6 Factors that May Help Predict Prepayment

New Call-to-action
     

In a recent post (Loan Prepayments and the Magic of Event Prediction), we discussed survival analysis and how it can be useful in predicting loan lifetimes. We’re excited to now share some of our initial findings with you on this topic! Don’t worry, we won’t get all mathy on you this time; we’ll instead focus on sharing the factors that we found to have the strongest relationship with loan prepayment: loan-to-value ratio, debt-to-income ratio, unpaid balance, unemployment rates, house price indices, and interest rate gap.

moneybag.jpg

 

  1. Loan-to-value ratio (LTV): Visible Equity calculates real-time market values on all real-estate collateral. Real-time collateral values are used to calculate real-time loan-to-value ratios. This variable captures two important economic factors. First, it captures the impact of borrower leverage on loan performance. Second, the variable captures macroeconomic meaningful economic changes in the housing market. The results of our model suggest that an increase of 17% in LTV will, on average, decrease the probability of prepayment in the first three years of loan life by 1.49%.

  2. Debt-to-income ratio (DTI):
    This variable provides the percentage of a consumer’s monthly gross income that contributes to paying debts. There are two main aspects of DTI, the first being the front-end ratio which indicates the percentage of income that goes toward housing costs. The second, or the back-end ratio includes other debts such as credit card payments, car loan payments, student loan payments, etc.. As you might suspect, we find that borrowers with higher DTIs are less likely to prepay. Specifically, when DTI increases 12%, the probability of prepayment in the first three years decreases, on average, by 1.61%.

  3. Unpaid balance:
    This variable captures the current total unpaid balance for any specific loan. It is positively associated with prepayment such that an increase of $9800 in unpaid balance decreases the probability of prepayment in the first three years by 0.39%.

  4. Unemployment (MSA-specific):
    The Bureau of Labor Statistics reports MSA-specific unemployment rates each month. We find that when unemployment increases by 2%, the three-year probability of prepayment decreases by 0.52%.

  5. House price indices (MSA-specific):
    The Federal Housing Finance Agency reports MSA-specific house prices each month. The HPI is a weighted, repeat-sales index, meaning that it measures average price changes in repeat sales or multiple refinancing on the same properties. The index has a positive relationship with prepayments. Specifically, when the HPI increases by 7.8%, the three-year probability of prepayment increases by 6.0%.

  6. Interest rate gap:
    This variable calculates the difference between the real-time, loan-specific interest rate and the prevailing 30-year fixed mortgage rate in the market. This variable captures two important economic factors. First, it captures the impact of macroeconomic factors influencing interest rates in general. Second, it captures the propensity of borrowers to prepay on account of paying interest rates lower than the prevailing market rate. Interest rate gap has shown to be the most influential factor on prepayments. Borrowers with interest rates that are lower than prevailing market rates demonstrate a higher propensity to prepay. As the interest rate gap widens by 1%, the three-year probability of prepayment increases 14.33%. For example, a borrower’s current mortgage rate is 6% while the national mortgage rate is 6%. If the national rate drops to 5%, then the estimated probability of prepayment increases by 14.33% in the first three years.

 

Most of these results probably align well with your intuition and expertise about the industry, but it’s valuable to not only see data-based evidence of these relationships, but to actually see them quantified.


Rachel Messick

Product Manager/Data Scientist at Visible Equity


COMMENTS