J K Rowling
Agreed. While the branch infrastructure is extremely costly, they can’t get rid of most of their branches – at least for now…
Thanks Abdoulaye. You’re the only one that points out that borrowers might want to meet the lending officer for advice. I completely agree, and some of the work-around solutions – like chatbots – are insufficient for many applicants.
Agreed. Definitely good that originators now have to retain a chunk of each securitization.
There are some work-around solutions to address the valuation challenges. For example, even if the process is perceived to be 100% digital, end-to-end, for the consumer, Wells could still send someone to visit each house (costly, but possible).
Thanks. Some of the underwriting data points that you brought up – e.g. race – would never be considered in an application – and some would definitely have to be taken into account to determine ability of the borrower to pay – e.g. income, credit rating. However, you bring up an interesting point about disparate treatment vs. disparate impact. That is, even if your credit models do not explicitly take into account protected/discriminatory data points (e.g. race, gender), the impact (i.e. who gets loans and at what price) might be discriminatory (i.e. applicants from one race are systematically being offered higher rates).
Leigh, interesting perspective. From a credit perspective, I believe the more that Wells underwrites risk based on available data and not human interactions, the less likely they are to see underperforming credit across the portfolio.
Agreed that they have to always be careful not to create “another 2008.” To that end, they have to be mindful about how they fund these loans – e.g. to what extent through securitization markets – and how transparent the cash flows and underlying credit characteristics quality really is.
HJ – good points.
I actually think that from a credit risk and fraud perspective, digitizing the process is risk-mitigating. Given Wells’ vast amounts of historical customer, loan and default data, they can create strong data-driven models that are not vulnerable to poor underwriting decisions on the part of human underwriters. On the fraud side, Wells can license strong third-party softwares that detect fraud based on identity-based questions, facial recognition, IP address, device fingerprint of applicants mobile/PC, and geolocation, among other tools not always leveraged by traditional lenders.
Cyber is a huge risk, and – given the recent stumblings at Uber and Equifax – I’m not sure how well-prepared financial services companies really are to protect our data.
On the customer side, I can understand why a borrower might be skeptical of taking out a mortgage from a no-name upstart online lender. I think Wells Fargo will face less, albeit some, resistance given their brand. At the same time, Wells does not plan to close most of their brick-and-mortar mortgage branches – at least not immediately; they just want to provide a convenient alternative for their tech-savage, time-constrained customers.
I strongly agree with Michael above that there is not a strong market demand for increased design customization in sneakers. Sneaker customization has been offered by Nike and its competitors since I was in elementary school, when I designed both running shoes and basketball shoes with my name and number on them. While it’s true that Nike could invest in expensive new 3D printing technology to accommodate customization at scale, I don’t think it would be addressing any customer pain point at scale; rather, I cultish sneaker-heads and design buffs are still niche.
Secondly, while I agree with you that it would be ideal for Nike to be integrated with their customers’ IT systems in order to analyze sales data in real-time, I question the viability of this plan. Nielsen and IRI are two data companies that already provide robust and fairly reliable retail sales data (on sell-through rates and distribution, for example), and they publish updated data every few weeks; One Click Retail offers the same data services for the online retail world. I believe it would be a tall ask for Nike to demand that their retail customers integrate their IT systems for data sharing, given that (a) other manufacturers do not do so and (b) good alternatives like Nielsen, IRI and One Click exist.
I agree with Michael’s point that Uber and other gig-economy platforms will not transition from contractors to full-time employees. Uber is already burning ~$2B/year, so, as you said, they’d have to pass along this ~30% cost increase along to their customers. Rather than “setting the standard” for corporate responsibility as you suggested, I think Uber would risk being undercut on price by your competitors and/or shrinking the overall size of the pie that they created for ridesharing in the first place. Moreover, it would be difficult to distinguish between full-time and part-time drivers, and enforce hours for full-time employees. It is possible that they reach an intermediate classification – like Germany’s “dependent contractors” for freelancers who work primarily for one business – but I don’t think this is a focus right now.
Moreover, I think there’s another interesting labor dimension that is introduced by an autonomous Uber fleet. While the need for drivers will certainly shrink with autonomous vehicles, Uber will have to add a significant number of full-time employees for other positions, like maintenance and filling up with gas. (Maintenance labor costs will be significant due to the high vehicle utilization rates of autonomous cars.) Do you think Uber, in an effort to create some positive PR buzz, should lead the way in job retraining (e.g. for mechanics) in anticipation of a shift toward driverless cars?
Thanks Daniela. You mentioned Go-Jek as being a similar company in South-East Asia (also shares some VC investors I believe). I’ve noticed that Go-Jek has a worked to develop more products – in addition to their original on-demand services – to build a moat around their business, making switching more difficult for Go-Jek users. Specifically, in addition to their payments app, Go-Jek has created their own loyalty program through which Go-Jek customers can redeem earned points and tokens at partner retailers and restaurants. This is very similar to credit card or hotel loyalty programs in the US, as it disincentivizes home-sharing and encourages more transaction through Ge-Jeks apps and payment processors. I believe a similar loyalty program would help Rappi reach profitability by increasing customer lifetime value. Do you know if Rappi is moving in this direction?
I agree that the overall industry should move toward value-based, consumer-driven healthcare, and that collecting patient data through wearables has many interesting applications.
My cousin ran a study at UCSF last year to evaluate whether giving away free Fitbits to adult Type 2 diabetic women would increase weight loss. In addition to incentivizing healthy behavior (which it effectively did), it provided health care providers with accurate, reliable data on consumer behavior.
HBS professor Regina Herzlinger has long-advocated for consumer-driven healthcare, which she illustrates through a case on South African insurer Vitality . Essentially, she thinks that consumers should get paid if they take care of their diseases. If you have asthma and quit smoking, you get paid. If you have diabetes, and go to the gym, you should be compensated. The financial return to employers for participating in these systems is in increased employee productivity; for example, fewer employees have back problems that distract them from work.
Tying this back to wearables – if you subsidize the cost of wearables, you can collect better data on whether patients are actually participating in some of these activities, and compensate them for that. On the mental health side, innovative upstarts like Ginger.io have developed algorithms that monitor your activity through mobile or wearable devices and notify your close-ones to intervene if you’re exhibiting behavior predictive of a mental health episode.
Great article, and timely given last week’s IPO.
I’m very worried about the threat that Amazon poses to Stitch Fix. Right now, the equity story that Stitch Fix is telling the public markets is that its competitive edge – the “moat” it has built – is its level of personalization, achieved through advanced data techniques; they’ve even said that investors need education to understand how importance data science is to their model.
As you cited, Amazon unveiled Prime Wardrobe this summer. My sense is that Amazon can replicate this level of personalization in their Prime Wardrobe through machine learning fairly quickly. After all, they have the machine learning know-how, engineering and data talent, a huge existing customer base, and ample data on their existing customers’ preferences. (As the story goes, they recommend diapers to you before you even know you’re pregnant.) As seen with Blue Apron, Amazon’s decision to go into markets in a focused way can crush a new public challenger’s stock price.
Moreover, I’m dubious about Stitch Fix’s claims that data science can, in addition to customize recommendations, effectively gauge consumer trends and forecast fashions. In my view – and as we learned in the GAP case, predicting trends is more of an art than a science. Moreover, they would be in even more of a position of strength if they worked to dictate trends, which would be done by creative directors, not data analysts.
Finally, I think an interesting dynamic here is the fact that Katrina Lake and a couple of her early investors still – post-IPO – control the company through a special class of stock with higher voting rights. What affect do you think this will have on management, strategic priorities, and do you think this fact is baked into/bringing down the stock price at all?
Very interesting, and lots of good parallels to the development of the SMB online lending space. While I agree that SMB insurers offer a strong value prop to business owners, I think the conclusion that you’ve drawn – about mid-sized insurers inevitably dying off – is a bit draconian.
Proponents of the online lending space drew similar conclusions a couple years ago about community and regional banks’ abilities to respond to marketplace lenders offering a less cumbersome application process. What industry observers have seen, however, over the last year is that smaller incumbents have actually built their own online capabilities faster than expected (we’re doing a case on one such MA-based community bank in Entrepreneurship Management this spring). One way to do so here without significant upfront technology costs is to license white-label or co-branded online application services from a sort of “insurance-as-a-service” company. (If such a company doesn’t exist yet, I might start it 😉 )
To respond to your last question, I’d also be curious about whose balance sheet the insurance liability actually sits for digital insurers. If it sits on the balance sheet of an upstart digital insurer, you’re in essence asking the VCs to subsidize high losses rather than exercise underwriting discipline from the onset. On the other hand, some upstart digital insurers are partnering with large insurers or reinsurers including Hanover Re and Munich Re to use their balance sheets; I do not think these companies will tolerate significant upfront losses on claims to accumulate data for future underwriting models.