Linguistic Features and Peer-to-Peer Loan Quality: A Machine Learning Approach

Gao, Qiang; Lin, Ming-Chyuan

doi:10.2139/ssrn.2446114

Cited by 28 publications

(21 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We follow the procedure mentioned in Tweedie and Baayen () in calculating a variant of the Simpson's Diversity Index to approximate the diversity of language used in a description. Originally developed for use in ecology to quantify the level of biodiversity in an environment, researchers in other fields have adapted Simpson's measure of diversity for a variety of other applications, including textual analysis (Gao and Lin ). The formula used to calculate Simpson's Diversity Index is:

{SIMPSON}_{i} = 1 - ((), \sum_{w} n_{w} ((), n_{w} - 1) true/ N ((), N - 1))

where n w is the total number of times word w appears in d i and N is the total number of unique words in all of the d i .…”

Section: Methodsmentioning

confidence: 99%

Small Business Borrowing and Peer‐to‐peer Lending: Evidence From Lending Club

Nowak

Ross

Yencha

2017

Contemporary Economic Policy

View full text Add to dashboard Cite

We investigate the ability of small business borrowers to signal to investors their credit worthiness through the use of text descriptions in the peer‐to‐peer lending market. Specifically, we examine the relationship between the loan description written by a borrower and whether or not the project is funded by investors. Using textual analysis, we find that small business loan descriptions can be used to predict the likelihood that the loan will be funded. We also find that an index, created from a textual analysis of the words used in the loan description, can forecast the performance of the loan, specifically whether or not the loan defaults. This index has the strongest impact when we focus on borrowers with low FICO scores, suggesting that for these individuals the description can signal information that standard measures used for lending purposes cannot. Overall, it appears as though investors are making investment decisions based on proper and relevant signals given by the borrowers through the loan description. (JEL D47, D53, D82, D83, G14, G21)

show abstract

{SIMPSON}_{i} = 1 - ((), \sum_{w} n_{w} ((), n_{w} - 1) true/ N ((), N - 1))

where n w is the total number of times word w appears in d i and N is the total number of unique words in all of the d i .…”

Section: Methodsmentioning

confidence: 99%

Small Business Borrowing and Peer‐to‐peer Lending: Evidence From Lending Club

Nowak

Ross

Yencha

2017

Contemporary Economic Policy

View full text Add to dashboard Cite

show abstract

“…Furthermore, previous research found that the length of descriptions in crowdfunding campaigns have a significant positive effect on the campaign outcome (Greiner and Wang 2010;Gao and Lin 2014). Longer descriptions can deliver more information about the project, the start-up, or the product and can help to reduce information asymmetries between the start-up and potential investors.…”

Section: Clarity Of Updates and Its Effects On Crowd Participationmentioning

confidence: 99%

Which updates during an equity crowdfunding campaign increase crowd participation?

2017

View full text Add to dashboard Cite

Start-ups often post updates during equity crowdfunding campaigns. However, little is known about the effects of such updates on crowd participation. We investigate this question by using hand-collected data from 71 funding campaigns and 39,399 investment decisions on two German equity crowdfunding portals. Using a combination of different empirical research techniques, we find that posting an update has a significant positive effect on the number of investments made by the crowd and the investment amount collected by the start-up. This effect does not occur immediately in its entirety; rather, it lags the update by a few days. Furthermore, the effect of updates loses statistical significance with the number of updates posted during a campaign. We also find that an easier language used in updates increases crowd participation, whereas the length of updates has no effects. With respect to the update's content, we find that the positive effect can be attributed to updates about new developments of the start-up such as campaign developments, new funding, business developments, and cooperation projects. Updates on the start-up team, business model, product developments, and promotional campaigns do not have meaningful effects. Our paper contributes to the literature on the effects of information disclosure on equity crowdfunding participation. Furthermore, our results have practical implications for start-ups and their investor communication during equity crowdfunding campaigns.

show abstract

“…Herzenstein, Sonenshein, and Dholakia (2011), through text analysis of borrower narratives, confirm limited usefulness. Gao and Lin (2012) use text mining and determine that more complex narratives are correlated with higher default rates. Yencha, Nowak, and Ross (2018) also used text mining and find that text descriptions of small businesses can predict whether a small business loan will be funded.…”

Section: The Literaturementioning

confidence: 99%

The roles of alternative data and machine learning in fintech lending: Evidence from the LendingClub consumer platform

Jagtiani

Lemieux

2019

Financial Management

219

View full text Add to dashboard Cite

There have been concerns about the use of alternative data sources by fintech lenders. We compare loans made by LendingClub and similar loans that were originated by banks. The correlations between the rating grades (assigned by LendingClub) and the borrowers’ FICO scores declined from about 80% (for loans originated in 2007) to about 35% for recent vintages (originated in 2014–2015), indicating that nontraditional data (not already accounted for in the FICO scores) have been increasingly used by fintech lenders. The rating grades perform well in predicting loan default. The use of alternative data has allowed some borrowers who would have been classified as subprime by traditional criteria to be slotted into “better” loan grades, allowing them to obtain lower priced credit.

show abstract

Linguistic Features and Peer-to-Peer Loan Quality: A Machine Learning Approach

Cited by 28 publications

References 29 publications

Small Business Borrowing and Peer‐to‐peer Lending: Evidence From Lending Club

Small Business Borrowing and Peer‐to‐peer Lending: Evidence From Lending Club

Which updates during an equity crowdfunding campaign increase crowd participation?

The roles of alternative data and machine learning in fintech lending: Evidence from the LendingClub consumer platform

Contact Info

Product

Resources

About