CPR Report: SBA 7(a) Prepays “Skyrocket” Due to Principal Repayment Changes

November 29, 2017

CPR Report: SBA 7(a) Prepays “Skyrocket” Due to Principal Repayment Changes

By Bob Judge
Editor, CPR Report

Download the CPR Report, November 2017

Quite a month it has been. A change to how the SBA and Colson will pass-through principal payments on pools originated from FY 2005 to FY 2017 has had the effect of significantly speeding up prepayment speeds during its first two months of implementation (Please refer to the Glossary for more information about this change).

For this reason, I have combined the October and November reports to discuss this change and its impact so far.

For a more detailed analysis of the change and what it means, please see the companion article below.

Through this past September, we had been running at an average of CPR 8.43% for the first nine months of 2017. This represented an increase from 2016, which showed an average CPR of 7.85% for the entire year.

So, it is safe to say that 7a loan prepayment speeds have been slowly rising, and were currently in the mid-8s as of the end of FY 2017 (The SBA fiscal year ends of 9/30). These measurements were based on the “old” way of distributing prepaid principal, which goes back to the beginning of FY 2005.

Basically, the SBA went from distributing loan proceeds and the loan’s own excess principal within FY 2005—FY 20017 pools to distributing loan proceeds and pool excess principal on a prorata basis across the remaining loans inside those same pools.

On how excess principal grows inside a pool, please refer to the article “The SBA Master Reserve Fund: Is SBA Pooling in Trouble?” from the September, 2015 CPR Report.

While it would seem that this change would have only modest consequences to the return of excess principal, it in fact has had a significant impact in October and November.

For the remainder of this article, we will go over the actual results, provide some insight into how much excess was paid out and calculate a “corrected” prepayment speed that strips out said excess. The companion article in this report will go into further detail on what the future holds based on what we currently know.

Resuming the discussion, the reality is that prepayment speeds in the FY2005 to FY 2017 pools have moved up considerably, pushing overall speeds much higher, since this group of pools currently represents 98.5% of all outstanding pool balances.

For those who are attempting to calculate a base-line prepay speed for 7a pools, we offer our version of a corrected speed that allows for an “apple-to-apples” comparison to past historical numbers. I leave it to the reader to judge the value of the results.

For October, actual speeds came in at 14.55%, an 82% increase over September’s reading of 7.99%. November was a mirror of October, coming in at 14.54%.

As you would imagine, most maturity buckets witnessed steep increases, except, curiously, the 8-10 sector fell 19% (CPR 15.20% to CPR 12.28%) in October. It also declined by another 12% in November, coming in at CPR 10.75%.

Enough about current prepay speeds. Suffice to say they will remain elevated for an extended period of time, with the “Why” to be answered in the companion piece. Instead, let’s turn to our model for stripping out the excess principal and calculate a pre-change CPR for each maturity bucket, as well as the overall.

Fortunately for this analysis, there is a file on the Colson website that shows all secondary market defaults and voluntary prepayments pre-MRF, of which pooled loans are a significant subset. As you might imagine, these two datasets are highly correlated, allowing us to predict what the prepayments might look like using the same rules that were in place prior to October.

A simple regression of prepayment amounts between the paidoff loan file and the factor report over the past two years produces the results for the above chart, “POOL PREPAY PREDICTION RESULTS”.

As you can see, the predictive power pre-change was very high. In fact, it produced a r-squared of 95.3% for the period of 1/2015 through 9/2017 (for more details, see the table on the next page).

By this measurement, $183 million and $174 million of excess principal was distributed from the MRF in the October and November factor reports, respectively. Subtracting these amounts lowered the “corrected” overall CPR to 8.68% and 8.96% for both months.

You will notice in the CPR charts further into the report that I have added a corrected CPR chart, in addition to the actual prepayment speeds you are used to seeing. Again, actual prepayment speeds for both months were in the 14s, but for those who need corrected numbers for modeling purposes, I offer the corrected speeds.

The next step was to distribute the excess principal over the six maturity buckets that we track. In order to get to the appropriate proportions, I modeled the excess distribution at the pool level, based on the amount of excess principal that is imbedded in each pool. This was possible by running balances on all 64,898 loans that were still active as of October 31st and November 13th which was the date of the November Factor Report. We then totaled the loan balances by pool and compared it to the actual pool balances for both October and November factor reports.

At this stage, this is not an exact comparison, due to the delay imbedded in the factor reports versus up-to-date loan balances available on the Colson website. The October loan balances will match up with the December factor report, which is an important distinction when we discuss modeling the future in the companion piece. Suffice to say that our modeling will get better as we continue to track loan balances versus pool balances over time.

The results of this distribution of the paidout excess by maturity bucket is also in the corrected CPR table, so we can get a better handle on how each bucket is currently prepaying. We did not distribute the excess by age, as seen in those tables, leaving them with the actual numbers for this report.

As we get better data and sharpen our pencils, we should be able to provide this type of distribution in future issues of the report. Fortunately, we do have results from previous months, that will allow those individuals looking to make their judgement as to how the excess payments have impacted the pool age tables.

Announcing the 10th Annual 2017 SBA Secondary Market Summit
Join this high-level gathering of public and private stakeholders to review strategies for capitalizing on emerging opportunities and avoiding new threats and pitfalls.

One Day: Tuesday, December 5, 2017
Blank Rome, LLP
Washington D.C., 20006