Interdependence between online peer‐to‐peer lending and cryptocurrency markets and its effects on financial inclusion

Simple analytical framework

We develop a simple analytical framework to illustrate how the cryptocurrency market affects a self‐interested platform's decision‐making and payoff and consequently financial inequality of borrowers in the P2P lending market. With lax screening and 24/7 operations, in real time, the lenders are not the only ones who consider the cryptocurrency market, but the borrowers can also consider this market as an attractive investment source. To incorporate this potentiality, we consider two‐sided outside options. We first exploit key considerations in the posted price P2P lending model (Einav et al., 2018; Wei & Lin, 2017), and then include the cryptocurrency market as an outside option for both the lender and the borrower.

At stage 1 of the game sequence, the platform sets an interest amount, p, under the posted price mechanism to maximize its own payoff, as a self‐interested player, by attracting both sides to participate in the transaction (Einav et al., 2018; Wei & Lin, 2017). At stage 2, the borrower, given p, decides whether to accept the interest if it is less than or equal to the threshold interest amount,

p ̂ $\hat{p}$

For each loan, the borrower incurs a variable cost of c and obtains a random value of v. Without a new outside option, the borrower is willing to pay a maximum interest rate for a loan of

v − c $v-c$

π B = v − p − c $\pi _B = v -p-c$

π B < 0 $\pi _B <0$

δ ( 1 + p ) $\delta (1+p)$

δ ∈ [ 0 , 1 ] $\delta \in [0,1]$

π E = p − δ ( 1 + p ) − u $\pi _E = p - \delta (1 + p) - u$

π E ≥ 0 $\pi _E \ge 0$

γ ≥ 0 $\gamma \ge 0$

max p π S P = D B · D E . $$\begin{equation} \text{max}_p \pi _{SP} = D_B \cdot D_E. \end{equation}$$

1

π B = v + τ γ s − p − c ≥ 0 . $$\begin{equation} \pi _B = v + \tau \gamma s - p - c \ge 0. \end{equation}$$

2

π E = p − δ ( 1 + p ) − u − τ s ≥ 0 . $$\begin{equation} \pi _E = p -\delta (1+p) - u - \tau s \ge 0. \end{equation}$$

3

The borrower and the lender decide to participate in the loan transaction if they obtain a positive payoff. Let

D B $D_B$

D E $D_E$

k · D B · D E $k \cdot D_B \cdot D_E$

p ∗ $p^*$

To capture the impact of the outside shock on a lender's decision, we first consider the heterogeneous reservation utility of lenders for the loan. For mathematical tractability, we assume a uniform distribution of

[ u ̲ , u ¯ ] $[\underline{u}, \overline{u}]$

D E = ∫ u ̲ u ̂ 1 d u $D_E=\int _{\underline{u}}^{\hat{u}} 1 du$

u ̂ = p − δ ( 1 + p ) − τ s $\hat{u}=p-\delta (1+p)-\tau s$

D E $D_E$

u ̲ $\underline{u}$

u ¯ $\overline{u}$

[ v ̲ , v ¯ ] $[\underline{v}, \overline{v}]$

Lemma 1

(a)

∂ p ∗ ∂ τ > 0 $\frac{\partial p^*}{\partial \tau } >0$

, (b)

∂ 2 p ∗ ∂ τ ∂ δ > 0 $\frac{\partial ^2 p^*}{\partial \tau \partial \delta } >0$

, and (c)

∂ 2 p ∗ ∂ τ ∂ γ > 0 $\frac{\partial ^2 p^*}{\partial \tau \partial \gamma } >0$

.

For the sake of space, we describe the interpretation of Lemma 1 in Appendix B.2 in the Supporting Information. Proposition 1

(a) If and only if

δ ≤ 1 − 1 γ $\delta \le 1-\frac{1}{\gamma }$

∂ D B ∗ ∂ τ ≥ 0 $\frac{\partial D_B^*}{\partial \tau } \ge 0$

∂ D E ∗ ∂ τ ≥ 0 $\frac{\partial D_E^*}{\partial \tau } \ge 0$

∂ D B ∗ ∂ τ < 0 $\frac{\partial D_B^*}{\partial \tau } < 0$

∂ D E ∗ ∂ τ < 0 $\frac{\partial D_E^*}{\partial \tau } < 0$

∂ 2 D B ∗ ∂ τ ∂ δ < 0 $\frac{\partial ^2 D_B^*}{\partial \tau \partial \delta } < 0$

∂ 2 D B ∗ ∂ τ ∂ γ > 0 $\frac{\partial ^2 D_B^*}{\partial \tau \partial \gamma } > 0$

∂ 2 D E ∗ ∂ τ ∂ δ < 0 $\frac{\partial ^2 D_E^*}{\partial \tau \partial \delta } < 0$

∂ 2 D E ∗ ∂ τ ∂ γ > 0 $\frac{\partial ^2 D_E^*}{\partial \tau \partial \gamma } > 0$

Proposition 1 summarizes under what conditions

D B ∗ $D_B^*$

D E ∗ $D_E^*$

δ ≤ 1 − 1 γ $\delta \le 1-\frac{1}{\gamma }$

p ∗ $p^*$

τ · γ $\tau \cdot \gamma$

p ∗ $p^*$

1 − δ $1-\delta$

With Propositions 1 (b) and 1 (c), we explain the case of

δ > 1 − 1 γ $\delta > 1-\frac{1}{\gamma }$

D B ∗ $D_B^*$

D E ∗ $D_E^*$

∂ 2 p ∗ ∂ τ ∂ δ > 0 $\frac{\partial ^2 p^*}{\partial \tau \partial \delta }>0$

p ∗ $p^*$

D B ∗ $D_B^*$

p ∗ $p^*$

( 1 − δ ) $(1-\delta )$

D E ∗ $D_E^*$

δ = 1 − 1 γ $\delta =1-\frac{1}{\gamma }$

Regarding the moderation effect of γ on the relationship between τ and

D B ∗ $D_B^*$

D E ∗ $D_E^*$

δ ≤ 1 − 1 γ $\delta \le 1-\frac{1}{\gamma }$

δ > 1 − 1 γ $\delta > 1-\frac{1}{\gamma }$

γ ≥ 1 ( 1 − δ ) $\gamma \ge \frac{1}{(1-\delta )}$

γ < 1 ( 1 − δ ) $\gamma < \frac{1}{(1-\delta )}$

p ∗ $p^*$

γ < 1 ( 1 − δ ) $\gamma < \frac{1}{(1-\delta )}$

γ · τ $\gamma \cdot \tau$

p ∗ $p^*$

p ∗ $p^*$

Hypothesis development

On the basis of Proposition 1, we develop hypotheses related to the dynamics between cryptocurrency availability and financial inclusion. Prior studies on financial inclusion characterized financially advantaged groups as high‐creditworthy individuals who live in urban areas and/or have high‐skill occupations and finally disadvantaged populations as low‐creditworthy individuals residing in rural areas and/or employed in low‐skill occupations (e.g., Ozili, 2018; Salampasis & Mention, 2018). Financial inclusion is intended mainly to extend financial services to financially disadvantaged groups. Building on these previous works, we focus our analytical model on two factors that can help differentiate the financially disadvantaged from their advantaged counterparts: (1) the creditworthiness of borrowers (denoted as δ), and (2) the value of the outside option for borrowers (relative to lenders) (interpreted as technological savviness and denoted as γ).

We first associate δ with the creditworthiness of the borrowers, which has been the most salient factor for classifying individuals as finally advantaged (Buchak et al., 2018; Maskara et al., 2021). Prior studies showed that as creditworthiness improves, access to credit becomes easier. We thus incorporate this factor into our model. Another important, subtle factor is the relative value of the outside option for borrowers relative to lenders (denoted as γ). On the basis of this definition, borrowers with a high γ are likely to use the cryptocurrency market as a new investment option. We argue that borrowers should have enhanced technical knowledge about cryptocurrencies and blockchains in general to enable them to more effectively seize and capitalize on this opportunity. Because the cryptocurrency market is based on blockchain technology, technology sophistication should play an important role in the exploration of financial inclusion (Liu et al., 2021). Accordingly, to complement previous research, we conceptualize high γ borrowers as tech‐savvy individuals that represent a financially advantaged group. This clears the way for us to examine whether technological savviness (especially on the cryptocurrency market) can inhibit financial inclusion.

By reorganizing the condition underlying Proposition 1(a), we could observe that a bullish (bearish) cryptocurrency market may increase (decrease) the likelihood of participation by high creditworthy individuals (i.e.,

∂ D B ∗ D E ∗ ∂ τ ≥ 0 $\frac{\partial D_B^* D_E^*}{\partial \tau } \ge 0$

δ ≤ 1 − 1 γ $\delta \le 1-\frac{1}{\gamma }$

∂ D B ∗ D E ∗ ∂ τ ≥ 0 $\frac{\partial D_B^* D_E^*}{\partial \tau } \ge 0$

γ ≥ 1 ( 1 − δ ) $\gamma \ge \frac{1}{(1-\delta )}$

Contrastingly and interestingly, we can observe from Proposition 1(b) that a bullish (bearish) cryptocurrency market may decrease (increase) demand from low creditworthy or non‐tech‐savvy borrowers. First, a lender who targets low‐creditworthy borrowers may be attracted to the cryptocurrency market—a situation that may not be sufficiently offset by an increased interest rate, especially under a high default rate. Second, low‐creditworthy borrowers are discouraged from participating in the market by a high interest rate. Non‐tech‐savvy borrowers may not recognize opportunities in the cryptocurrency market, thus resulting in limited benefits from a bullish cryptocurrency economy. However, a platform will increase the interest rate to retain a lender seduced by the outside option, but such a rate is insufficient for the lender to stay because the increased interest comes mostly from lender demand and less from that by non‐tech‐savvy borrowers.

In sum, we theorize possible changes in financial inclusion in two FinTech markets by examining how P2P borrowing by financially advantaged (disadvantaged) groups is influenced by opportunities in the cryptocurrency market (see Appendix B.3 in the Supporting Information for graphical illustration). Specifically, we inquire into whether a bullish (bearish) cryptocurrency market provides more opportunities to either low‐creditworthy or non‐tech‐savvy borrowers who have been excluded from traditional financial services or whether the market reduces such exclusion. With Proposition 1, especially 1(c), as a basis, we submit the following: Hypothesis 1

Positive (negative) shocks in the cryptocurrency market are associated with higher (lower) access and usage from high‐creditworthy borrowers than low‐creditworthy borrowers.

Hypothesis 2

Positive (negative) shocks in the cryptocurrency market are associated with higher (lower) access and usage from tech‐savvy borrowers than non‐tech‐savvy borrowers.

Because we use the two attributes that distinguish financially advantaged groups from disadvantaged populations in the P2P lending market, another interesting yet unexplored question is which factor has a stronger impact on the link between cryptocurrency shocks and P2P borrowing. Thus, we compare the moderating effect of the technological savviness and creditworthiness of borrowers. The cross partial derivatives of borrower demand with respect to shock and creditworthiness and technological savviness are

| ∂ 2 D B ∗ ∂ τ ∂ δ | = s 2 ( 1 − δ ) 2 ( v ¯ − v ̲ ) $\vert \frac{\partial ^2 D_B^*}{\partial \tau \partial \delta } \vert = \frac{s}{2(1-\delta )^2(\overline{v}-\underline{v})}$

| ∂ 2 D B ∗ ∂ τ ∂ γ | = s 2 ( v ¯ − v ̲ ) $\vert \frac{\partial ^2 D_B^*}{\partial \tau \partial \gamma } \vert = \frac{s}{2 (\overline{v}-\underline{v})}$

| ∂ 2 D B ∗ ∂ τ ∂ γ | ≥ | ∂ 2 D B ∗ ∂ τ ∂ γ | $\vert \frac{\partial ^2 D_B^*}{\partial \tau \partial \gamma } \vert \ge \vert \frac{\partial ^2 D_B^*}{\partial \tau \partial \gamma } \vert$

D B ∗ $D_B^*$

Data and variables

We downloaded data from Prosper.com on April 12, 2019. The data set includes all fund seekers on the company platform since its inception in 2006. For each listing, we obtained a comprehensive set of list characteristics, including the requested amount, borrower's interest rate, loan term, starting and ending time, and loan purpose. Furthermore, for successfully funded listings, we tracked the loan origination date and repayment status each month, which helps define whether a loan is defaulted. The data set also includes borrower‐specific characteristics such as Prosper rating (AA‐A‐B‐C‐D‐E‐HR) and debt‐to‐income ratio. ¹⁰ Because we are interested in differences in borrowers' reactions to cryptocurrency markets, we also collected their occupations and residential states.

We collected aggregate data from Coinmarketcap.com, which provide aggregate information on trading activities on key exchanges for Bitcoin, Ethereum, and other major cryptocurrencies. As of May 20, 2019, 2186 coins were trading in the market. Bitcoin and Ethereum account for 68% of the total cryptocurrency market capitalization. Thus, we focused on Bitcoin and Ethereum in our main analysis. In a robustness test, we included the top 10 cryptocurrencies in terms of market capitalization, which covers 83% of total market capitalization.

To examine cryptocurrency market effects on borrowing in the P2P lending market, we matched all listings from Prosper created from January 1, 2017 to February 28, 2019, with the cryptocurrency market data. We used the listing creation date because it should be closest to the actual borrower reaction in the cryptocurrency market. We then conducted analyses at two levels. To examine the effect on the overall volume of borrowing (i.e., the extensive margin), we aggregated and analyzed both data sets at the day level. To examine the effect on individual loans (i.e., the intensive margin) and resulting outcomes, we analyzed at the listing level. Using a period of more than 2 years allowed us to cover both boom‐and‐bust cryptocurrency periods. Figure 1 shows that the Bitcoin price rapidly increased in 2017 and then dropped in 2018. Thus, we can examine how cryptocurrency market shocks over the course of the cycle affect borrowers' incentives for using P2P lending markets.

OPEN IN VIEWER

FIGURE 1

The price of Bitcoin and exogenous shocks in the cryptocurrency markets

To capture the overall interest in P2P lending, we first computed the total amount of loan application at the day level. We also constructed the daily number of listings and the daily number of unique borrowers as dependent variables to test how the cryptocurrency market affects overall borrowing on the P2P lending market. Our results are robust to using different measures to capture the overall interest in P2P lending.

We also examined how the cryptocurrency market affects individual loans and outcomes. We used loan amount to see whether a hot cryptocurrency market induces individuals to borrow more from P2P lending. We also constructed two measures to capture loan outcomes. The default rate is measured 540 days after the loan initiation. If a loan has any missed interest payment at least once during the period, the loan is considered a default. The borrower rate is a yearly interest rate. Lower default rates and interest rates represent better loan performance.

Our key independent variables of interest are cryptocurrency market trends. For that, we constructed and used three measures in our main analysis. We first extracted three daily transaction outcomes (closing price, market capitalization, and trading volume) of Bitcoin and Ethereum. ¹¹ We then took a weighted average of the metrics of each cryptocurrency (weighted by their loadings in the underlying principal components). ¹² To quantify the average price of cryptocurrency over a given period, we further used a 10‐day exponential moving average ¹³ (e.g., Brock et al., 1992).

To mitigate the omitted variable bias, we include an extensive set of control variables that may be associated with online P2P lending. Using Prosper.com, we added relevant listing characteristics and borrower credit risk variables known to affect listing and loan outcomes (e.g., Caldieraro et al., 2018; Morse, 2015; Wei & Lin, 2017). Finally, we added several macroeconomic variables. Using the borrower location, we included monthly unemployment rates for each state from the Bureau of Labor Statistics. We also added the monthly Zillow home value index measured for each state from Zillow.com. Both are used to control for economic conditions that may affect borrowing from P2P lending markets (e.g., Wei & Lin, 2017). We also included daily Dow Jones Industrial Average numbers, monthly Federal fund rates, foreign exchange rate (USD/CNY), and daily gold price, because they are arguably natural investment channels. Table C.1 in the Supporting Information summarizes our sample and descriptive statistics.

Model‐free evidence

Figure C.1 in the Supporting Information shows how the P2P lending and the cryptocurrency markets evolved during our study period. In 2017, we observed an exponential increase in the cryptocurrency market in terms of trading volume and market capitalization and a subsequent steep decrease. Although the trend is less dramatic than in the cryptocurrency market, the P2P lending market showed a similar pattern. P2P lending increased in total listing amount until early 2018. It has since decreased slightly. Similar trends in both markets indicate that they may positively relate to each other.

Figure C.2 in the Supporting Information confirms that high‐creditworthy borrowers are the main causes of increased online P2P market demand. Specifically, during our study period, high‐creditworthy borrowers (Prosper ratings of AA, A, or B) increased (decreased) their demands at bullish (bearish) markets, whereas low‐creditworthy borrowers (Prosper ratings of D, E, or HR) changed slightly or insignificantly in response to cryptocurrency market trends (Figure C.3 in the Supporting Information for details), supporting our prediction.

Loan demand increased more for loans that seem potentially intended for cryptocurrency market investment. Based on the broad borrower‐reported loan purpose, we classified loans as having financing (i.e., debt consolidation and personal loan) or nonfinancing purposes (e.g., home improvement, medical, and vacation). Figure C.4 in the Supporting Information shows that financing‐related listings increased the most. Considering that loans with nonfinancing purposes were created for uses unrelated to cryptocurrency investing, the result supports that P2P lending is a potential source of capital for investing in cryptocurrency.

Furthermore, in Figure C.5 in the Supporting Information, the increase is stronger among borrowers with occupations that are generally highly skilled, tech‐savvy, and more likely to be interested in the cryptocurrency market (Ennis, 2016; Stray, 2017), such as computer programmers, executives, and professionals. In contrast, borrowers in low‐skilled jobs such as administrative assistants, car dealers, and laborers are expected to have minimal knowledge about the new cryptocurrency market.

Our model‐free evidences show that increased P2P borrowing comes mostly from borrowers with good credit ratings and high‐skilled occupations who are seeking loans related to cryptocurrency. This implies that when the cryptocurrency market increased rapidly, the increase might have made P2P lending less inclusive.

How cryptocurrency market shocks affect borrowers' incentives to use P2P lending

To examine how cryptocurrency markets affect the role and effectiveness of P2P lending in enhancing financial inclusion, we first focused on the usage of P2P borrowing. That is, we examined whether cryptocurrency markets affect borrowers' loan amounts on the P2P lending market, and how the effect varies between financially advantaged and disadvantaged groups. Our main empirical strategy was to exploit several natural shocks on the cryptocurrency markets, along with propensity score matching (PSM). We first identified two positive and two negative exogenous shocks (Figure 1). We wanted to include both positive and negative shocks in boom‐and‐bust periods to dampen concerns about picking up general trends during the periods if we examined only positive (negative) shocks in boom (bust) periods.

We then compared transaction outcomes on P2P lending before and after the natural shocks within a short‐term period. As the shocks essentially affect all the listings in the P2P lending market, we consider the listing in the 5‐day after‐shock period as the treatment and consider the listing in the 5‐day before‐shock period as the control group. ¹⁴ We used a PSM method to find a control group that most resembled the treatment group and then estimated the treatment effects associated with the cryptocurrency market changes (e.g., Wei & Lin, 2017). This allows a direct comparison of transaction outcomes between two matched listings with similar characteristics (propensity score).

We used a logit regression to calculate the propensity score for the probability of being in the treatment group,

λ ( x i ) = Pr ( D i = 1 | X = x i ) , $$\begin{equation} \lambda ({\bf x}_i) = \text{Pr}(D_i =1 | {\bf X} = {\bf x}_i), \end{equation}$$

4

D i $D_i$

5

y i $y_i$

X i ${\bf X}_i$

We first observed two positive exogenous shocks on the cryptocurrency market. On May 2, 2018, Goldman Sachs became the first Wall Street bank to open a Bitcoin trading operation, in which they used Sachs money for various contracts linked to the price of Bitcoin (Popper, 2018). In addition, on October 2, 2017, Christine Lagarde, IMF CEO, announced that cryptocurrency could be the future (Hackett, 2017). Both shocks led to sudden and huge increases in the price of Bitcoin (Figure 1). We observed that unlike cryptocurrency markets, stock markets did not move much before and after the shocks, implying that the shocks are specific to the cryptocurrency rather than the general financial market (see Figure C.6 in the Supporting Information). Thus, we used natural shocks to help identify effects of the cryptocurrency market on the P2P lending market.

We then examined two negative exogenous shocks on the cryptocurrency market. On September 13, 2017, Jamie Dimon, JPMorgan CEO, warned that Bitcoin “is a fraud and will eventually blow up.” Bitcoin immediately slid by more than 10% in a single day (Kelly, 2016). On January 15, 2018, China clamped down on cryptocurrency and planned to block access to foreign cryptocurrency exchanges (Yang, 2018). Both shocks led to a huge decrease in the price of cryptocurrency (Figure 1 and Figure C.6 in the Supporting Information).

The matching estimates for the ATET of positive shocks in Panel A of Table 1 show that the two positive shocks overall led to an increase in loan amounts. Specifically, the shock related to Goldman Sach's crypto trading open (IMF head's positive mention) increased the loan by an average of 4.7% (3.3%). Negative shocks to the cryptocurrency markets were associated with smaller loan requests (Panel B of Table 1). China's cryptocurrency market regulation was associated with a 0.9% decrease in the listing amount within a short‐term 10‐day period. Dimon's statement had similarly negative effects. Together, the results show that positive cryptocurrency shocks led the same borrower to borrow more.

OPEN IN VIEWER

TABLE 1

Matching estimates of the cryptocurrency shock effects on online P2P lending markets

Panel A: Positive cryptocurrency shocks
	Listing amount (USD)
Dep. var:		Credit rating							Occupation
Borrower characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others
1(Goldman Sachs's Crypto	0.047***	929.424***	547.987*	907.935*	206.652	−130.531	−145.528	−81.915	248.734*	−2.755
Trading Open)	(3.33)	(2.79)	(1.64)	(1.84)	(0.92)	(−0.40)	(−0.54)	(−0.36)	(1.70)	(−0.01)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	8710	1501	1110	1991	2114	1106	659	229	1702	386

	Listing amount (USD)
Dep. var:		Credit rating							Occupation
Borrower characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others
1(IMF Head Positive	0.033*	100.359	662.570**	943.634*	−865.469**	−406.444**	−402.952**	−168.911***	407.266***	−591.557
Mention)	(1.78)	(0.35)	(1.71)	(1.25)	(−2.33)	(−2.24)	(−2.45)	(−3.50)	(5.31)	(−1.10)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	5560	476	568	731	957	1727	641	460	675	660

Panel B: Negative cryptocurrency shocks
	Listing amount (USD)
Dep. var:		Credit rating							Occupation
Borrower characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others
1(JP Morgan CEO Negative	−0.008*	−867.231	−346.235**	−918.088***	216.018***	273.569**	618.248**	355.993***	−299.707*	369.935
Mention)	(1.66)	(−1.08)	(−2.15)	(−2.93)	(5.90)	(2.12)	(2.10)	(6.08)	(−1.42)	(1.28)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	7590	558	1273	1517	1642	1838	538	224	1361	724

	Listing amount (USD)
Dep. var:		Credit rating							Occupation
Borrower characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others
1(China Negative Regulation)	−0.009*	−277.464*	−925.868**	−432.061	406.904**	498.454	524.333**	459.558***	−811.855***	562.809*
	(1.95)	(−1.39)	(−2.23)	(−1.18)	(2.24)	(1.41)	(2.28)	(2.80)	(−3.78)	(1.83)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	7712	900	1526	1942	1279	1547	387	131	361	393

Note: Robust t‐statistics are shown in parentheses. We follow Abadie and Imbens (2006) to calculate the standard errors of the matching estimates (AI Robust Standard Error). We used teffects psmatch2 syntax in STATA allowing for the estimate average treatment effect on the treated (i.e., ATET). We employed three‐nearest neighbors matching with replacement. Tech‐savvy occupations include investors, computer programmers, executives, professionals, and analysts who might be sensitive to the cryptocurrency market. Others occupations include administrative assistants, car dealers, and laborers who might not be sensitive to the cryptocurrency market, in general. Sensitivity analyses of treatment effects by Rosenbaum bounds indicate that our treatment effects are insensitive and less likely to be over/under‐estimated. *

p < 0.1 $p < 0.1$

; **

p < 0.05 $p < 0.05$

; ***

p < 0.01 $p < 0.01$

.

We then examined whether credit rating and occupation affected borrower responses to positive cryptocurrency shocks (Table 1). We found that high‐creditworthy borrowers largely drove increases in loan amounts. For instance, borrowers with a Prosper rating of AA increased the requested amount of loans by about US $930 after the positive shock from Goldman Sachs, which corresponds to 7.01% of the average loan amount during the 10‐day window. In contrast, low‐creditworthy borrowers generally responded negatively to positive shocks, although some were not statistically significant. This implies that growth in the cryptocurrency market induced high‐creditworthy borrowers to use P2P borrowing more aggressively and exploit the full potential of the market better. In addition, the increase was more positive for borrowers in Type 1 occupations–investors, computer programmers, executives, professionals, and analysts–who are likely to have more familiarity with and have better access to the cryptocurrency market. That is, high‐creditworthy and tech‐savvy borrowers increased (decreased) their demands when the cryptocurrency market was in a good (bad) situation. In contrast, low‐creditworthy and low‐skilled borrowers showed the opposite behaviors, although statistically weaker. ¹⁶ These findings support our Hypotheses 1 and 2. Together, these results suggest that as the cryptocurrency market grows, P2P lending becomes less inclusive because advantaged borrowers that likely have less difficulty accessing traditional credit make more use of P2P borrowing for that opportunity.

Impact of cryptocurrency market on the P2P lending market

In this section, we turn to loan access and examine whether and how the cryptocurrency market affects borrowers' access to P2P loans. We could not test whether and how the chances of loan acquisition by different borrowers change in response to cryptocurrency market conditions, because once an application is approved for listing, the success of obtaining a loan is almost 100%. Few applications are unsuccessful. We instead explored whether the cryptocurrency market affects individuals who are willing to borrow and, if so, how much they are prepared to borrow. To investigate the effects of the cryptocurrency market on the incentive of borrowers to use P2P lending, we estimated the impact of the former on the latter at the market level by running ordinary least squares (OLS) regressions on daily aggregated data. Note that although OLS regressions provide initial base results, this estimation approach may be confronted with identification issues. For instance, we could not completely rule out whether some omitted variables (e.g., updated market news) may have affected both the cryptocurrency and P2P lending markets. To alleviate these concerns, we incorporated a large number of control variables into Table C.1 in the Supporting Information, including loan‐ and borrower‐related characteristics and macroeconomic variables. We also used instrumental variables to minimize the endogeneity issue driven by developments in the cryptocurrency market. Our OLS regression models for the overall borrowing of P2P lending are delineated as follows:

y t = b t + α 1 Cryptocurrency Index t − 7 + α 2 ′ X t + e t , $$\begin{equation} y_t =b_t + \alpha _1\ \text{Cryptocurrency Index}_{t-7} + \alpha _2^{^{\prime }} {\bf X}_t +e_t, \end{equation}$$

6

y t $y_t$

X t ${\bf X}_t$

OPEN IN VIEWER

TABLE 2

Daily‐aggregated associations between cryptocurrency markets (Bitcoin and Ethereum) and online P2P lending markets

Model	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Borrower		Credit rating		Occupation		Credit rating x Occupation
Characteristics	Overall	High‐credit	Low‐credit	Tech‐savvy	Non‐tech‐savvy	High‐credit and tech‐savvy	High‐credit and non‐tech‐savvy	Low‐credit and tech‐savvy	Low‐credit and non‐tech‐savvy
Panel A
Dep. var:	ln(Daily Listing Amount)
ln(Market Capitalizations) t − 7 $_{t-7}$	0.123***	0.182***	−0.038	0.071***	−0.92*	0.140***	0.102***	−0.112***	−0.247
	(2.60)	(3.48)	(−0.79)	(2.76)	(−1.88)	(4.44)	(3.62)	(−3.11)	(−6.66)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	789	789	789	789	789	789	789	789	789
Panel B
Dep. var:	ln(Daily Listing Number)
ln(Market Capitalizations) t − 7 $_{t-7}$	0.131***	0.192***	−0.063	0.065**	−0.82*	0.151***	0.174***	−0.069**	−0.115***
	(2.80)	(3.73)	(−1.33)	(2.56)	(−1.80)	(4.92)	(6.64)	(−2.21)	(−3.69)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	789	789	789	789	789	789	789	789	789
Panel C
Dep. var:	ln(Daily Unique Borrower Number)
ln(Market Capitalizations) t − 7 $_{t-7}$	0.126***	0.178***	−0.064	0.060**	−0.076*	0.131***	0.116***	−0.103***	−0.151***
	(2.73)	(3.46)	(−1.36)	(2.39)	(−1.66)	(4.40)	(4.93)	(−3.46)	(−5.75)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	789	789	789	789	789	789	789	789	789

Note: Robust t‐statistics are shown in parentheses. All models are based on daily‐aggregated borrowers' activities in the P2P lending market. In our main estimation, we use each of the three measures (i.e., market capitalization, trading volume, and exponential moving average) for the cryptocurrency index one by one due to the high correlation among them); this yielded similar results and we reported with market capitalization for brevity. To allow for the possibility of delayed impact of cryptocurrency markets on online P2P markets, we re‐estimated our model using 1‐day or 1‐week lagged values of our key independent variables—cryptocurrency market‐related variables. The results are similar to the results based on contemporaneous values. The current table allows a 1‐week lag between two markets. We estimate our modes in a hierarchical manner (i.e., estimating our model with control variables only then including independent variables on control variables); this yielded similar results. High‐credit group includes Prosper ratings AA, A, and B, and low‐credit group includes Prosper ratings E and HR. Tech‐savvy occupations include investors, computer programmers, executives, professionals, and analysts who might be sensitive to the cryptocurrency market. Non‐tech‐savvy occupations include administrative assistants, car dealers, and laborers who might not be sensitive to the cryptocurrency market, in general. *

p < 0.1 $p < 0.1$

; **

p < 0.05 $p < 0.05$

; ***

p < 0.01 $p < 0.01$

.

In Table 2, the market growth of cryptocurrencies is significantly positively associated with the daily listing amount (Model 1 of Panel A) and daily number of listings (Model 1 of Panel B). This suggests that when the cryptocurrency market is hot, borrowing on the P2P lending market increases in the following week. Interestingly, we also observed an increase in the daily unique number of borrowers (Model 1 of Panel C), implying that hot cryptocurrency markets attract new borrowers to the P2P lending market. Economically, for a 1% increase in the market capitalization of the cryptocurrency market, a daily listing amount on the P2P lending market increases on average by 0.12%. We observed similar economic effects on the daily listing number and the daily unique number of borrowers.

To investigate how the cryptocurrency market affects the distribution of P2P borrowers, we broke down borrowers into two groups by credit rating or occupation. First, in Models 2 and 3, we broke down borrowers into high‐credit and low‐credit borrowers. Our results show that hot cryptocurrency markets are more positively associated with borrowing of high‐credit borrowers than that of low‐credit. Actually, we did not observe a significant association for low‐credit borrowers. This indicates that as the cryptocurrency market becomes hot, the P2P lending market has proportionally more borrowing from high‐credit borrowers. When we classified borrowers by their occupation, we further observed a positive association only in tech‐savvy borrowers. Together, the results show that borrowers responded differently to cryptocurrency market developments. High‐credit and tech‐savvy borrowers were more responsive to hot cryptocurrency markets. Given that they have been traditionally in a better condition in terms of access to credit, traditionally advantaged borrowers are more likely to issue P2P loans to seize new investment opportunities.

Importantly, our results show that a hot cryptocurrency market is more positively associated with the daily unique borrower number of high‐credit (

α = 0.178 $\alpha = 0.178$

,

p < 0.01 $p < 0.01$

in Model 2 of Panel C) and tech‐savvy borrowers (

α = 0.06 $\alpha = 0.06$

,

p < 0.05 $p < 0.05$

in Model 4 of Panel C) than with that of low‐credit and non‐tech‐savvy borrowers. This pattern reflects that the probability of participation by high‐credit and tech‐savvy borrowers can increase in response to a hot cryptocurrency market. This result supports H1 and H2. In addition, we found that the moderating effect of creditworthiness on the relationship between cryptocurrency trends and demand from P2P borrowers is greater than that of technological savviness. These differences were confirmed when we divided borrowers into four groups on the grounds of a combination of credit rating and occupation. Models 6–9 of Table 2 indicate that the marginal effects of creditworthiness are greater than those of technological savviness. The marginal effects are positive only when borrowers have high creditworthiness. Also, the level of technological savviness does not change the marginal effects of technological savviness significantly. Together, these findings support H3.

To dampen endogeneity concerns, we implemented the instrumental variable approach (IV approach, also known as a two‐stage least squares estimation). As instruments for the cryptocurrency market index, we used weather‐related variables employed in prior studies as instruments for trading assets (Hirshleifer & Shumway, 2003; Loughran & Schultz, 2004). Although weather affects stock market returns and trading volumes, it is arguably exogenous to investment behaviors. ¹⁷ We measured our instruments and conducted the analyses at the state–month level based on the smallest granularity of the weather data and the P2P lending data. Specifically, we used (1) average temperature by state and month and (2) precipitation by state and month as IVs for both the market index of cryptocurrencies (i.e., market capitalizations) and the stock market index (i.e., S&P500 Index; SNP500). We obtained monthly statewide average‐temperature and precipitation data from the National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Information.

We then constructed a panel data set containing 1168 state–month observations with monthly aggregated data of two FinTech markets. Table 3 reports the results of the IV estimation. In the first stage, models show that average temperature (precipitation) is significantly positively (negatively) associated with both the market capitalizations of Bitcoin and Ethereum cryptocurrencies and the stock market (SNP500), corroborating that sunny and warm weather affects investor behavior (Hirshleifer & Shumway, 2003; Kaustia & Rantapuska, 2016; Loughran & Schultz, 2004). In the second stage, we regressed the overall borrowing of P2P lending (i.e., listing amount) on the fitted value of market capitalization of Bitcoin and Ethereum. Overall, the results of the IV estimation are qualitatively similar to the main results. In other words, a greater market capitalization of the cryptocurrency market is positively associated with the daily P2P borrowing. The association is observed mainly in high‐credit or tech‐savvy borrowers.

OPEN IN VIEWER

TABLE 3

Two‐stage least squares results: Considering endogeneity of Cryptocurrency Market Index

Model	(1)	(2)	(3)	(4)	(5)	(6)	(7)
		Credit rating		Occupation
Borrower characteristics	Overall					Overall	Overall
		High‐credit	Low‐credit	Tech‐savvy	Others
Panel A: Second‐stage regression
Dep. var:	ln(Monthly Listing Amount at Each State)					1(Defaulted)	Borrower interest rate
ln(Market Capitalizations	1.281**	0.690***	−3.015***	1.625***	−4.139	−0.054***	−1.89e‐14***
of Cryptocurrencies)	(2.56)	(3.05)	(−2.85)	(2.68)	(−1.30)	(−2.08)	(−3.75)
ln(Stock Market Index	−8.022	−3.918	41.684***	−14.787*	45.226	0.446	2.05e‐13***
, SNP500)	(−1.48)	(−1.42)	(3.36)	(−1.75)	(1.54)	(1.19)	(2.81)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	1168	1168	1168	1168	1168	1168	1168
Panel B: First‐stage regression
Dep. var:	ln(Market Capitalizations)
Average temperature (F)	0.008***	0.007***	−0.001	0.007***	0.003***	0.011***	0.011***
	(8.73)	(7.35)	(−0.07)	(6.77)	(2.83)	(11.04)	(11.04)
Precipitation (inch)	−0.022***	−0.030**	−0.044***	−0.033***	−0.042***	−0.020**	−0.020**
	(−2.83)	(−3.60)	(−4.69)	(−3.30)	(−4.20)	(−2.48)	(−2.48)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	1168	1168	1168	1168	1168	1168	1168
Panel C: First‐stage regression
Dep. var:	ln(Stock Market Index)
Average temperature (F)	0.0007***	0.0004***	0.0003***	0.001***	0.001***	0.001***	0.001***
	(9.51)	(3.86)	(3.27)	(6.06)	(5.45)	(9.51)	(9.51)
Precipitation (inch)	−0.003***	−0.004***	−0.003***	−0.003***	−0.003***	−0.003***	−0.003***
	(−5.57)	(−6.07)	(−5.03)	(−5.15)	(−5.42)	(−5.57)	(−5.57)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	1168	1168	1168	1168	1168	1168	1168

Note: Robust t‐statistics are shown in parentheses. The table reports results from 2SLS using two instrument variables (IVs) about statewide weather. Specifically, we use (1) average temperature by state and month and (2) precipitation by state and month as IVs for both the market index of cryptocurrencies (i.e., market capitalizations, trading volume, and exponential moving averages) and the market index of stock (i.e., S&P500 Index; SNP500). Models are based on monthly statewide aggregated borrowers' activities in the P2P lending market and market capitalizations of cryptocurrencies (Bitcoin and Ethereum). We also estimate our models with other market indexes of cryptocurrency (e.g., trading volume and exponential moving averages) and stock market indexes (e.g., Dow Jones Industrial Average or Nasdaq); this yielded similar results but are not reported for brevity. High‐credit group includes Prosper ratings AA, A, and B, and low‐credit group includes Prosper ratings E and HR. Tech‐savvy occupations include investors, computer programmers, executives, professionals, and analysts who might be sensitive to the cryptocurrency market. Others occupations include administrative assistants, car dealers, and laborers who might not be sensitive to the cryptocurrency market, in general. *

p < 0.1 $p < 0.1$

; **

p < 0.05 $p < 0.05$

; ***

p < 0.01 $p < 0.01$

.

Although we saw an increase in overall borrowing in response to hot cryptocurrency markets, we do not know whether loan outcomes would be better or worse. To address this, we conducted OLS or logistic regressions at the loan level:

z i = α i + β 1 Cryptocurrency Index i + β 2 ′ X i + φ i , $$\begin{equation} z_i = \alpha _i + \beta _1\text{ Cryptocurrency\nobreakspace Index}_i+ \beta _2^{^{\prime }} {\bf X}_i + \varphi _i, \end{equation}$$

7

where

z i $z_i$

are the two transaction outcomes of a loan‐the indicator of default for a funded loan and borrower rates.

X i $X_i$

denotes the control variables, as shown in Table C.1 in the Supporting Information, aggregated at the state–month level.

In Models 6 and 7 of Table 3, we find that better cryptocurrency market conditions are associated with lower default rates and lower interest rates on the P2P lending market. In terms of economic significance, on average, a 1% increase in market capitalization of the cryptocurrency market is associated with a 0.003% decrease in borrower interest rates and a 3.03% decrease in default rates (in terms of marginal effects,

z = − 31.80 $z = -31.80$

,

p < 0.01 $p < 0.01$

), indicating that as the cryptocurrency market became hot, this led to lower default rates and borrower interest rates at the market level. This result should not be surprising because hot cryptocurrency markets attracted proportionally more high‐credit borrowers.

Overall, the findings from the two identification strategies suggest that hot cryptocurrency markets motivates creditworthy and tech‐savvy borrowers to borrow more actively and large amounts, thus skewing the market toward financially advantaged borrowers. This further implies that financially disadvantaged borrowers have difficulty accessing both FinTech markets, thereby diminishing financial inclusion, at least in the short run.

Robustness checks

We conducted a series of robustness checks to further ensure the validity of our results. First, we used measures to capture a broad scope of cryptocurrency markets. Our baseline measures of cryptocurrency markets based on Bitcoin and Ethereum account for about 60% of the total cryptocurrency market capitalization. To increase comprehensiveness, we observed the market capitalization of top 10 cryptocurrencies: Bitcoin, Ethereum, Ripple, Bitcoin Cash, EOS, Litecoin, Cardano, TRON, Dash, and Monero, which covered 83% of the total market capitalization as of June 27, 2018. We constructed three measures of cryptocurrency market trends and re‐estimated Equation (4) at the listing level. Table 4 indicates that the results based on the alternative measures of top 10 cryptocurrencies (Panel B) are qualitatively similar to the results with Bitcoin and Ethereum (Panel A).

OPEN IN VIEWER

TABLE 4

Listing‐level associations between cryptocurrency markets and transaction outcomes of online P2P lending markets

Panel A: Bitcoin and Ethereum markets
Model	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	ln(Listing Amount)
Dep. var:		Credit rating							Occupation		1(Defaulted)	Borrower interest rate
Borrower characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others	Overall	Overall
ln(Market Capitalizations) t − 7 $_{t-7}$	0.010***	0.038***	0.008**	0.003	0.002	−0.012***	0.002	−0.011*	0.017***	0.003	−0.103***	−0.003***
	(7.20)	(9.14)	(2.44)	(0.95)	(0.58)	(−2.90)	(0.35)	(−1.86)	(5.59)	(0.53)	(−31.81)	(−233.12)
ln(Trading Volume) t − 7 $_{t-7}$	0.008***	0.024***	0.013***	0.073***	0.004**	−0.002	−0.001	−0.007*	0.013***	0.001	−0.074***	−0.002***
	(8.14)	(7.62)	(5.10)	(3.32)	(2.27)	(−0.57)	(−0.05)	(−1.60)	(5.96)	(0.14)	(−33.19)	(−165.61)
ln(Exponential Moving Averages) t − 7 $_{t-7}$	0.010***	0.044***	0.007**	0.003	0.002	−0.015***	−0.002	−0.007	0.018***	0.002	−0.116***	−0.003***
	(6.86)	(9.68)	(1.93)	(0.79)	(0.58)	(−3.25)	(−0.36)	(−1.14)	(5.42)	(0.26)	(−32.94)	(−224.78)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	537,878	48,031	81,858	118,992	141,294	78,952	48,956	19,601	103,376	24,021	444,313	444,313

Panel B:Top 10 cryptocurrencies markets
Model	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	ln(Listing Amount)
Dep. var:		Credit Rating							Occupation		1(Defaulted)	Borrower interest rate
Borrower Characteristics	Overall	AA	A	B	C	D	E	HR	Tech‐savvy	Others	Overall	Overall
ln(Market Capitalizations) t − 7 $_{t-7}$	0.009***	0.034***	0.008**	0.003	0.002	−0.010***	0.002	−0.008	0.016***	0.003	−0.097***	−0.003***
	(7.42)	(8.89)	(2.58)	(1.14)	(0.72)	(−2.64)	(0.66)	(−1.53)	(5.71)	(0.56)	(−32.52)	(−229.55)
ln(Trading Volume) t − 7 $_{t-7}$	0.007***	0.023***	0.012***	0.007***	0.004**	−0.002	0.001	−0.006*	0.012***	0.001	−0.071***	−0.002***
	(8.09)	(7.56)	(5.41)	(3.26)	(1.96)	(−0.76)	(0.24)	(−1.65)	(6.09)	(0.02)	(−33.42)	(−167.82)
ln(Exponential Moving Averages) t − 7 $_{t-7}$	0.013***	0.045***	0.009**	0.003	0.002	−0.015***	0.006	−0.010	0.020***	0.006	−0.122***	−0.004***
	(8.01)	(9.25)	(2.32)	(0.77)	(0.58)	(−2.82)	(0.99)	(−1.34)	(5.70)	(0.86)	(−31.32)	(−226.42)
Covariates	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Number of observations	537,878	48,031	81,858	118,992	141,294	78,952	48,956	19,601	103,376	24,021	444,313	444,313

Note: Robust t‐statistics are shown in parentheses. Top 10 cryptocurrencies are Bitcoin, Ethereum, Ripple, Bitcoin Cash, EOS, Litecoin, Cardano, TRON, Dash, and Monero, which covered 83% of the total market capitalization as of June 27, 2018. In our main estimation, we use each of the three measures for the cryptocurrency index one by one due to the high correlation among them. All models are based on either listing‐ or loan‐level analysis. For comparisons of default rate, we estimated logit models. To allow for the possibility of delayed impact of cryptocurrency markets on online P2P markets, we re‐estimated our model using 1‐day or 1‐week lagged values of our key independent variables—cryptocurrency market‐related variables. The results are similar to the results based on contemporaneous values. The current table allows a 1‐week lag between two markets. We estimate our modes in a hierarchical manner (i.e., estimating our model with control variables only then including independent variables on control variables); this yielded similar results. Tech‐savvy occupations include investors, computer programmers, executives, professionals, and analysts who might be sensitive to the cryptocurrency market. Others occupations include administrative assistants, car dealers, and laborers who might not be sensitive to the cryptocurrency market, in general. *

p < 0.1 $p < 0.1$

; **

p < 0.05 $p < 0.05$

; ***

p < 0.01 $p < 0.01$

.

To further address potential endogeneity issues, we employed a vector autoregressive model with exogenous variables (VARX) used previously (e.g., Kök et al., 2020) to examine dynamic interactions between cryptocurrency markets and online P2P markets. To create time series data, we aggregated daily listings in P2P lending by Prosper credit ratings. The specific equation we estimate is given as follows:

8

where

y r , t $y_{r,t}$

is a two‐element column vector for Prosper rating r (AA‐A‐B‐C‐D‐E‐HR) at day t, Φs indicate 4 X 4 matrices of slope coefficients for endogenous variables, p represents the number of lags,

X r , t ${\bf X}_{r,t}$

denotes a column vector of covariates,

θ t $\theta _t$

refers to a column vector of time dummies,

μ t $\mu _t$

stands for a column vector of unobserved individual effects, and

ε i , t $\epsilon _{i,t}$

signifies a two‐element vector of errors (error terms are serially uncorrelated when a sufficient number of lags p is used).

The estimation results of our VARX model in Table C.3 in the Supporting Information indicate that the cryptocurrency market has a significant positive effect on the demand of high‐creditworthy borrowers ($776.298,

p < 0.001 $p < 0.001$

), whereas the coefficient estimate for low‐creditworthy borrowers is not significant ($340.82,

p > 0.010 $p > 0.010$

), after controlling for lagged endogenous and exogenous variables. Furthermore, the coefficients on the listing amount are insignificant (

p s > 0.10 $p\text{s} > 0.10$

). The VARX results are consistent with our main findings that the cryptocurrency market positively affects P2P lending, mostly through high‐quality borrowers.

Finally, to address concerns that Prosper may fail to represent the online P2P lending market, we replicated our estimations using Lending Club data. Because this data set provides only monthly listing and loan data, we matched monthly aggregated data with the monthly cryptocurrency data set. Lending Club data yielded similar results (Table C.4 in the Supporting Information).

Additional analyses

We utilize data on individuals' investments in multiple assets, such as P2P loans, cryptocurrency, and others. These analyses are provided in Appendix D in the Supporting Information. Our archival data analysis shows that the cryptocurrency market affects P2P lending, but we surveyed P2P lending participants to gain a better understanding of the relationship between the two markets. A detailed description of the survey is provided in Appendix E in the Supporting Information. Finally, additional analyses on possible alternative explanations are provided in Appendix F in the Supporting Information.

DISCUSSION AND CONCLUSION

Key findings

Online P2P lending and cryptocurrency markets have emerged as two of the most active and popular FinTech platforms that capitalize on innovative technologies and efficient business processes to complement established finance. In particular, P2P lending has contributed to the growth of financial inclusion by which many individuals and firms, who are excluded by formal financial institutions, can gain access to an array of financial services (i.e., loans). As financial resources move from one market to another, the rapid growth of cryptocurrency markets may affect the operation of P2P lending and its capacity to exercise financial inclusion.

Controversies and polarized viewpoints continue in regard to the value proposition and future outlook of cryptocurrencies as viable FinTech initiatives. Before the advent of cryptocurrency markets, online P2P lending received considerable practical and academic attention as a new attractive platform for alternative financing, particularly for borrowers who are excluded by conventional financial institutions. Numerous empirical explorations abound in respect to the economics of P2P lending and cryptocurrency platforms, but most were conducted in isolation. The inter‐market relationships and cross‐platform dynamics of these markets should be scrutinized, as they affect the welfare of many investors, especially those who are financially disadvantaged.

Our results indicated that as an upsurge occurs in cryptocurrency markets, P2P lending markets acquire more borrowers and a greater volume of loan requests. Loans initiated in response to bullish cryptocurrency markets are associated with reduced default rates and interest rates, as shown by increases in the number of borrowing activities among high‐creditworthy borrowers and decreases in such transactions among low‐creditworthy borrowers. These effects are manifested primarily among borrowers with tech‐savvy occupations, suggesting that rising cryptocurrency markets drive high‐creditworthy and tech‐savvy borrowers toward P2P lending as an avenue for investment in cryptocurrency markets. Therefore, cryptocurrency markets may, in the long run, undermine the democratization of access that P2P lending platforms foster by serving superfluous consumers instead of imperative consumers.

Theoretical and managerial implications

Our study contributes to the literature in several ways. Empirical works on P2P lending examined the behaviors of participants and the market mechanism within P2P lending markets (e.g., Lin & Viswanathan, 2016; Liu et al., 2015; Wei & Lin, 2017). We extended these endeavors by demonstrating that a spillover effect occurs between the two dominant FinTech platforms and that such cross‐dynamics affect the welfare of many investors, including those who are constrained in accessing traditional financial services. A recent study found that the gig economy (i.e., Uber) and crowdfunding (i.e., Kickstarter) platforms have a negative relationship and might compete against each other (Burtch et al., 2018). In contrast, by focusing on cross‐platform‐related issues, our study provided evidence that a positive externality exists between P2P lending and cryptocurrency markets. Second, we broadened the scope of inquiry to cover financial inclusion via technology. A few studies have suggested that P2P lending and blockchain‐based applications potentially enhance financial inclusion (Arifin et al., 2018; Maskara et al., 2021; Schuetz & Venkatesh, 2020). We, however, showed that newer FinTech platforms (i.e., cryptocurrency) may bring significant change to the financial inclusion cultivated in an incumbent FinTech platform (i.e., P2P lending) by increasing inequality in access to these avenues.

In terms of practice, our study presents important implications through its revelation that creditworthy and tech‐savvy investors use P2P lending to invest in cryptocurrency and suggestions on how FinTech strategies should be formulated to protect the wealth of investors and sustain focus on financial inclusion. For instance, in addition to measuring their financial risk (Wang et al., 2015), in the short term, online P2P lending managers can extract the most value by tracking and managing high‐creditworthy and tech‐savvy borrowers. Nevertheless, despite the possible short‐term gains derived from the influx of additional loan requests from high‐quality borrowers, the increased inequality in access to FinTech platforms originating from service to certain groups may hurt the long‐term sustainability of P2P lending. This double‐edged sword of FinTech platforms should be considered with care with respect to people with low credit scores and technology literacy to ensure sustainable operations. For example, policy makers can provide appropriate technological education to individuals and firms with low credit ratings and technological knowledge to prevent the potential transition from technological illiteracy to financial exclusion. From an individual perspective, people should embrace fast‐changing financial technologies and actively look for possible opportunities from FinTech platforms under a proper understanding of these modern technologies.

Conclusion

This research is not without limitations. We identified several directions that can be pursued in future studies. Although our investigation focused on the effects of cryptocurrency on debt‐crowdfunding platforms, future work should also illuminate how cryptocurrency markets influence the transaction outcomes of other crowdfunding platforms, such as reward‐based (e.g., Burtch et al., 2021) and donation‐based crowdfunding (e.g., Ge et al., 2016; Weinmann & Mishra, 2019) or the stock market (e.g., Tafti et al., 2016). Further, given that there is heavy manipulation in the cryptocurrency market (e.g., Griffin & Shams, 2020), finding a way to isolate authentic transactions might be an interesting pursuit. Next, we obtained and analyzed P2P lending data from Prosper.com, which is considered the largest channel for crowdfunding in the United States. However, our sample company does not represent the entire population of P2P lending firms, and thus other researchers can extend the generalizability of our findings by including additional lending platforms in analyses. In addition, the effects of the interdependencies between the two FinTech platforms on financial inclusion may vary depending on the economic conditions of countries. Countries establish different regulations for governing P2P lending and cryptocurrencies in a manner that best suits their financial and economic conditions. Policy makers and government agencies may interpret and use our findings to conceive of the best way to protect investors and preserve their welfare, particularly those who do not have equal opportunities to gain financial access. Finally, while our study used weather variables as instrumental variables and exploits several exogenous shocks to the cryptomarket, teasing out alternative explanations is fundamentally difficult. Although we believe that the whole set of tests we conducted can rule out many possible alternatives, future studies should also investigate alternatives.

Notwithstanding these caveats, our study lays an important foundation for future research related to the role of emerging FinTech platforms in advancing efforts by a disadvantaged segment of society to obtain equal opportunities and access to financial services at affordable costs. Our work casts light on the inter‐dynamics of FinTech markets and implications revolving around social welfare, but it also raises a number of new issues. For instance, to what extent should regulatory frameworks be enforced to protect the interests of the disadvantaged? Likewise, how should government agencies discourage the use of P2P lending as a source channel for speculative investment in alternative markets? It may be necessary that the goals of FinTech and crowdfunding should evolve as competition intensifies and regulations strengthen, but its social value and responsibility over those in need of equal financial access and opportunities must also be sustained and expanded.

Footnotes

1

https://globalfindex.worldbank.org/

2

Table A.1 in the Supporting Information provides the literature review on the financial inclusions in terms of accessibility and usage of financial services.

3

https://www.alliedmarketresearch.com/peer‐to‐peer‐lending‐market

4

P2P lending provides unique borrower benefits. Typical P2P lending markets leverage various big data and machine learning technologies and offer attractive interest rates, especially to borrowers who lack credit histories or have negative histories that would make it difficult for them to borrow at reasonable rates. P2P lending significantly hastens loan processing: Borrowers can acquire loans within a few minutes with a few clicks on a mobile app. We calculated that the average duration for loan success on Prosper in 2018 was only 3.8 h.

5

Examining the effects of cryptocurrency markets on P2P lending markets is relevant for at least two reasons. First, from the perspective of investors, cryptocurrencies and P2P loans are alternative assets that are driven by similar FinTech technologies, such as decentralized networks. Thus, the two markets may draw similar users. Second, cryptocurrency markets are significantly more volatile than traditional investment assets. They operate continuously while stock markets are closed on holidays. Thus, short‐term traders who want to invest in cryptocurrencies must react quickly. If they want to use financial leverage, they must borrow in a short period of time. Consequently, online P2P lending would be better than conventional lending channels as it allows borrowers to rapidly acquire loans anytime, anywhere, with a few clicks. Thus, we believe that the two markets are closely linked.

6

In the Lend Academy Forum (forum.lendacademy.com), an online community where P2P lending participants share their investment and borrowing experiences, many users consider investment in cryptocurrencies to be an alternative to P2P lending.

7

Of particular interest to this study is the interdependence between a financial market (e.g., the cryptocurrency market) and a credit market (e.g., the online P2P lending market). Therefore, we conducted a thorough literature review of this stream of literature and highlight the incremental contribution of this study. We thank the senior editor for providing this insight. Details are provided in Appendix A in the Supporting Information.

8

Table A.3 in the Supporting Information summarizes this stream of research.

9

Uniform distributions are assumed for traceability purposes. Ceteris paribus, relaxation of the distribution assumption, does not change our results.

10

Prosper does not provide lenders' transaction history, identity, investment amount, and actual returns for 2017–2019. To derive additional insights into lenders, we utilize individual‐level investment data on both P2P lending and cryptocurrency (Subsection 4.8 provides details).

11

We obtained daily trading information for the remaining top 10 cryptocurrencies (Ripple, Bitcoin Cash, EOS, Litecoin, Cardano, TRON, Dash, and Monero). Using this measure yields similar results (see Panel B in Table 4).

12

Using only Bitcoin yielded similar results but accounted for less than 50% of the total market capitalization of cryptocracies. Therefore, we focused on the two major cryptocurrencies accounting for 68% of the total market capitalization: Bitcoin and Ethereum.

13

Moving averages are called “moving” because, as the price moves, new data are added into the calculation, therefore changing the average. They help traders isolate the trend or lack of trend in a security or market, and can also signal when a trend is reversing. Two of the most common types are simple and exponential. A 10‐day simple moving average (SMA) adds the closing prices of the last 10 days divided by 10, whereas the exponential moving average (EMA) focuses more on recent prices than on a long series of data points, as the SMA required. This difference makes the EMA react more rapidly to price changes and the SMA react more slowly. Many short‐term traders prefer EMAs because they want to be alerted as soon as the price moves. We present our main results with EMAs, but using SMAs yielded similar results. We also employed the Relative Strength Index (RSI) as the volatility measure of cryptocurrency markets. Results in Table C.6 in the Supporting Information corroborate our main results. We thank an anonymous reviewer for providing this insight.

14

In addition, we tried 7‐day or 10‐day windows before and after shocks. The results from both periods were qualitatively similar to our 5‐day window results.

15

Using coarsened exact matching (CEM) yielded similar results. For all PSM estimates in , we conducted covariate imbalance testing and found no major differences between treated and control groups in terms of their means (e.g., Table C.5 and Figure C.8 in the Supporting Information). Details are available on request. We thank an anonymous reviewer for providing this insight.

16

To explore the possibility that high‐ versus low‐creditworthy borrowers show different sensitivities to cryptocurrency booms and busts (i.e., 2017 vs. 2018) but learn to leverage P2P lending when the bull market is over, we conducted univariate comparisons of the main key variables including macroenvironmental factors. Table C.2 in Appendix C of the Supporting Information shows results for difference‐of‐means tests showing that high‐creditworthy borrowers kept their demands constant with boom and bust markets, but low‐creditworthy borrowers increased their demands in bust markets. The number of unique borrowers was stable over boom and bust markets regardless of credit levels (Figure C.7 in the Supporting Information), and low‐creditworthy borrowers showed smaller demands than high‐creditworthy borrowers. Therefore, we conjecture that low‐creditworthy borrowers did not experience learning effects that would significantly affect our main results.

17

For instance, when it is sunny, people are in a good mood and tend to evaluate future prospects more optimistically. This then makes people consider positive information more salient and psychologically available, and boost stocks. However, P2P borrowing is very different from stock investing. When a borrower wants to issue a P2P loan, the interest rate is already determined by the P2P lending platform. Moreover, the lending platform does not change its interest rates frequently. Thus, the payoff for a borrower is rather expected. Thus, we doubt that any psychological mood has a significant role in P2P borrowing. In contrast, returns of stocks and cryptocurrencies change frequently even within a day. Thus, psychological mood can be important for investors in such assets. If people use P2P borrowing to invest in cryptocurrencies, weather can affect P2P borrowing, but only via cryptocurrency prices. As long as the weather effect is indirect, we believe that our instruments are valid. As long as the weather effect is indirect, we believe that our instruments are valid. In other words, we acknowledge that our weather instruments are only valid based on the assumption that they affect P2P borrowing only through investment in stock markets and cryptocurrencies. We thank anonymous reviewers for providing this insight.

References

1.

Abadie A. Imbens G. W. (2006). Large sample properties of matching estimators for average treatment effects. Econometrica, 74(1), 235–267.

+ Show References

2.

Alexander K. (2017). Financial inclusion: The role of Fintech and digital financial services. In UNCTAD Multi‐year Expert Meeting on TRADE, SERVICES AND DEVELOPMENT , Geneva.

Go to Reference

3.

Alstyne M. V. (2014). Why Bitcoin has value. Communications of the ACM, 57(5), 30–32.

Go to Reference

4.

Arifin A. Arshad N. Muneeza A. (2018). The application of blockchain technology in crowdfunding: Towards financial inclusion via technology. International Journal of Management and Applied Research, 5(2), 82–98.

+ Show References

5.

Babich V. Hilary G. (2020). Distributed ledgers and operations: What operations management researchers should know about blockchain technology. Manufacturing Service Operations Management, 22(2), 223–240.

+ Show References

6.

Balyuk T. (2022). FinTech lending and bank credit access for consumers. Management Science , Articles in Advance 22 Feb 2022.

Go to Reference

7.

Bartlett R. Morse A. Stanton R. Wallace N. (2022). Consumer‐lending discrimination in the Fintech era. Journal of Financial Economics, 143(1), 30–56.

Go to Reference

8.

Basu S. Garimella A. Han W. (2019). Impact of artificial intelligence on human decision making on ico platforms. In International Conference on Information Systems , Munich, Germany.

Go to Reference

9.

Berg T. Burg V. Gombovi A. Puri M. (2020). On the rise of Fintechs: Credit scoring using digital footprints. The Review of Financial Studies, 33(7), 2845–2897.

Go to Reference

10.

Bernstein F. DeCroix G. Keskin N. (2021). Competition between two‐sided platforms under demand and supply congestion effects. Manufacturing Service Operations Management, 23(5), 1043–1061.

+ Show References

11.

Bianchetti M. Ricci C. Scaringi M. (2018). Are cryptocurrencies real financial bubbles? Evidence from quantitative analyses. Available at SSRN. https://ssrn.com/abstract=3092427

Go to Reference

12.

Brock W. Lakonishok J. LeBaron B. (1992). Simple technical trading rules and the stochastic properties of stock returns. Journal of Finance, 47(5), 1731–1764.

Go to Reference

13.

Buchak G. Matvos G. Piskorski T. Seru A. (2018). Fintech, regulatory arbitrage, and the rise of shadow banks. Journal of Financial Economics, 130(3), 453–483.

+ Show References

14.

Burtch G. Carnahan S. Greenwood B. N. (2018). Can you gig it? An empirical examination of the gig economy and entrepreneurial activity. Management Science, 64(12), 5497–5520.

+ Show References

15.

Burtch G. Gupta D. Martin P. (2021). Referral timing and fundraising success in crowdfunding. Manufacturing Service Operations Management, 23(3), 676–694.

Go to Reference

16.

Caldieraro F. Zhang J. Z. Cunha M. C. Jr Shulman J. D. (2018). Strategic information transmission in peer‐to‐peer lending markets. Journal of Marketing, 82(2), 42–63.

+ Show References

17.

Chen Y. Bellavitis C. (2020). Blockchain disruption and decentralized finance: The rise of decentralized business models. Journal of Business Venturing Insights, 13, e00151.

Go to Reference

18.

Chen Y. Dai T. Korpeoglu C. Körpeolu E. Sahin O. Tang C. Xiao S. (2020). Innovative online platforms: Research opportunities. Manufacturing Service Operations Management, 22(3), 430–445.

Go to Reference

19.

Du N. Li L. Lu T. Lu X. (2020). Prosocial compliance in P2P lending: A natural field experiment. Management Science, 66(1), 315–333.

Go to Reference

20.

Duarte J. Siegel S. Young L. (2020). Trust and credit: The role of appearance in peer‐to‐peer lending. Review of Financial Studies, 25(8), 2455–2484.

Go to Reference

21.

Einav L. Farronato C. Levin J. Sundaresan N. (2018). Auctions versus posted prices in online markets. Journal of Political Economy, 126(1), 178–215.

+ Show References

22.

Ennis P. (2016). The four types of Bitcoin users. CoinDesk . https://www.coindesk.com/four‐types‐bitcoin‐users

Go to Reference

23.

Foley S. Karlsen J. R. Putniņš T. J. (2019). Sex, drugs, and bitcoin: How much illegal activity is financed through cryptocurrencies. Review of Financial Studies, 32(5), 1798–1853.

Go to Reference

24.

Fry J. Cheah E.‐T. (2016). Negative bubbles and shocks in cryptocurrency markets. International Review Financial Analysis, 47, 343–352.

Go to Reference

25.

Fuster A. Plosser M. Schnabl P. Vickery J. (2019). The role of technology in mortgage lending. The Review of Financial Studies, 32(5), 1854–1899.

Go to Reference

26.

Ge L. Guo Z. Luo X. (2016). Intermediaries vs. peer‐to‐peer: A study of lenders' incentive on a donation‐based crowdfunding platform. In Americas Conference on Information Systems , San Diego, USA.

Go to Reference

27.

Ge R. Feng J. Gu B. Zhang P. (2017). Predicting and deterring default with social media information in peer‐to‐peer lending. Journal of Management Information Systems, 34(2), 401–424.

Go to Reference

28.

Geva H. Barzilay O. Oestreicher‐Singer G. (2019). A potato salad with a lemon twist: Using a supply‐side shock to study the impact of opportunistic behavior on crowdfunding platforms. MIS Quarterly, 43(4), 1227–1248.

Go to Reference

29.

Ghosh S. (2016). Does mobile telephony spur growth? Evidence from indian states. Telecommunications Policy, 40(10–11), 1020–1031.

Go to Reference

30.

Goldstein I. Jiang W. Karolyi G. A. (2019). To Fintech and beyond. The Review of Financial Studies, 32(5), 1647–16618.

Go to Reference

31.

Gomber P. Kauffman R. Parker C. Weber B. (2018). On the Fintech revolution: Interpreting the forces of innovation, disruption, and transformation in financial services. Journal of Management Information Systems, 35(1), 220–265.

Go to Reference

32.

Griffin J. M. Shams A. (2020). Is Bitcoin really untethered? The Journal of Finance, 75(4), 1913–1964.

Go to Reference

33.

Hackett R. (2017). IMF head: Cryptocurrency could be the future. really. Fortune . http://fortune.com/2017/10/02/bitcoin‐ethereum‐cryptocurrency‐imf‐christine‐lagarde/

Go to Reference

34.

Hastig G. Sodhi M. (2020). Blockchain for supply chain traceability: Business requirements and critical success factors. Production and Operation Management, 29(4), 935–954.

+ Show References

35.

Hatzakis E. Nair S. Pinedo M. (2010). Operations in financial services—An overview. Production and Operation Management, 19, 633–664.

+ Show References

36.

Hildebrand T. Puri M. Rocholld J. (2017). Adverse incentives in crowdfunding. Management Science, 63(3), 587–608.

Go to Reference

37.

Hirshleifer D. Shumway T. (2003). Good day sunshine: Stock returns and the weather. Journal of Finance, 58(3), 1009–1032.

+ Show References

38.

Honohan P. (2008). Cross‐country variation in household access to financial services. Journal of Banking and Finance, 32(11), 2493–2500.

Go to Reference

39.

Iwamura M. Kitamura Y. Matsumoto T. Saito K. (2014). Can we stabilize the price of a cryptocurrency? Understanding the design of Bitcoin and its potential to compete with central bank money, Institute of Economic Research, Hitotsubashi University, Tokyo.

Go to Reference

40.

Iyer R. Khwaja A. I. Luttmer E. F. P. Shue K. (2016). Screening peers softly: Inferring the quality of small borrowers. Management Science, 62(6), 1554–1577.

Go to Reference

41.

Jiang J. Liao L. Wang Z. Zhang X. (2019). Government affiliation and peer‐to‐peer lending platforms in China. Available at SSRN. https://ssrn.com/abstract=3116516

Go to Reference

42.

Kaustia M. Rantapuska E. (2016). Does mood affect trading behavior? Journal of Financial Markets, 29, 1–26.

Go to Reference

43.

Kelly J. (2016). Bitcoin slides on ‘fraud’ warning from JPMorgan's dimon. Reuters . https://reut.rs/3nYyMLN

Go to Reference

44.

Kendall J. (2017). Fintech companies could give billions of people more banking options. Harvard Business Review, 6(4), 8–24.

Go to Reference

45.

Kim K. Hann I.‐H. (2019). Crowdfunding and the democratization of access to capital—An illusion? Evidence from housing prices. Information Systems Research, 30(1), 276–290.

Go to Reference

46.

Kök A. Shang K. Yücel (2020). Investments in renewable and conventional energy: The role of operational flexibility. Manufacturing Service Operations Management, 22(5), 925–941.

Go to Reference

47.

Krijestorac H. Garg R. Mahajan V. (2020). Cross‐platform spillover effects in consumption of viral content: A quasi‐experimental analysis using synthetic controls. Information Systems Research, 31(2), 449–473.

+ Show References

48.

Lichtfous M. Yadav V. Frantino V. (2018). Can blockchain accelerate financial inclusion globally? Inside Magazine . https://theblockchaintest.com/uploads/resources/file‐36385816098.pdf

Go to Reference

49.

Lieber R. (2011). The gamble of lending peer to peer. The New York Times . https://www.nytimes.com/2011/02/05/your‐money/05money.html

Go to Reference

50.

Lin M. Prabhala N. R. Viswanathan S. (2013). Judging borrowers by the company they keep: Friendship networks and information asymmetry in online peer‐to‐peer lending. Management Science, 59(1), 17–35.

Go to Reference

51.

Lin M. Viswanathan S. (2016). Home bias in online investments: An empirical study of an online crowdfunding market. Management Science, 62(5), 1393–1414.

+ Show References

52.

Liu A. Xu S. Zhang M. Zhao X. (2015). Do social media attenuate outsiders' information disadvantage? Evidence from Wikipedia and insider trading . In American Accounting AssociationFinancial Accounting and Reporting Section (FARS)Mid‐Year Meeting, Nashville, Tennessee.

Go to Reference

53.

Liu Y. Sheng J. Wang W. (2021). Technology and cryptocurrency valuation: Evidence from machine learning. Available at SSRN. https://ssrn.com/abstract=3577208

+ Show References

54.

Loughran T. Schultz P. (2004). Weather, stock returns, and the impact of localized trading behavior. Journal of Financial and Quantitative Analysis, 39(2), 343–364.

+ Show References

55.

Malinova K. Park A. (2018). Tokenomics: When tokens beat equity. CNBC . https://ssrn.com/abstract=3286825

Go to Reference

56.

Maskara P. K. Kuvvet E. Chen G. (2021). The role of P2P platforms in enhancing financial inclusion in the united states: An analysis of peer‐to‐peer lending across the rural–urban divide. Financial Management, 50(3), 747–774.

+ Show References

57.

Mbiti I. Weil D. N. (2015). Mobile banking: The impact of M‐Pesa in Kenya. NBER Chapters, In African successes, volume III: Modernization and development (pp. 247–293). National Bureau of Economic Research, Inc.

Go to Reference

58.

Morse A. (2015). Peer‐to‐peer crowdfunding: Information and the potential for disruption in consumer lending. Annual Review of Financial Economics, 7(1), 463–482.

Go to Reference

59.

Ozili P. (2018). Impact of digital finance on financial inclusion and stability. Borsa Istanbul Review, 18(4), 329–340.

+ Show References

60.

Paravisini D. Rappoport V. Ravina E. (2017). Risk aversion and wealth: Evidence from person‐to‐person lending portfolios. Management Science, 63(2), 279–297.

Go to Reference

61.

Philippon T. (2019). On Fintech and financial inclusion. National Bureau of Economic Research No. w26330.

Go to Reference

62.

Pope D. G. Sydnor J. R. (2011). What's in a picture? Evidence of discrimination from prosper.com. Journal of Human Resources, 46(1), 53–92.

Go to Reference

63.

Popper N. (2018). Goldman Sachs to open a Bitcoin trading operation. The New York Times . https://www.nytimes.com/2018/05/02/technology/bitcoin‐goldman‐sachs.html

Go to Reference

64.

Pun H. Swaminathan J. M. Hou P. (2021). Blockchain adoption for combating deceptive counterfeits. Production and Operations Management, 30(4), 864–882.

+ Show References

65.

Salampasis D. Mention A.‐L. (2018). FinTech: Harnessing innovation for financial inclusion. In Chuen D. L. K. Deng R. , (Eds.), Handbook of blockchain, digital finance, and inclusion (pp. 451–461). (Academic Press).

+ Show References

66.

Schuetz S. Venkatesh V. (2020). Blockchain, adoption, and financial inclusion in India: Research opportunities. International Journal of Information Management, 52, 101936.

+ Show References

67.

Seth S. (2020). Explaining the crypto in cryptocurrency. Investopedia . https://www.investopedia.com/tech/explaining‐crypto‐cryptocurrency/

Go to Reference

68.

Stray K. (2017). Bitcoin users: Who they are and what they do? Cointelegraph . https://cointelegraph.com/news/bitcoin‐users‐who‐they‐are‐and‐what‐they‐do

Go to Reference

69.

Tafti A. Zotti R. Jank W. (2016). Real‐time diffusion of information on Twitter and the financial markets. PLoS One, 11(8), e0159226.

Go to Reference

70.

Tang H. (2019). Peer‐to‐peer lenders versus banks: Substitutes or complements? Review of Financial Studies, 32(5), 1900–1938.

+ Show References

71.

Taylor A. (2017). The broken promises of peer to peer lending. Lending Times . https://lending‐times.com/2017/12/13/the‐broken‐promises‐of‐peer‐to‐peer‐lending/

Go to Reference

72.

Vallée B. Zeng Y. (2019). Marketplace lending: A new banking paradigm? Review of Financial Studies, 32(5), 1939–1982.

Go to Reference

73.

Wang J. Gupta M. Rao H. R. (2015). Insider threats in a financial institution: Analysis of attack‐proneness of information systems applications. MIS Quarterly, 39(1), 91–112.

Go to Reference

74.

Wang J. Zhao L. Huchzermeier A. (2021). Operations‐finance interface in risk management: Research evolution and opportunities. Production and Operation Management, 30, 355–389.

Go to Reference

75.

Wei Z. Lin M. (2017). Market mechanisms in online peer‐to‐peer lending. Management Science, 63(12), 4239–4257.

+ Show References

76.

Weinmann M. Mishra A. (2019). The effect of social distance in donation‐based crowdfunding. In International Conference on Information system , Munich, Germany.

Go to Reference

77.

Yang S. (2018). China escalates crackdown on cryptocurrency trading. Bloomberg . https://www.bloomberg.com/news/articles/2018‐01‐15/china‐is‐said‐to‐escalate‐crackdown‐on‐cryptocurrency‐trading

Go to Reference

78.

Zhang J. Liu P. (2012). Rational herding in microloan markets. Management Science, 58(5), 892–912.

+ Show References