the characteristics of teams joined by our lenders. Table 2 displays the results of eight conditional logit specifications with odds ratios reported, with one specification per treatment.
In our regressions, we use whether each lender joined each team as our dependent variable, and location similarity, loan history similarity, team status, team size, and experimenter recommendation as our independent variables.,Interestingly, we find that the provision of a location similarity recommendation mitigates the influence of team status, leading lenders to join recommended teams (columns 3 and 4) or teams with higher history similarity (column 4). By contrast, our recommendations based on loan history similarity (columns 5 and 6) do not substantially change how lenders choose their teams. Finally, recommendations based on team status (columns 7 and 8) seem to decrease the importance of lending history.,This effect is also insignificant beyond 1 wk.

One possible reason for the lack of an observed long-term effect is that lenders may wait until initial loans are repaid before lending again, a process which may take 12 to 18 mo. However, even the 1-wk effect ($392) is more than 15 times the lifetime contribution of the median Kiva lender ($25), indicating that team membership is effective in increasing member contributions on those lenders who would join a team because of our email.,This paper reports the results of a large-scale field experiment designed to test the hypothesis that team membership can increase participation and lending for an online crowdlending community, Kiva. We find that emails increase the likelihood that a lender joins a team, and that joining a team increases lending in an 1-wk window following the decision to join. Although this experiment does not explore the mechanism through which joining a team increases giving, our prior empirical analyses and field experiment point to two mechanisms at work (4).
First, joining a team increases information sharing about specific borrowers on the team forum, which reduces team members’ search costs and increases their lending. Second, joining a team increases the pressure to help improve the team’s ranking on the Kiva leaderboard.

Therefore, effective teams share information and coordinate their loans to reduce search costs, and emphasize team competition through goal setting. Our results suggest that recommending teams to members of an online lending community based on homophily is an effective mechanism to engage community members and increase theircontributions.

free payday loans for first time customers