Credit Risk: Vintage Analysis

This tutorial explains the concept of vintage analysis and how it is used in banking. Vintage analysis is also called 'cohort' analysis.

Introduction to Vintage Analysis

In credit risk, it is a popular method for managing credit risk. The term 'Vintage' refers to the month or quarter in which account was opened (loan was granted). In simple words, the vintage analysis measures the performance of a portfolio in different periods of time after the loan (or credit card) was granted. Performance can be measured in the form of cumulative charge-off rate, proportion of customers 30/60/90 days past due (DPD), utilization ratio, average balance etc.

The vintage analysis is used for a variety of purposes. Some of them are listed below -

  • Identify if accounts opened in a particular month or quarter are riskier than others
  • Determine the optimal period of performance window in development of scorecard
  • Monitor or track risk of a portfolio
  • Estimate minimum months required after that we can cross-sell to new customers
  • Forecast risk
  • Can be used in stress testing

How Vintage Analysis is used in credit risk modeling?

It is used to determine the number of months' data you should consider for performance window. If customer defaults (90 days or more past due) during the performance window, borrower would be considered as a 'bad' customer and labeled as 'event' in dependent variable.

To do this process, we first need to take multiple periods. Let's take 6 different time periods for demonstration -

  1. 36 months window starting from Q1,2016. Q1,2016 is the period when accounts were opened and we would see their performance in the next 36 months.
  2. 30 months window starting from Q3,2016. Why 30 months? For accounts opened in 3rd quarter of 2016, we only have 30 months of data.
  3. 24 months window starting from Q1,2017
  4. 18 months window starting from Q3,2017
  5. 12 months window starting from Q1,2018
  6. 6 months window starting from Q3,2018

Next step is to calculate the cumulative % of bad customers or cumulative charge-off rate (also known as cumulative loss rate) against months on books (MOB) which is the number of months that have completed since the loan origination Date. New Vintage means recently opened accounts.
Let's understand the table above
In the first row, it shows 1.1% of customers who opened their account in first quarter of 2016 were 90 days past due after 15 months. Similarly, 2.5% of accounts opened in the first quarter of 2017 turned out to be bad customer after 21 months.
For graphical representation, let's plot this data in line chart and see the growth trend of bad rate.
The chart above shows the bad rate growing at a fast pace before 24 months and then stabilize after 24 months. Hence we can use 24 months as a time period for performance window.
Trend Analysis
We can also analyse trend using Vintage Analysis. For example, average balance for customers opened in different quarters and see their trend in the subsequent months after account opening date. In the charts shown below, we are showing vintage analysis of average balance and utilization ratio. The variable reaches maximum in 6-7 months and after that it starts decreasing. The charts shown below are for demonstration purpose. It's not related to the chart shown in the previous section.
Related Posts
About Author:

Deepanshu founded ListenData with a simple objective - Make analytics easy to understand and follow. He has over 10 years of experience in data science. During his tenure, he has worked with global clients in various domains like Banking, Insurance, Private Equity, Telecom and Human Resource.

10 Responses to "Credit Risk: Vintage Analysis"
  1. Hi Deepanshu,

    The contents are very good.If you Could add the SAS code for credit risk model,vintage analysis,roll rate analysis etc.. then it would be really a great help to everyone and i really appreciate for that.

    Thanks & Regards,

  2. Hi Deepanshu, i am regular follower of your website. This is really good content for learning. Please add some data set and sas code so that we can see the output as well. This will be really helpful.

  3. Dear sir ,

    Please share the python codes for the cohort table and also the trend analysis.

    Beat regards ,
    Somenath Ghosh.

  4. Depanshu thanks for such a wonderful blog I would like you to write about general analytic insights which one can definitely use in common to any industry.

  5. Hi Deepanshu,

    Very nice explantion, it would be very helpful if you can share the SAS code for Credit risk modelling, Rolerate and vintage anaysis.


  6. This comment has been removed by a blog administrator.

  7. Hi Deepanshu,

    Thanks for making this useful website! I've learned a lot.

    I am working in a bank ,too.
    And in realistic, I sometimes finds that the vintage of some products never be stabilized, it is still growing. Although the maximal MoB is over 36 months.
    Do you know why?
    And in that case, how can we decide the best performance window?

    I will be very appreciate if you answer my questions. I've been troubled for weeks.


    1. Hi Samantha,

      I work in Credit Bureau and we see this pattern all the time.
      In this case we check Rate of change of bads. Suppose when you plot graph and Cum bads are increasing continuously and never the curve is flatten out then we check %change per month basis and we have seen after some specific time like 12 months or so that %change is minimal and that is where we make the cut for performance window.

  8. Thank you so much for this information. Any reason why we don't consider Q2 in 2016 or 2017 as my base but only Q1 and Q3?

  9. Hi,
    I'd like to compare the performance in a particular month on book of the loan granted in difference year how can I calculated? Should the closed account at that particular month be considered or only active account? for example 100 accounts booked in Year 2019 and 10 accounts closed in mob 10 and 3 accounts was DQ30, what is the formula for %DQ30 at mob 10? ==>3/90 or 3/100


Next → ← Prev
Love this Post? Spread the Word!