Subscribe: Random Thoughts
http://andirog.blogspot.com/feeds/posts/default
Added By: Feedage Forager Feedage Grade B rated
Language: English
Tags:
borrowers  club  credit grade  credit line  credit  issued  lending club  lending  loan  loans issued  loans  month 
Rate this Feed
Rate this feedRate this feedRate this feedRate this feedRate this feed
Rate this feed 1 starRate this feed 2 starRate this feed 3 starRate this feed 4 starRate this feed 5 star

Comments (0)

Feed Details and Statistics Feed Statistics
Preview: Random Thoughts

Random Thoughts



Cloud, Crowdfunding and everything else ...



Updated: 2018-01-10T03:25:48.326-08:00

 



Building a Python/Django Development Virtual Machine

2014-02-25T08:00:05.864-08:00

Recently for a project, I needed to build a Python, Django, PostgreSQL, NGINX Development Virtual Machine(VM). Below are the steps that I followed to build this VM in VMware Fusion on MacbookPro (MBP). This post is as much about sharing my build experience as documenting the steps for my future use and potential automation.The guidance for this procedure came from How To Install and Configure Django with Postgres, Nginx, and Gunicorn.InstallationUbuntu Server OSThe steps were very similar to the ones I covered in my prior post OpenStack: Quick Install using DevStack.Download Ubuntu 12.04 "Precise Pangolin" x86_64 Minimal CD ISO Image mini.iso.Start VMware Fusion and select Virtual Machine Library in Windows option on VMware Fusion toolbar. This will bring up Virtual Machine Library window showing all the Virtual Machine already available.Click Add button and select New. This will bring up New Virtual Assistance Window showing Create New Virtual Machine.Click Continue without disc as we will be using the downloaded ISO image. This will bring up Installation Media section.Select Use operating system installation disc or image and click on arrows next to Choose a disc or disc image.... Select the Ubuntu ISO image and then click Continue.Choose Operating System section should show Linux as Operating System and Ubuntu 64-bit as Version. Click Continue.The Finish section will show Virtual Machine Summary, Click Finish. Select the location where we want to save the VM file and name the file.A console window will be launched and OS install will start. Answered the prompts during the install process. Once update and reboot completes, a login prompt will appear. Log in to VM.OpenSSH ServerAfter login, install OpenSSH Server to enable access to Ubuntu VM over SSH.$sudo apt-get install openssh-serverCheck whether SSH process is running.$service ssh statusEither note down the IP address of VM from login screen (shown above) or using ifconfig command to be able to SSH into the VM remotely.$ssh anil@172.16.191.158You may need to remove SSH key if there is a fingerprint mismatch between the VM and remote client.$ssh-keygen -R 172.16.191.158Update PackagesYou need to make sure all installed packages are current. Download any package updates and install.anil@django:~$ sudo apt-get update[sudo] password for anil: Hit http://us.archive.ubuntu.com precise Release.gpgGet:1 http://us.archive.ubuntu.com precise-updates Release.gpg [198 B]...At this point VM is ready for installation of Python Virtualenv, Django, PostgreSQL, NGINX, and Gunicorn.Python VirtualenvVirtualenv is Virtual Python Environment builder to create separate Python environments. This enables to keep installations, dependencies, versions and permissions separate for different applications across different virtual environments.Install python-virtualenv.anil@django:~$ sudo apt-get install python-virtualenvReading package lists... DoneBuilding dependency tree Reading state information... DoneThe following extra packages will be installed: python-pip python-setuptoolsThe following NEW packages will be installed: python-pip python-setuptools python-virtualenv...Now, we need to create a virtual environment for our project (in this case, lendcafe) where we can install Python and Django packages.anil@django:~$ sudo virtualenv /opt/lendcafeNew python executable in /opt/lendcafe/bin/pythonInstalling distribute...........................................................done.Installing pip...............done.You can name anything you like for your virtualenv.DjangoDjango is a Python web framework. It enables rapid development of common web application tasks and adheres to DRY principle (Don't Repeat Yourself).To install Django, first we need to activate virtualenv.anil@django:~$ source /opt/lendcafe/bin/activate(lendcafe)anil@django:~$ The activate script modifies shell prompt to show the currently active environment.Install Django(lendcafe)anil@django:~$ sudo pip install django[sudo] password for anil: Downloading/unpacking django Downloading Django-1.6.2.tar.gz (6.6Mb): 6.6Mb do[...]



OpenStack: Virtual Image Instances using Horizon Dashboard

2014-02-11T08:00:04.630-08:00

Install AddendumEnable VT in BIOSAn addendum to install steps defined in my previous post OpenStack: Quick Install using DevStack is required to avoid a surprise that I encountered after the install. Please check to make sure BIOS is at latest version available from the system manufacturer and Intel's Virtualization Technology (VT) is enabled in BIOS.anil@OSCloud:~$ sudo apt-get install cpu-checkeranil@OSCloud:~$ sudo kvm-okINFO: /dev/kvm existsKVM acceleration can be usedIf CPU doesn't support VT, the output will show CPU does not support KVM extensions.The OpenStack Horizon Dashboard is implemented as a Python/Django web application that provides admin and user interface to OpenStack services.Horizon DashboardLog inIn web browser, type the IP address for the dashboard. On Log In page enter User Name and Password and click Sign In. When signing in as Admin, the home page shows the Admin panel - System Panel - Overview.Existing Virtual Machine ImagesBy clicking Images category in Admin - System panel on the left, a list of available images are viewed. In default installation, CirrOS x86_64 image is made available in AMI/ARI/AKI format.CirrOS images are tiny cloud guest images with minimal Linux distribution that can also be downloaded from LaunchPad. The AMI/ARI/AKI is the image format supported by Amazon EC2. AMI (Amazon Machine Image) is a virtual machine raw image. ARI (Amazon Kernel Image) is a kernel file (vmlinuz) that will load initially to boot image. ARI (Amazon Ramdisk Image) is ramdisk file (initrd) mounted at boot time.Launch InstancesClicking on the Project tab in left panel shows the overview of current project.To launch an instance from an image, click Images and Snapshot category in Project - Manage Compute panel on the left.Select an image and click Launch. A Launch Instance modal pop-up appear. Enter a name in Instance Name field in Details tab.In Access & Security tab, enter a passphrase in Admin Pass and Confirm Admin Pass fields.Upon clicking Launch, Horizon dashboard switches to Project - Manage Compute - Instances page and shows the Instances running.Clicking on Instance Name hyperlink shows the Instance Details for that specific instance with three tabs for Overview, Log and Console.Though the Project - Manage Compute - Instances page shows instance to be Active and Running, the console for the instance is displaying an error message.This kernel requires an x86-64 CPU, but only detected an i686 CPU.Unable to boot - please use a kernel appropriate for your CPU.Error TroubleshootingA little bit of googling suggested to check whether the 64-bit PC (amd64, x86_64) or 32-bit PC (x86) version of host operating system is installed. Sure enough, the Ubuntu version installed on OSCloud host is x86 and not x86-64 version. I can't use x86-64 instance images on OSCloud host.anil@OSCloud:~$ uname -aLinux OSCloud 3.2.0-58-generic-pae #88-Ubuntu SMP Tue Dec 3 18:00:02 UTC 2013 i686 i686 i386 GNU/LinuxAfter terminating the newly created instance test1 and deleting all x86_64 Images, the next step was to either find or build x86 images and start a new x86 instance.Prebuilt Virtual Machine ImagesAs OSCloud host is using QEMU Hypervisor, it made sense to look for qcow2 (QEMU copy-on-write) format x86 images. At CirrOS download page, I found a bootable qcow disk image for i386 and decided to try it out.Create ImagesTo create images, on Admin tab, select Images and then click Create Image button in right pane. On Create An Image page, enter Name for the image, select Image Source, Image Location, and Format. Select the Public checkbox to make available this image to everyone. Then click Create Image. The image will be queued for creation.Once images are created, they will be available to launch instances in projects following the steps listed above in Launch Instances section.In next blog post, I will start to dig deeper into high level solution design using OpenStack. Your feedback and comments are welcome.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive [...]



OpenStack: Quick Install using DevStack

2014-02-04T08:00:02.955-08:00

Though not a recommended method for installing OpenStack for production, DevStack offers an easy method to install and run an OpenStack cloud either on hardware or even within virtual machine.In this post, I walk through installing OpenStack using DevStack on an old Sony Vaio laptop on local network. The DevStack site provides instructions for installing OpenStack on Virtual Machines and on Hardware. The detailed instructions for installing OpenStack on a single hardware machine, that I followed, are available at DevStack.Install Ubuntu Server OSI repurposed a SONY VAIO laptop for DevStack install that was originally wiped clean with Darik's Boot and Nuke in preparation for disposal.As DevStack downloads and installs all dependencies, I downloaded 64-bit PC (amd64, x86_64)Ubuntu 12.04 "Precise Pangolin" using Minimal CD mini.iso and burnt a CD on my MacBook Pro.Booted Sony laptop using the Ubuntu minimal CD and selected Install from Installer boot menu, followed the prompts, and accepted default options for most prompts. It takes over an hour for the installation to download and install the base system.Installed OpenSSH Server to enable access to Ubuntu server over SSH from my MBP. $sudo apt-get install openssh-serverChecked whether SSH process is running by using one of the two commands listed below. The output shows process running. anil@OSCloud:~$ ps aux | grep sshroot 1394 0.0 0.0 6684 2416 ? Ss 21:43 0:00 /usr/sbin/sshd -D anil 1534 0.0 0.0 4384 836 pts/0 S+ 22:06 0:00 grep --color=auto sshanil@OSCloud:~$ service ssh statusssh start/running, process 1394Checked the IP address of Ubuntu server so that I can remotely access the server. anil@OSCloud:~$ ifconfigeth0 Link encap:Ethernet HWaddr 00:1d:ba:23:9a:c5 inet addr:10.0.1.25 Bcast:10.0.1.255 Mask:255.255.255.0 inet6 addr: fe80::21d:baff:fe23:9ac5/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:875 errors:0 dropped:0 overruns:0 frame:0 TX packets:202 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:522168 (522.1 KB) TX bytes:19796 (19.7 KB) Interrupt:16 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)While trying to access Ubuntu server from MPB over SSH, I received following error. ANILs-MacBook-Pro:~ anilgupta$ ssh anil@10.0.1.25@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!Someone could be eavesdropping on you right now (man-in-the-middle attack)!It is also possible that a host key has just been changed.The fingerprint for the RSA key sent by the remote host is26:54:a3:4e:cd:a3:6c:80:f3:36:2c:b3:c9:17:f0:db.Please contact your system administrator.Add correct host key in /Users/anilgupta/.ssh/known_hosts to get rid of this message.Offending RSA key in /Users/anilgupta/.ssh/known_hosts:6RSA host key for 10.0.1.25 has changed and you have requested strict checking.Host key verification failed.This is well-known error when there is fingerprint mismatch between the host (Ubuntu Server) and client (MBP). In this case, as there is no malicious attempt, I removed the offending key from MBP using ssh-keygen -R command.ANILs-MacBook-Pro:~ anilgupta$ ssh-keygen -R 10.0.1.25# Host 10.0.1.25 found: line 6 type RSA/Users/anilgupta/.ssh/known_hosts updated.Original contents retained as /Users/anilgupta/.ssh/known_hosts.oldAfter removing the offending key, I was able to successfully SSH into Ubuntu Server.ANILs-MacBook-Pro:~ anilgupta$ ssh anil@10.0.1.25The authenticity of h[...]



OpenStack: Overview of Service Components

2014-01-28T08:00:03.705-08:00

OpenStack is made up of several projects that together build up the OpenStack cloud. Three core components are Compute, Storage, and Network.OpenStack Training Guides provide a nice conceptual architecture of OpenStack service components, created by Solinea.Compute (Nova)Nova services provide computing resources through virtual machine (VM) instances and works with several virtualization technologies, such as KVM, QEMU, VMWare ESX, Xen, Hyper-V, and LXC. A complete list of supported Hypervisors is listed at OpenStack Wiki.Nova is also called Cloud Controller as it provides framework for provisioning and managing VMs. Nova also provides ephemeral storage.With Nova services, IT departments can offer private cloud services to internal departments that can scale computing resources on-demand as workload varies. It is analogous to Amazon EC2.Nova is made up of several service components:nova-compute: Runs the VM instances.nova-scheduler: Decides which host will run the requested instance.nova-api: Service interface to Nova, such as a call to start up a Nova instance.nova-network: Network servicesnova-objectstore: File storage servicesnova-common: The underlying common librariesnova-cert: Certificate management service used to authenticate to NovaObject Storage (Swift)By default, all storage disappears when VM instance is terminated. Swift services provide cost-effective scale-out redundant persistent storage to VM instances and responsible for ensuring data replication and integrity. The object storage is suitable for static data and stored as Objects. The objects are stored and replicated on disks spread across storage cluster nodes. The stored data persists until deleted by users.Swift is analogous to Amazon S3. Swift provides RESTful API for integration with other application, storage and services.Swift is made of several service components:swift-proxy: Accepts, authorizes, authenticates incoming requests.swift-account: Manages database of accounts.swift-container: Contains mapping of containers.swift-object: Contains mapping of objects.Block Storage (Cinder)Cinder provides high performance persistent block storage for use with VM instances. Cinder volumes are appropriate for database, file system and raw block storage. Cinder is supported by several storage platforms including Ceph, Nexenta, SolidFire, Zadara, CloudByte, Coraid and Scality to name a few startups in this area. A complete lists of storage devices supported by Cinder drivers is available at Cinder Support Matrix.Cinder is analogous to Amazon EBS. It also provides snapshot management with capability to either restore snapshot or use snapshot to create a new block storage volume.Following services are available with Cinder:cinder-api: Authenticates and routes requests to block storage.cinder-scheduler: Scheduling and routing requests to volume service.cinder-volume: Managing back-end block storage devices.cinder-backup: Backup Cinder volume to Swift.Network (Neutron)Neutron (formerly Quantum) provides virtual network service for connectivity and addressing used by other services. OpenStack networking can have multiple private networks with overlapping IP addressing schemes and relies on Keystone for authentication and authorization for API access. A standard implementation includes management network, data network, external network, and API network. Though started with basic Linux VLANs and IP tables, it now includes plugins which are pluggable back-end implementation of OpenStack Networking API: Open vSwitch, Linux Bridge, NEC OpenFlow, etc.Following services are available with Networking:quantum-server: Passing user requests to the configured Plugin for processing.plugin agent (quantum-*-plugin-agent): Perform local vSwitch configuration on hypervisor.dhcp agent (quantum-dhcp-agent): Provides DHCP services to internal networks.l3 agent (quantum-l3-agent): Provides L3/NAT forwarding for access to external network.Image (Glance)Glance provides catalog and repository for disk, ser[...]



OpenStack: The Building Block for Private Cloud

2014-01-23T08:00:05.060-08:00

I am back after an extended break from blogging about data storage topics. As I am no longer working at Quantum, I am free to blog about the recent developments in data storage without any concerns.

Last year, I started using cloud servers on Digital Ocean for PeerCube and worked with Amazon Web Services during Coursera's Introduction to Data Science course. Since then, I became very interested in private clouds and methods to establish and manage them.

With the rise of server virtualization, users and applications ability to spin up and spin down pre-built images as needed, and the success of Amazon Web Services (AWS) public cloud, there are fewer reasons for IT administrators to actively manage pooling and allocation of IT infrastructure resources. While there are several commercial (VMware vCloud) and open source platforms (Apache CloudStack, Eucalyptus, OpenStack) that let you build Infrastructure as a Service (IaaS) private cloud, I am particularly excited with the developments happening in OpenStack community.

I was sold on OpenStack as soon as I was able to install a DevStack environment on an old laptop and spin up images within couple of hours. Since then, I have been exploring OpenStack documentation and source code to understand this platform better with private cloud as a specific application in mind.

OpenStack Components

The modularized architecture of OpenStack includes following service components:
  1. Compute (Nova)
  2. Object Store (Swift)
  3. Block Storage (Cinder)
  4. Neutron, formerly Quantum (Network)
  5. Image (Glance)
  6. Identity (Keystone)
  7. Dashboard (Horizon)
In the next blog post, I will elaborate on each services components. Being from data storage industry, I am very interested in learning about the storage service components of OpenStack in-depth.

As I realized that I learn best by writing and sharing, future blog posts will be my journey toward understanding OpenStack and how to use it to establish and manage private clouds.

Book Resources




Lending Club Borrower's Income Verification, Loan Issued Year, and Initial List Status

2013-04-08T07:30:01.136-07:00

Recently Lending Club modified the historical loan data file, included several new loan and borrower attributes and removed a few. One of the new fields is whether a borrower's income was verified by Lending Club during the loan application process. Recently, there was a discussion about verified income at LendAcademy forum. Some of the questions and concerns raised during the discussion were:Has the number of available loans with verified income gone down recently?Does the income verification really matter with loan performance?Does Lending Club verify income for all loans?Which borrowers are more likely to have their income verified?It will be interesting to find answers to some of these questions and supporting data from the new historical loan data file.Loan Issued YearThe chart below shows the borrowers' income verification for the loans by issued year. The loans with verified income are listed as TRUE and with unverified income are listed as FALSE. Two questions listed above are right away addressed from this chart.Lending Club doesn't verify borrower's income for all loans issued.The percentage of loans issued with verified income has gone up recently. Whether the retail lenders are seeing the loans with verified income at the time of offering is an open question.In 2007, borrowers' income was not verified at all. Since then, the percentage of loans with verified income has been rising. In 2013 year to date, more loans were issued that borrower's income was verified than the loans with unverified income.Initial List Status of LoanLending Club reserves a few loans for 12 hours and offers them to the institutional and large retail lenders who want to lend the whole amount for a loan. I am not sure whether the historical loan data file includes the loans that were offered and picked up by lenders as 'whole' loans. But, the loans that were initially offered as whole, designated with 'w', but not picked up as 'whole' loans are listed in historical loan data file.The chart below shows the percentage of loans with verified and unverified income of borrowers with initial listing status and issued year of the loans. Lending Club started offering the 'whole' loans only since late 2012. With the limited data, there doesn't appear to be any significant difference in percentage of loans with verified income between the loans that were initially offered as whole or fractional.Key TakeawaysLending Club is verifying borrower's income for greater percentage of loans issued on its platform recently.There is no significant difference in income verification for loans initially listed as fractional or whole.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loans Issued Since 2010 - Principal Paid Back and Months of Payment

2013-03-21T07:30:00.727-07:00

This post is the last in the series of posts discussing when default of loans start to peak (Part 1, part 2, part 3, and part 4).Months of PaymentThe chart below shows the percentage of 36 month and 60 month loans defaulted as a function of months of payment for loans issued since 2010. By reviewing both 36 month and 60 month loans issued in same time frame, we may be able to better compare such loans. While the default patterns are very similar for first 10 months of payment, the rate of defaults increases rapidly for 60 month loans after 10 months of payment. 50% of defaults for both 36 month and 60 month loans occurred within 8 months or so, the 80% of defaults for 60 month loans occurred within 13 months compared to 15 months for 36 month loans.Principal Paid BackSimilar chart for Principal paid back is shown below. It is clear from the chart that while 50% of loans of both maturities defaulted within 8 months, the 60 month loans paid back (9%) only half of principal compared to the principal paid back by 36 month loans (18%). Can the 60 month loans that continue to make payment make up for this extra loss in principal with longer repayment duration and/or higher interest rate?Another interesting observation from above chart is the increasing difference in principal paid back between loans of 36 month and 60 month maturities. For example, the 20% of defaulted 60 month loans paid back 4% of principal little more than half of 7% principal paid back by 20% of defaulted 36 month loans. In comparison, the 80% of defaulted 60 month loans paid back 16% of principal less than half of 37% principal paid back by 80% of defaulted 36 month loans.The chart below shows the scatter plot of Principal paid back and Months of payment for 36 month and 60 month loans issued since 2010. A second order polynomial trend line is shown on the chart separately for 36 month and 60 month loans. As the principal portion in monthly repayments for 60 month loans is much smaller than that for similar 36 month loans, the increasing difference between principal paid back with months of payment is understandable.Key TakeawaysIn the end, the months of payment is much more straightforward method to determine when defaults peak.For 36 month loans, 50% of defaults are expected to occur within 10 months of payment, and 80% of defaults within 20 months. The default trend for both 36 month and 60 month loans is very similar for first 8 months of payment.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loans - Principal Paid Back and Defaults

2013-03-18T07:30:02.553-07:00

In this post, I will analyze the defaults based on principal paid back by the borrowers before default. For earlier posts in the series, please refer to Lending Club Loans - Defaults with Loan Age, Lending Club Loans - Months of Payment before Default, and Lending Club Loans - Principal Paid Back and Months of Payment.Loan LengthThe chart below shows the percentage of loans defaulted as a function of principal paid back for 36 and 60 months loans. Similar to previous methodologies, the curve for 60 month loan is exaggerated, i.e. higher defaults at lower principal paid back because all of 60 month loans are less than 3 years old.One interesting observation from this chart is that 2.7% of 36 month loans that default pay back less than 0.04% of principal, i.e. borrower only makes one or two payments on the loan. A few other interesting observations for 36 month defaulted loans are:About 44% of defaulted loans pay back less than 20% of principal.About 80% of defaulted loans pay back less than 50% of principal.About 95% of defaulted loans pay back less than 80% of principal.Put another way, we can expect 50% of our defaulted loans to pay back less than 24% of principal. This causes a double whammy for peer to peer lenders who are using a non-retirement account for lending. The interest earned is taxed at higher ordinary income tax rate while the principal lost to default is deducted from capital gains that are taxed at much lower rate. The monthly repayments during the first year consist mostly of interest resulting in lender paying higher percentage of interest income in taxes while lower percentage of principal loss to offset the capital gains if borrower defaults within a year.Credit GradeThe chart below shows the percentage of loans defaulted as a function of principal paid back for various credit grades. There are no surprises here.The defaults with principal paid back for loans with credit grades E, F, and G behave very similar to each other. The similar trend is also seen for  loans with credit grades B, C, and D. This may suggest that better returns to be have by investing in the loans with higher interest rate within each group.As expected the defaults of higher quality loans, i.e. credit grade A loans, tend to pay back larger portion of original principal. The open question is whether the lower interest payments for such loans cover the principal loss.Key TakeawaysOverall, I am disappointed that principal paid back didn't prove to be as effective of methodology as months of payment in determining when loan defaults peak. It appears sometime simpler measurements are much more effective in describing the trends. Also, none of the methodologies discussed able to explain the defaults of 60 month loans without observing such loans to maturity.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loans - Principal Paid Back and Months of Payment

2013-03-16T16:31:55.377-07:00

In the last post, I analyzed the defaults based on number of monthly payments made by borrower before default. In next post, I will analyze the defaults using another methodology of principal paid back before default. In this post, I will review the relationship between principal paid back and months of payment.Principal Paid BackWhile months of payments methodology provides us a time frame when most defaults occur, it doesn't provide a decent comparison for loans in different maturity cycle and loans of different maturity length. My expectation is that the principal paid back will serve as an appropriate proxy for matching the loans at the same point in maturity cycle. For example, a 3 year loan at 10% interest rate would take about 20 months (55% of maturity length) to pay off 50% of principal while same loan with 5 year term would take about 34 months (56% of maturity length) to pay off 50% of principal. In this analysis, I assume that defaults for these two loans should behave similarly when each has paid off same percentage of principal.Principal Paid Back is one of my favorite data point in evaluating a lending strategy as it is a good method to gage the risk tolerance of a lender. It quickly communicates on average what percentage of principal is going to be recovered from defaulted loans. Combined with the return from fully paid loans, it also communicates how many loans a lender needs to recover principal lost from defaulted loans and just break-even.For example, the screen capture above from PeerCube shows historical performance of loans issued between 2007 and 2009 for a specific lending strategy that only includes borrowers who own their home. On average, each defaulted loan paid back only about 42% of principal. To break-even, this strategy need to recover 58% of lost principal from other loans in the portfolio. With, on average, only 14% return (ROI to be discussed in future blog post) from fully paid loans, a lender need at least 4 loans to be fully paid to just break-even. If a lender invested in all loans meeting this criteria between 2007 and 2009, 55% of loans contributed 0% to return and only 45% loans contributed to achieving 7.32% ROI from this strategy.Months of PaymentThe chart below shows the Principal Paid Back as a function of Months of Payment for both 36 and 60 month loans. Actually axis are swapped as certain observations listed below are easier to see this way. [Edits 03/16/2013: As requested by Andrew in comments below, updated the chart to include linear trend lines and screen capture of trend model description.]The 36 month loans and 60 month loans clearly have two different paths on  the chart. It is clear that the principal payback schedule is different for 36 month and 60 month loans. The scatter plot for 60 month loans shows most data points in the region below 30 months of payment and left of 50% principal paid back. This is primarily due to 60 month loans issued only since second quarter of 2010. The solid color at 100% principal paid back mark for loans with both terms is due to loans that are fully paid either on schedule or ahead of schedule.The chart below shows the Principal Paid Back as a function of Months of Payment for loans with various credit grade.This chart is very similar to the previous chart. From density of different colors, it can be observed that the 60 month loans are primarily carry credit grade E, F, and G. Also, it appears that majority of 60 month loans that are paid off early carried credit grade A, B, and C.Key TakeawaysThe Principal Paid Back would be a good data point to gage the risk tolerance of a lender and variance in return of a lending strategy over the loan maturity cycle.Reviewing the loans that have paid back less than 50% of principal may offer better comparison between 36 and 60 month loans. border="0" frameborder[...]



Lending Club Loans - Months of Payment before Default

2013-02-27T07:30:00.791-08:00

In the last post, I reviewed the defaults based on loan issued date. As I mentioned in the previous post, due to point in time snapshot of historical loan data that Lending Club provides, it is difficult to determine exactly when a loan actually was charged off or defaulted.Months of PaymentIn this post, I will review the defaults using a different methodology. Before a loan is charged off or defaulted, borrowers stops making monthly payments on the loan. Based on the total payments made by borrower, we can determine approximately the number of months or number of times monthly payments were made before loan was charged off.The months of payment will always be smaller than when actually loans was charged off as Lending Club can take significant time to write off a loan once payment stops. Also, as there is no information about partial payments and any late fees in historical loan data file, this analysis assumes that all payments were made toward monthly payments.Loan LengthThe chart below shows the percentage of defaulted loans as a function of number of monthly payments made for 36 month and 60 month loan terms.At first glance, someone may make following observations:The 60 month loans default quicker than 36 month loans.Most 60 month loans default within first 24 months.Both 36 and 60 month loans have similar default trend within first six months after loan issued date.Such conclusion may not be correct. While 36 month loans of at least three vintage issued years have reached full maturity, none of the 60 month loans have gone through complete maturity cycle. The 60 month loans were first issued in early 2010. This is the main reason why the curve for 60 month loans gives the impression that most defaults happen within first two years.Following observations can be made for default behavior of 36 month loans:About 20% of all defaulted loans make five or less monthly payments.About 50% of all defaulted loans make ten or less monthly payments.About 80% of all defaulted loans make twenty or less monthly payments.Put another way, we can expect half of our default loans to occur before a borrower makes ten complete monthly payments. Assuming Lending Club takes on average four months (120 days) after payments stop to charge off loan, we can expect half of our defaults to occur by 14th month after issue date.Credit GradeThe chart below shows the percentage of defaulted loans as a function of number of monthly payments made for different Credit Grades.There doesn't appear to be any significant difference in default trend for loans with different credit grade. The separation of curves for Grade E, F, and G from rest of the pack at 15 months of payment and higher may suggest that greater number of such loans are defaulting earlier. Majority of E, F, and G grade loans are of 60 month term. As mentioned earlier none of the 60 month loans have reached full maturity yet. The separation in curve is most likely result of incomplete defaults data for 60 month loans to maturity.Key TakeawaysWe can expect half of our default loans to occur before a borrower makes ten monthly payments, i.e. by 14th month after loan issue date.It is too early to compare default trend of 60 month loans with that of 36 month loans.There is no significant difference in default trend for loans with different credit grade.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loans - Defaults with Loan Age, Part I

2013-02-25T07:30:01.021-08:00

February has been slow month from blogging perspective for me. I spent majority of time making improvements to PeerCube. Three major enhancements released this month are ability to invest in multiple loans together from the list of loans, more robust BLE Risk Index that now includes 19 different loan and borrower attributes, and showing a list of loans with similar risk profiles as the one being viewed by a user.When Default of Loans Start to Peak?Recently, a participant on LendAcademy forum asked when defaults of loans start to peak. It is an interesting question so I decided to look further into historical data for Lending Club loans and see if I can find patterns for loan defaults. Unfortunately, Lending Club only provides point-in-time snapshot of historical loan data so I can't observe when a loan entered in default state. But, there are several different methods that can help provide insights from point-in-time snapshot.One such method is to review the loans that are currently in default and when they were issued. This method assumes that similar pattern, in aggregate, will persist for loans in the future. One challenge with this method is that any variation in loan volume will skew the pattern. For example, if 1,000 loans defaulted out of 100,000 loans issued last year versus 400 loans out of 10,000 loans issued prior year, we may erroneously assume that loans default more within a year of being issued. In following analysis, I use the percentage of loans with default status to smooth out any effect of volume.The chart below shows the percentage of loans with default status as a function of loan issued date. I only included 3 year term loans primarily because I have the historical data that covers the loans from issued date to maturity. Also, the 5 year term loans may have different default pattern. A peculiarity you may notice is the way I have chosen to plot X-axis (loan issued date). Instead of ascending issued year, I have reversed the axis and plotted percentage loan defaults with descending issued year. In fact, the chart below is a mirror image. As we are more interested in knowing when a loan may default, I believe flipping the X-axis better communicates visually the trend.While reviewing this chart, think of that you are trying to find out how many 3 year term loans issued in January 2013 may default by the end of 2016. Think of right side of 2012 being end of 2013, right side of 2011 being end of 2014 and so on. The major assumption here is that future monthly default trend is exactly represented by the past monthly default trend.ObservationsWhile reviewing the above chart, two observations right away stand out:No loans are charged off or defaulted within first six months. This is understandable as majority of loans will go through stages of In Grace Period, Late (16 - 30 days), and Late (31 - 120 days) before being charged off. So, earliest most loans can be charged off is at least 120 days (4 months) after being issued.Th peaks appear during the year at regular interval. I am not very sure but I suspect this may have to do either with the time (end of December) when this historical loan file was downloaded or the Lending Club defaulting loans in batches at regular interval.Based on the trend line models in this chart, for all 3 year term loans purchased in January 2013, we can expect 2.3% loans in default by the end of first year, 6.2% loans in default by the end of second year and 10.5% loans in default by the end of third year. By the time all loans mature, we can expect 12.4% loans originally issued to default. These numbers were obtained by substituting 0 for Month of Issued Date in trend line model for each year.For 100 loans issued in January 2013, 2.3 loans can be expected to default in first year, 3.9 loans default [...]



Lending Club Borrower's Revolving Credit Utilization, Loan Purpose and Defaults

2013-02-04T07:30:02.010-08:00

While reviewing the loan volume with borrower's revolving credit utilization, I became curious to know how the revolving credit utilization of borrowers impact their reasons for borrowing on peer to peer platform. My initial thought was that borrowers with high revolving credit utilization most likely borrow for credit card refinancing purpose.Loan PurposeThe chart below shows the cumulative loan volume % by loan purpose as a function of borrower's revolving credit line utilization. The findings here don't surprise me. The percentage loan volume for credit card refinancing and debt consolidation purposes is much higher (steepest slope) for borrowers with high revolving credit line utilization. The borrowers with low revolving credit utilization are more likely to borrow for house buying, major purchase, and educational purposes.Lending Club Loan Volume by Purpose and Borrower's Revolving Credit Line UtilizationLoan StatusThe chart below shows the moving average of loan volume by loan status as a function of borrower's revolving credit line utilization. There is not much of a surprise here either. In general, the loan defaults and charged off rise with rising revolving credit line utilization of borrowers. Even though the volume of fully paid loans declines with rising revolving credit line utilization, the volume of fully paid loans appears to be somewhat constant for lower revolving credit line utilization.Lending Club Loan Volume by Status and Borrower's Revolving Credit UtilizationThe chart below is similar to the one above. In this chart, the revolving credit line utilization is divided into buckets. Each bucket (bin) is 10% wide. For example, the first bin includes all loans issued to borrowers who have revolving credit line utilization between 0 and 9.99%. The loans issued to borrowers who have revolving credit line utilization either below 10% or above 90% seem to default much more.Lending Club Loan Status and Borrower's Revolving Credit Line UtilizationHigher number of loans are fully paid off that were issued to borrowers with revolving credit line utilization below 20%. This observation leads to the question of whether borrowers with low revolving credit line utilization tend to pay off loans early.The chart below shows the loans that were charged off or fully paid for issued year 2009 through 2012 as function of revolving credit line utilization. The ratio of loans charged off to fully paid appears to be about 7 for borrowers with lower revolving credit line utilization, i.e. such loans are seven times more likely to be paid off early than charged off.Lending Club 2009-2012 Loan Status and Borrower's Revolving Credit Line UtilizationAnother observation worth highlighting is that unlike the earlier chart above, this chart doesn't show that defaults and charged off are higher for loans issued to borrowers with very low revolving credit utilization. The reason of discrepancy may be due to much higher loan volume in recent years that skews the default rate in the earlier chart.Key TakeawaysThe borrowers with high revolving credit line utilization are more likely to borrow on Lending Club platform for debt consolidation and credit card refinancing purposes.The default rate of loans rises with rising revolving credit line utilization of the borrowers. In contrast, the loan pay off rate declines with rising revolving credit line utilization of the borrowers.The borrowers with low revolving credit utilization are seven time more likely to pay off loan early than to default on the loan.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loan Volume and Borrower's Revolving Credit Line Utilization

2013-02-01T07:30:03.440-08:00

In next few posts, I will review the loan characteristics and default rate with respect to the revolving credit line utilization of borrowers. I believe revolving credit line utilization is one of the major borrower attribute that influences the chances for borrowers to default.Loan VolumeThe chart below shows the Loan Volume (right Y-axis) and Cumulative Loan Volume % (left Y-axis) as a function of borrower's revolving credit line utilization. Almost 20% of loans are issued to borrowers who have revolving credit utilization less than 29% and 20% of loans are issued to borrowers who have revolving credit utilization greater than 80%. The loan volume rises with rising revolving credit utilization up to about 70% revolving credit utilization. A few loans have also been issued to borrowers whose revolving credit utilization was greater than 100%.Lending Club Loan Volume and Borrower's Revolving Credit Line UtilizationThe chart below shows the Cumulative Loan Volume % by loan issued year as a function of borrower's revolving credit line utilization. Do you notice a wide gap between the lines for loans issued in 2012 from the lines for loans issued in prior years? This gap indicates that borrower profile based on revolving credit line utilization for loans issued in 2012 is very different from prior years. In 2011, 18% of loans were issued to borrowers who used up to 20% of their revolving credit. The share of loan volume to such borrowers dropped almost half to 9% in 2012. Similarly, about 30% of loans in 2011 were issued to borrowers who used up 70% or higher of available revolving credit. The share of loans volume to such borrowers rose about 20% to 36%.Lending Club Loan Volume by Issued Year and Borrower's Revolving Credit Line UtilizationThese trends may indicate that quality borrowers with low revolving credit utilization are not much interested in borrowing through peer to peer lending platform. Also, peer to peer lending platform being attractive to borrowers with higher revolving credit utilization may have resulted in Lending Club relaxing the minimum credit criteria late last year.Key TakeawayIf defaults and returns are closely related with borrower's revolving credit line utilization, I expect the loans issued in 2012 to behave very differently than the loans issued in prior years.Expanded Credit Utilization Information on PeerCubeThe Loan Details page on PeerCube contains additional information related to revolving credit line utilization that provides better context to lenders about borrower. For example, the screen capture below shows such information for a currently available loan that carries F2 credit grade. I typically gravitate toward reviewing information highlighted below. This borrower is carrying, on average, about $10,000 balance on each of his revolving accounts. All of his bankcards are maxed out and total credit balance exceeds $100,000.Loan Details page on PeerCube with expanded credit utilization information.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Recent Changes in Minimum Credit Criteria by Lending Club

2013-01-21T07:30:02.146-08:00

Minimum Credit CriteriaIn November, Lending Club not only hide the details of underwriting process, as described in my previous post Lending Club Loans - Impact of Recent Changes, but also changed the minimum credit criteria for the borrowers. The new minimum credit criteria is much more lenient and directed at attracting lower quality borrowers.The minimum credit criteria in the latest prospectus filed in November 2012 is listed as follows:Under the current credit policy, prospective borrower members must have among other elements:a minimum FICO score of 660 (as reported by a consumer reporting agency);a debt-to-income ratio (excluding mortgage) below 35%;minimum credit history of 36 months;6 or less inquiries in the last 6 months; andat least 2 revolving trade accounts.The minimum credit criteria in the previous prospectus filed in August 2012 is listed as follows:Under the current credit policy, prospective borrower members must have among other elements: a minimum FICO score of 660 (as reported by a consumer reporting agency);a debt-to-income ratio (excluding mortgage) below 35%;a credit report (as reported by a consumer reporting agency) without any current delinquencies, recent bankruptcy, tax liens or non-medical related collections opened within the last 12 months, and reflecting:at least two accounts currently open;for credit credits 740 and higher, no more than 8 credit inquiries on the credit report in the past six months and for credit scores below 740, no more than 3 inquiries on the credit report in the past six months;a revolving credit balance of less than $150,000;utilization of credit limit not exceeding 98%; anda minimum credit history of 36 months.The bold text above indicates the new condition added in the minimum credit criteria and the strikeout text above indicates the conditions removed. The new condition is nothing more than the lenient version of number of credit inquiries in the past six months.Public RecordsThere was some discussion on LendAcademy forum about Increase in Applicants With Public Records. The table below shows the monthly loan volume with respect to number of public records for loans issued in 2012. In November, 127 loans were issued to borrowers who had at least one public record. The number of such loans increased to 202 in December, an increase of almost 60% over previous month.I believe the changes in the minimum credit criteria as mentioned in November prospectus may have resulted in increase of borrowers with public records in December. No longer Lending Club excludes borrowers with current delinquencies, recent bankruptcies, tax liens, and non-medical related collections in past 12 months.Did Lending Club modify 'proprietary' credit grade model to accommodate this change in minimum credit criteria? I was expecting the credit grade shifting to the right toward grade G for such loans. The table below shows the monthly volume and credit grade of loans issued to borrowers with more than 2 public records. There is no shift in credit grade of loans to borrowers with public records. It doesn't appear that Lending Club credit grade model accounts for the number and type of public records.Accounts Now DelinquentThe table below shows the monthly loan volume  with respect to borrowers' number of accounts currently delinquent. It is clear from the table that prior to December, Lending Club didn't issue any loans to borrowers who had delinquent accounts at the time of applying for loan. In December 2012, Lending Club issued 15 loans to borrowers who had one or two accounts delinquent at the time of loan application.The credit grade for loans to borrowers with one account delinquent ranged from A5 to F1 while the one [...]



Lending Club Loans - Impact of Recent Changes

2013-01-14T07:30:02.970-08:00

Recently, in response to my post Lending Club 2012 in Review, Part I: Loan Volume and Amount Funded, a commentator pointed out a few recent changes related to Credit Grade in Lending Club's November 2012 prospectus and scarcity of F credit grade loans. Also, on LendAcademy forums, there were a few related discussions: High demand for D-G grades, Increase in Applicants With Public Record?, and Has LC loan quality dropped?. So, I decided to review the prospectus to see if I can find these changes.Below is a highlight from the November 2012 prospectus available at Lending Club website:Q: What are LendingClub loan grades?A: For borrower members who qualify, we assign one of 35 loan grades, from A1 through G5, to each loan request, based on the borrower member’s:FICO score; our proprietary scoring model which takes into account many of the attributes previously used by us and also allows borrowers to have delinquencies and public records; loan term and loan amountIn addition to replacing previously listed loan and borrower attributes with the proprietary scoring model, the current prospectus also mentions allowing borrowers with delinquencies and public records. Both terms are highlighted above in bold.The fewer low quality D-G grade loans may be result of the proprietary scoring model modifying weight of previously used loan and borrower attributes or incorporation of new 'unknown' attributes. Allowing borrowers with delinquencies and public records may be resulting in increase in borrowers with public record.Such changes are not surprising as Lending Club is gearing up for IPO. It's focus has been shifting to non-lending financial institutions as lenders and attracting more borrowers to its platform becoming higher priority.Can we confirm these changes using historical data from 2012?Credit Grade Distribution of Monthly Loan VolumeThe chart below shows the monthly loan volume in 2012 in relation to Credit Grade. The monthly volume lines for most months follow the same pattern except for December (Green line). There is a shift in the line for December from regular pattern for other months indicating there may have been a change in December how loans were allocated to different Credit Grade buckets.Lending Club Monthly Loan Volume and Credit Grade in 2012For most businesses, December tend to be a unique month due to holidays and end of calendar year. The chart below shows the monthly loan volume for 2011. The hypothesis being that if December is somehow unique month for share of loan volume across Credit Grade, it should show up in previous years too. As the chart shows, there is no deviation in line for December compared to the pattern for other months in 2011. This confirms that something changed in how credit grades are assigned for loans issued in December 2012.Lending Club Monthly Loan Volume and Credit Grade in 2011Monthly Loan Volume ChangeThe chart below shows the percentage change in loan volume for each credit grade in December from previous month in 2012. The large spreads in percentage change for credit grades that normally has low loan volume is understandable. The interesting observation is that the percentage volume change for credit grade D through G is consistently negative in the month of December, indicating the loan volume for credit D through G was much smaller in December than previous month.Lending Club Loans - Percentage Monthly Volume Change in December 2012As a comparison, the chart below shows the percentage change in loan volume in November from previous month in 2012. The loans with credit grade D through G show large percentage increase in volume from previous month. Was the volume drop in December for l[...]



Lending Club 2012 in Review. Part III: Loan Title, Loan Description and State of Residence

2013-01-10T07:30:01.870-08:00

Continuing the year-end review of Lending Club loans issued in 2012 from Part I and Part II where I discussed Loan Volume, Amount Funded, Interest Rate, Credit Grade and Loan Purpose ...Loan TitleIn 2012, the number of loans issued in relation to number of characters in Loan Title continue to follow similar pattern as previous years. Almost 25% of loans issued in 2012 had exactly 18 characters in loan title. Almost 50% of loans issued had less than 17 characters and 90% had less than 24 characters.Length of Lending Club Loan Title, Loan Volume, and Total Amount FundedThe most popular 18 characters loan titles in 2012 are listed below. Considering 77% of all loans in 2012 were issued for debt consolidation and credit card refinancing purposes, it is not surprising to see related phrases dominating the loan titles.Debt ConsolidationCredit Card PayoffConsolidation Loan2012 Consolidation / Consolidation 20122012 Personal LoanBill ConsolidationCard ConsolidationLoan DescriptionIn 2012, 38% loans issued had no loan description. 50% loans issued had loan description with less than 90 characters. 90% of loans issued had loan description with less than 332 characters.Length of Loan Description, Loan Volume, and Total Amount FundedThere was not much surprise with 53 character length of loan description, majority of them mentioned debt consolidation or similar variation and a few words. Similarly, 335 character length of loan description was primarily borrowers who wrote a complete sentence.State of ResidenceIn 2012, borrowers from 45 different states took out loans on Lending Club platform. Only borrowers from state of Iowa (IA), Idaho (ID), Maine (ME), Mississippi (MS), Nebraska (NE), North Dakota (ND), and Tennessee (TN) were absent. Borrowers from Indiana (IN) joined in 2012. Iowa (IA) dropped off in 2011 and Mississippi (MS) dropped off in 2012.The average amount per loan continue to rise for borrowers from all states with most increase from 2011 was for borrowers from state of New Mexico (NM), Vermont (VT), and Delaware (DE). The borrowers from Alaska (AK), Massachusetts (MA), and New Mexico (NM) borrowed the highest average amount per loan.Borrower's State of Residence and Average Amount FundedBorrower's State of Residence and Rise in Average Amount FundedOne change in 2012 from 2011 was that Texas bumped Florida to be in third place for the highest number of loans issued and total amount funded. California and New York continue to maintain first two spots. There were no significant changes in share of loans issued and amount funded to borrowers in each state in 2012 from 2011.Borrower's State of Residence, Loan Volume, and Total Amount FundedIn future blog posts, I may take a short break from reviewing characteristics of Lending Club loans in 2012 to discuss a few developments at PeerCube.Web Hosting - $6.95/mo!(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club 2012 in Review. Part II: Interest Rate, Credit Grade, and Loan Purpose

2013-01-04T07:30:02.071-08:00

Continuing the year-end review of Lending Club loans from my previous post where I discussed Loan Volume and Amount Funded ...Interest RateIn 2012, Lending Club increased the interest rates for most credit grade three times, in January, March, and July of 2012. By the end of 2012, the interest rate for loans ranged from 6.03% for credit grade A1 to 24.89% for credit grade G3, G4, and G5. I am happy to see Lending Club continuing to tweak interest rates and making loans more expensive for lowest quality borrowers. It will be great if in 2013 Lending Club can start issuing a summary of discussions from their Interest Rate pow-wow at least every quarter. I'm interested in knowing what made Lending Club to decide to change the rates.2012 Interest Rate and Credit Grade for Lending Club LoansAs seen from the loan volume chart in my previous post, raising interest rates doesn't seem to have much impact on loan demand. But is the slowdown in loan volume growth in second half of the year due to interest rate hike in July? The average interest rate for loans has risen about 13% from 12.49% in January to 14.12% in December. The average interest rate (13.64%) in 2012 was the highest compared to previous five years; it has increased over 10% from the average interest rate in 2011. I'm afraid that higher average interest rate may lead to higher default rates.Average Interest Rate for Lending Club Loans in 2012Average Interest Rate, year over year for Lending Club LoansThe average interest rate for loans with 36 month term and 60 month term in 2012 was 12.63% and 18.08% respectively. The rise in average interest rate for both loan terms was similar from previous year.Average Interest Rate for Lending Club Loans with 36 month and 60 month termsCredit GradeThe Credit Grade and its relationship with Loan Volume, Total Amount Funded, and Interest Rate has already been discussed in previous post and earlier in this post. The lower quality loans with credit grade E, F, and G are continuing to be dominated by loans with 60 month terms in 2012. Since 2010 when Lending Club first issued 60 month term loans, 23.85% of all loans have been issued with 60 month term. Year 2013 will be a pivotal year in better understanding the defaults behavior of 60 month term loans as such loans are reaching mid-way point in their maturity cycle. This will also offer better insights into lower quality loans as such loans recently have been primarily 60 month term loans.Credit Grade and Loan Term for Lending Club Loans, 2010 - 2012Loan PurposeIn 2012, Lending Club borrowers reported loan purpose as debt consolidation and credit card refinancing 77% of the time. Year over year, the percentage of loans with reported loan purpose of debt consolidation and credit card refinancing continues to rise. 77.15% of total amount raised in 2012 was used to fund loans with reported loan purpose of debt consolidation and credit card refinancing. This is an increase of almost 25% over 2011.At this growth rate, soon loan purpose attribute will become irrelevant as a selection criteria for loans. There is nothing stopping borrowers from claiming debt consolidation and credit card refinancing by running expenses for other loan purposes through credit card.Loan Purpose Reported by Lending Club Borrowers, 2010 - 2012.Loan Purpose and Amount Funded for Lending Club Loans, 2010 - 2012While there has been significant growth in volume and total amount funded for loans with reported purpose of debt consolidation and credit card refinancing, the pattern for average loan amount has stayed the same for past three years. The highest average loan amount are for[...]



Lending Club 2012 in Review, Part I: Loan Volume and Amount Funded

2013-01-02T07:30:00.232-08:00

Happy New Year to all my readers and to lenders on Lending Club platform. As Peter mentioned in his post In 2012 U.S. P2P Lenders Issue $871 Million in New Loans, Lending Club had terrific 2012.Loan VolumeIn 2012, Lending Club issued 53,367 loans, more than double the number of loans issued (21,721) in 2011. On monthly basis, in recent months the number of loans issued seem to be reaching plateau of little over 6,000 loans.Lending Club 2012 Monthly Loan VolumeIn 2012, over 90% of loans issued were of higher quality with credit grade of A, B, C, and D. Only 8.92% of loans (4,761 loans) were issued with credit grade E, F and G. In comparison, Lending Club issued 12.26% of loans with credit grade E, F, and G in 2011. Lending Club continues to show commitment toward improving listing of higher quality loans on its platform.Lending Club Loan Volume by Credit Grade 2010 - 2012As Lending Club announced in its blog post Investor Updates and Enhancements, beginning October 2012, LC started to reserve some loans for institutional and other investors who wanted to fully fund loans themselves. Since then percentage of loans offered initially as Whole on monthly basis is continuing to increase. In October 2012, only 13.51% of loans were initially offered as Whole and that ratio more than doubled to 29.10% by December. Overall in Quarter 4, 20.85% of loans were initially offered as Whole.I will recommend lenders, who would like to filter loans using initial list status as criteria, to consider using PeerCube filters. PeerCube now enables users to filter loans using 41 different loan and borrower attributes including Initial List Status.Lending Club 2012 Loan Volume by Initial List StatusAs I expressed concerns in my blog post Lending Club Loan Length and Default Rate about excessive 60 month term loans in 2011, it appears that the share of 60 month term loans in 2012 dropped to 18.55% of total loans issued, almost half of the 35.08% share in 2011.Lending Club Loan Volume by Loan Term, 2010 - 2012Loan AmountLending Club lenders funded $718 million in loans issued in 2012. This amount is over two-and-a-half times more than $257 million funded in 2011. As mentioned earlier, while the number of loans issued in last four months of 2012 stagnated, the amount funded continued to rise in the same period.Lending Club 2012 Loan Amount FundedThe stagnant number of loans but higher loan amount funded in last quarter of 2012 indicates that most likely the amount funded per loan was much higher than loans issued prior to last quarter. In fact, on average $14,100 was lent per loan in last quarter, almost 10% increase from previous quarter. This also reverses the declining trend in average amount funded per loan for past three quarters.Lending Club 2012 Average Amount Funded per LoanWhile only 8.92% of loans with credit grade E, F, and G were issued in 2012, such loans received 15.29% of the funding. The funding for such loans was slightly lower compared to 18.74% in 2011.Lending Club Total Amount Funded by Credit Grade, 2010 - 2012The small reduction in funding for lower quality E, F, and G grade loans doesn't match the significant reduction in number of such loans issued in 2012. The average amount per loan increased across all credit grades but the increase in average amount was much higher for loans with credit grade E, F, and G.Lending Club Average Amount per Loan by Credit Grade, 2010 - 2012Going forward, I am particularly interested in monitoring Initial Listing Status and how it impacts retail lenders like myself. I was surprised to see that the percentage of total amount funded fo[...]



Lending Club Borrower's Home Ownership and Loan Characteristics, Part II

2012-12-21T07:30:02.559-08:00

Continuing the review of loan characteristics as a function of home ownership status of borrower from previous post ...Interest RateThe chart below shows the average interest rate of loans issued between 2007 and 2012 YTD as a function of borrower's home ownership status. Due to very small volume of loans issued to borrowers who claimed home ownership as Any or None, those loans are removed from this chart. The average interest rate is an arithmetic equal-weighted average and not loan amount-weighted average.The average interest rate is slightly higher for borrowers who rent homes compared to their counterparts who carry mortgage on their homes. This pattern has been quite consistent year over year. The reason for the slightly higher average interest rate may be due to lower FICO scores and shorter credit history because borrowers who rent home are likely to be younger and with limited credit history.Similarly, the average interest rate is slightly lower for borrowers who own homes compared to borrowers who carry mortgage, loans issued in 2009 being the exception. The reason may be due to long credit history and higher FICO scores because borrowers who outright own homes are likely to be older and with decent credit history.Lending Club Borrower's Home Ownership and Average Interest RateTo confirm the assumptions, I decided to chart the average credit age at the time of issue of loan with the home ownership status of borrowers as shown below. Sure enough, the average credit age of borrowers who have mortgage on their homes is significantly and consistently greater than their counterparts who rent home. For example, in 2012 YTD, the credit age for borrowers who have mortgage is 15.8 years versus 12.6 years for their counterparts who rent. But there is no such relation between the borrowers who outright own their homes and credit age.Lending Club Borrower's Home Ownership and Average Credit AgeNext I decided to chart the composition of home ownership status for each FICO score range of borrowers. It's clear that the lower FICO range are dominated by borrowers who rent and higher FICO score range by borrowers who have mortgage. For example, 56.1% of borrowers with FICO score between 660 and 664 rent their homes while only 19.75% of borrowers with FICO score between 800 and 804 rent their homes.Lending Club Borrower's Home Ownership and FICO RangeBoth Credit Age and FICO Range appear to offer a reasonable explanation for average interest rate to be slightly higher for borrowers who are renters. It is also not surprising considering that FICO Range is a key component of calculations used by Lending Club in setting Interest Rate for a loan.Loan LengthThe chart below shows the percentage of 36 months and 60 months loans issued between 2010 and 2012 YTD to borrowers as a function of their home ownership status. Prior to 2010, Lending Club issued loans with 36 months term only thus the data prior to 2010 is excluded from this analysis.The borrowers who carry mortgage on their home are more likely to request loans with longer terms compared to their counterparts who own outright or rent their homes. In 2012 YTD, 23.60% of borrowers with mortgage received loans of 60 months term. In comparison, only 13.4% of borrowers who rent home received loans of 60 months term. This pattern is very consistent since Lending Club started issuing loans with both 36 and 60 months term.A likely reason could be the management of debt repayment load. The 60 months loan will have lower monthly payment compared to 36 months loan of the same amount and interest rate. Typi[...]



Lending Club Borrower's Home Ownership and Loan Characteristics, Part I

2012-12-18T07:30:01.014-08:00

In this post, I will review the home ownership status of borrowers and loans issued at Lending Club platform. Lending Club categorizes the home ownership status of borrowers as Rent, Mortgage, Own, None, and Any. While Rent, Mortgage and Own are self-descriptive, I am not so sure what None and Any categories represent and how they differ from each other. For the purpose of analysis, I combined the None and Any categories together.Over the past year, I have come across multiple loan listings where a borrower indicated home ownership status of Own but had one or more mortgage accounts, loan description or Q&A mentioned having mortgage. I believe some borrowers may be confusing the Mortgage and Own categories.Loan VolumeThe chart below shows the percentage of loans issued in each application year to borrowers with each of the four categories of home ownership status: Any or None, Own, Mortgage, and Rent.Lending Club Borrower's Home Ownership Status and Loan VolumeThe combined category of Any or None accounted for less than 0.002% of loan issued in 2012 YTD and 0.004% in 2011. The highest percentage of loan issued to borrowers with home ownership status of Any or None was 1.90% in 2008. Between 2007 and 2012, only 102 loans were issued to borrowers with home ownership status of Any or None.Majority of loans are issued to borrowers who declared either renting home (on average 50% from 2007 to 2012) or having mortgage on home (on average 42%). Lenders excluding any one of these two categories in their loan selection criteria are ignoring almost half of the loans available on Lending Club platform.With the exception of 2012 YTD, the percentage of borrowers who rent home is declining while borrowers who has mortgage on their home is rising. Considering we went through a deep recession, primarily due to Real Estate bubble, the rise in borrowers who have mortgage on their homes is not surprising. The 2012 may be start of reversal in these trends.Loan Amount FundedThe chart below shows the percentage of total loan amount funded in each application year to borrowers with different home ownership status.Lending Club Borrower's Home Ownership Status and Total Loan Amount FundedWhat is interesting about this chart is that the percentage of loan amount funded for borrowers who rent home is lower than the percentage of loan volume for such borrowers as shown in the previous chart. For example, in 2012 YTD, borrowers who rent home received 47% of total loans but only 42% of total loan amount. The trend is reverse for borrowers who have mortgage on their home. This observation seems to indicate that the borrowers who have mortgage seem to request higher loan amount than the borrowers who rent home. These observations are also confirmed by the chart below that shows the average loan amount funded for borrowers with different home ownership status.Lending Club Borrower's Home Ownership Status and Average Loan Amount FundedAs the chart shows, in 2012 YTD, borrowers who have mortgage on their home borrowed almost $3,000 (~20%) more than their counterparts who rent home. This observation is perplexing as why a borrower who already have a mortgage (supposedly much larger than any other debt) would request loans for large amount. Only logical explanations I could come up with are that most borrowers don't consider mortgage same as other debt, they are more comfortable carrying additional debt, and they are likely to borrow for large value home improvement projects.To confirm whether borrowers who have mortgage borrow large amoun[...]



Lending Club Borrower's Credit Age and Loan Defaults

2012-12-10T07:30:02.155-08:00

In this post, I continue the analysis of borrower's earliest credit line and relationship with loan status. As Lending Club doesn't publish the age of borrower, in this analysis I use the year of earliest credit line, termed credit age, as proxy for borrower's age. I am particularly interested in finding out whether:The finding from past consumer finance research of older borrowers being higher credit risk holds true for peer-to-peer lending platforms, andThe borrowers with recent earliest credit line are higher credit risk, as asserted by White Coat Investor in his blog post Peer to Peer Lending Club Update.Personally, I believe that Lending Club borrowers with earliest credit line several decades old likely to be much higher credit risk. Such borrowers in need for loan with high interest rate are likely to be not very savvy in managing their finances.Earliest Credit Line and Loan StatusThe chart below shows the loan status as a function of the year of earliest credit line. The loans issued to borrowers who started their credit in early 70's or earlier seem to have higher charged off, defaults and late payments. With 40+ years of credit history, these borrowers are likely to be in their late 50's and early 60's.It may appear from the chart that loans issued to young borrowers who started their credit history in 2007 and later seem to have lower charged off and later payments. Considering that the fully paid loans are also lower for such borrowers, I believe most of these loans were recently issued. Even though I didn't find any explicit statement, I believe Lending Club doesn't issue loans to borrowers who have less than three years of credit history.The chart below shows the charged off and default loan status as a function of the year of earliest credit line by loan application year 2008 through 2010. The loans issued to borrowers that started their credit history in 1970's or earlier appear to have consistently higher charged off and default status. There is no such pattern for borrowers with recently established credit history.Credit Age and Loan StatusThe chart below shows the loan status as a function of credit age. The pattern of loans issued to borrowers with longest credit history (40+ years) with higher charged off and default loan status, observed in first chart appears to hold. Also, loans issued to borrowers with credit history less than four year old appears to have higher charged off and default loan status.The chart below shows the charged off and default status as a function of credit for application year 2008 through 2010. While borrowers with long credit history continue to show higher tendency to have loans charged off and defaults, there is no such consistent pattern for borrowers with short credit history.Key TakeawaysThe borrowers with long credit history (40+ years) tend to have more loans charged off or defaulted. Risk averse lenders may benefit by avoiding older borrowers.There is no consistent patterns of charged off and defaults with younger borrowers. Lenders may consider taking a cautious approach toward borrowers who have less than four years of credit history.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loan Borrower's Earliest Credit Line and Loan Volume

2012-12-05T07:30:03.197-08:00

After returning from vacation, I have started to analyze Lending Club historical loan data again and have improved PeerCube further. I am pleased to see PeerCube user base growing another 20% in my absence. Peter was kind enough to set up a section for PeerCube on Lend Academy forum. I'd like to encourage you to use the forum to suggest enhancements, request new features, and report any issues on PeerCube.In a few posts, I will analyze a correlation between borrower's earliest credit line and credit risk. The past consumer finance research indicates that age of the borrower is significantly related to credit risk. As Lending Club doesn't publish the age of the borrowers, I have been using Earliest Credit Line as a proxy for age in my loan selection at Lending Club.I am particularly interested in finding out whether the year of earliest credit line moves with loan application year as that would indicate a particular age group of borrowers more likely to use peer to peer lending for their credit needs.  My impression is that most people start their credit in late teens and typically get into credit problems in late 20's or early 30's if they don't manage money properly. I expect that most borrowers on Lending Club to have their earliest credit line opened 10 to 15 years ago.The chart below shows the loan volume as a function of the year of earliest credit line. The surprising pattern is that the borrowers with first credit line established in year 2000 appears to have most loans issued, year after year. The breadth of years when first credit line was established for which most loans were issued seem to be narrowing with application year.I wanted to see whether the distribution of loans by borrowers' earliest credit line is changing year over year. The chart below shows the distribution of loans by borrowers' earliest credit line for each application year. The percentage distribution for each application year appears to track closely. While the previous chart indicates that borrowers who established credit line in 2000 have the biggest share of loans, this chart indicates that the percentage of loans issued to borrowers with earliest credit line established in 2000 is rising with each application year and shifted from 1999 and 1996 in earlier application years.I also calculated the Credit Age in years, the number of days between the earliest credit line date and application date. The chart below shows the loan volume as a function of credit age. The chart is not that different from the first one above. Even though no additional insights I gained from this chart, I expect the credit age to be handy when I analyze loan defaults.In the next post, I will analyze loan defaults with respect to earliest credit line and credit age.(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]



Lending Club Loan Selection using PeerCube's BLE Risk Index

2012-11-08T07:30:03.644-08:00

One lesson I learnt from Peter's review of PeerCube and subsequent discussions is that I need to share how PeerCube can be used effectively during loan selection at Lending Club. In this post, I would like to provide some basic examples of strategies with BLE Risk Index.As a refresher, BLE Risk Index of 1.0 is considered average risk. The lower the number, the lower the risk and the higher the number, the higher the risk.Strategy #1: Loans with BLE Index below a specified valueThe easiest way to use the BLE Risk Index during loan selection process might be to select loans on the basis of BLE Risk Index below a specified value. Very conservative lenders may consider selecting loans with BLE Risk Index below 0.90, while moderately conservative lenders may select loans with BLE Risk Index below 1.10.PeerCube makes this strategy easier by listing loans with lowest BLE Risk Index under menu Lending Club -> Loan Review -> Low BLE Risk Loans. As the screen capture shows below, most available loans with BLE Risk Index below 0.90 are likely to be Grade A and B loans with Interest Rate as high as 12.12%. The moderately conservative lender can pick loans up to Grade D with interest rate as high as 19.72% while keeping the BLE Risk Index below 1.10.Selecting loans solely on the basis of BLE Risk Index is the quickest way to make lending decisions. But in my opinion, this strategy may not be appropriate for most lenders because it puts too much faith on the current applicability of methodology used in a 1940 research and PeerCube's ability to implement the methodology correctly after 70 years.At present, both concerns are valid ones and can't be ignored before adopting this strategy. The additional questions may arise about the "right" BLE Risk Index number for a lender and foregoing additional returns possible by increasing the threshold value, as previously discussed as part of BLE Index Caveats.Strategy #2: Loans with lowest BLE Index in Each Credit GradeThis strategy is slight variation of previous strategy. Instead of relying on absolute specific value of BLE Risk Index, it relies on the relative risk as represented by BLE Risk Index.One way to use relative risk is choosing loans with the lowest BLE Risk Index within a Credit Grade. This strategy will enable lenders to capture higher interest rate offered by higher credit grade while minimizing the BLE Risk Index within that credit grade.We plan to utilize similar strategy in our new account with Lending Club. Our lending strategy with new account will be to:Invest in total seven loans per week.Invest in loans from all credit grade per week.Invest in one loan per day.Invest in loan with the lowest BLE Risk Index in each credit grade.Not review loan details or any other quantitative and qualitative criteria.PeerCube makes implementing this strategy a breeze by listing loans with the lowest BLE Risk Index for each Credit Grade under menu Lending Club -> Loan Review -> Low BLE Risk Loans. The concerns with this strategy are very similar to the concerns with the previous strategy. By not considering a fixed value of BLE Risk Index, the strategy minimizes the impact of errors in BLE Risk Index calculations as such errors will manifest for all loans.Strategy #3: BLE Index as additional criteria with Filtered Loan ResultsI believe this strategy is the most prudent among the three strategies discussed in this post. With this strategy, a lender considers BLE Risk Ind[...]



PeerCube Update - Comprehensive Details for Lending Club Loans

2012-11-02T07:30:02.300-07:00

Last week, Peter wrote an excellent review of PeerCube on his blog. The number of PeerCube users has greatly increased since his coverage including interest from individual and institutional investors. The major focus of users has been on PeerCube Loan Filter and Peer Filters. Users are becoming more comfortable with sharing and using Peer Filters which have been used almost 250 times. Three users, excluding myself, have shared their filters with the broader community. I also posted half-a-dozen curated peer filters.Peer Insights functionality continues to be under-utilized. Originally, I saw peer insights features to be natural extension of crowdfunding nature of peer-to-peer lending. But as Peter suggested, quick churning of loans may be preventing users from rating and discussing loans. Any suggestions for increasing traction of peer insights are most welcome.New Features on PeerCubeIn last couple of weeks, I also learnt that I need to frequently update users on changes and improvements made to PeerCube and discuss ways to use PeerCube effectively. I have been quietly rolling out new functionalities and changes almost every week or two on PeerCube. Going forward, I will start writing blog posts to introduce new features and usage.Peter already introduced Bad Loan Experience (BLE) Risk Index that is now integrated on PeerCube. For past few weeks, I have been discussing on this blog the BLE Index according to various loan parameters (Loan Purpose, Loan Length, Interest Rate, Loan Amount, Credit Grade, Borrower's Location, and FICO Scores). In future posts, I plan to discuss more about the foundation underlying BLE Risk Index and how lenders can use it more effectively during loan selection process.In this post, I would like to introduce another change on PeerCube that I rolled out this week. Now the loan detail page on PeerCube includes comprehensive details about the loan and borrower. Except for the Q&A, almost all other available information is included on PeerCube loan detail page.I am not aware of any place online including Lending Club's own loan detail page where as much information about a loan is available as on PeerCube's loan detail page.Frequently Updated Funding ProgressUpon user's request, PeerCube does not display most 100% funded loans anymore. PeerCube has taken further steps in displaying more up-to-date funding progress. During business hours, funding progress is updated every four hours that results in reducing further the chances that a user will encounter fully funded loans during loan selection using PeerCube.Expanded Loan Request SummaryThe Loan Request Summary is significantly expanded with Initial List Status, Review Status, and various dates such as when Lending Club reviewed the application, pulled the credit report and borrower acceptance of the loan terms. The changes are highlighted below.Comprehensive Borrower DetailsThe Borrower Details section now includes lot more information about borrower's employment and items from borrower's credit report. The changes are highlighted below. Thank YouOverall I am very impressed with the progress PeerCube has made in last two months since introduction. I thank users of PeerCube who continue to engage with me and give suggestions and feedback. I have been enjoying all interactions and they stimulate ideas.Thank you to all users of PeerCube and thank you for following this blog. I'm looking forward to coming up [...]



Lending Club Loans - Borrowers' FICO Score and Bad-Loan Experience Index

2012-10-29T07:30:00.915-07:00

After reviewing the defaults as a function of the borrowers' FICO score in the previous post, I decide to review FICO range according to bad-loan experience (BLE) index. Please refer to my first post on BLE Index for background information.The table below shows the BLE Index according to borrowers' FICO score range. The range of BLE Index from 0.23 for FICO score 790-794 to 1.84 for FICO score 660-664 is not considerably wide. It appears that changes in FICO score is not significantly related to credit risk. The borrowers with FICO score below 709 appear to be greater credit risk.Are you interested in Retirement Planning and Investment Annuities? Learn more from the published works of Moshe MilevskyThe table below shows the BLE Index according to FICO score range for loans issued each year from 2007 to 2012. Except for loans issued in 2012, the pattern of borrowers with FICO score below 709 being greater credit risk is pretty consistent. For loans issued in 2012, even the borrowers with better FICO score have BLE Index greater than 1.10. In my opinion, this observation shows that borrowers with better FICO score are more likely to default during the first year of the loan.Key TakeawaysNo surprises that borrowers with lower FICO score are greater credit risk.The lenders who purchase notes on secondary market may be better of purchasing notes that have aged at least a year. frameborder="0" marginheight="0" marginwidth="0" scrolling="no" src="http://rcm.amazon.com/e/cm?lt1=_blank&bc1=000000&IS2=1&nou=1&bg1=FFFFFF&fc1=000000&lc1=0000FF&t=randothoug075-20&o=1&p=8&l=as4&m=amazon&f=ifr&ref=ss_til&asins=1118291530" style="height: 240px; width: 120px;"> frameborder="0" marginheight="0" marginwidth="0" scrolling="no" src="http://rcm.amazon.com/e/cm?lt1=_blank&bc1=000000&IS2=1&nou=1&bg1=FFFFFF&fc1=000000&lc1=0000FF&t=randothoug075-20&o=1&p=8&l=as4&m=amazon&f=ifr&ref=ss_til&asins=0470680997" style="height: 240px; width: 120px;"> frameborder="0" marginheight="0" marginwidth="0" scrolling="no" src="http://rcm.amazon.com/e/cm?lt1=_blank&bc1=000000&IS2=1&nou=1&bg1=FFFFFF&fc1=000000&lc1=0000FF&t=randothoug075-20&o=1&p=8&l=as4&m=amazon&f=ifr&ref=ss_til&asins=0521148030" style="height: 240px; width: 120px;"> frameborder="0" marginheight="0" marginwidth="0" scrolling="no" src="http://rcm.amazon.com/e/cm?lt1=_blank&bc1=000000&IS2=1&nou=1&bg1=FFFFFF&fc1=000000&lc1=0000FF&t=randothoug075-20&o=1&p=8&l=as4&m=amazon&f=ifr&ref=ss_til&asins=0137127375" style="height: 240px; width: 120px;">(c) 2014 Anil Gupta. Published at http://andirog.blogspot.com. I do not receive compensation from discussed vendors and advertisers unless a reader sign up or purchase by clicking through the banners and links provided.[...]