Over the last couple of months, the yield curve has fluctuated quite a bit. From mid-July to mid-August, it rallied hard, with the 10-year dropping a full 55 bps and the belly of the curve moving almost as much, pricing in a significant economic slowdown. Since then, on the heels of firmer growth and inflation numbers, coupled with slightly underwhelming Quantitative Easing in the European Union relative to expectations, it has backed up, reversing roughly 40% of the drop. After the Fed meeting this week, incoming wage, inflation and growth data promise to be highly scrutinized.
Against this backdrop, mortgage durations and how well model durations match up against empirical durations are a natural concern. In our first article, Andy, Feiqi, and Alex compared option-adjusted durations using both our base prepayment model and a prepay-risk neutralized model against empirical duration across a range of coupons. They find that in lower coupons, both duration measures perform reasonably well but that risk-neutral durations tend to do better across the board. This is not surprising since a risk-neutral view would tend to match widening in rallies due to increases in prepay model risk.
Our second article is the debut of Chris Widman, who recently joined our behavioral modeling group. Chris takes a look at the auto-loan market, which is the second largest asset-backed consumer loan sector, and behind only student loans in the non-mortgage consumer loan sector. He reviews the macro trends of the sector and provides an introduction to key features of individual loans worth knowing about. We look forward to more of his articles in this sector as we move towards the introduction of an auto loan model.
EMPIRICAL DURATIONS AND HEDGE RATIOS: CAN AD&CO’S MODELS EXPLAIN THEM?
by Andrew Davidson, Feiqi Jiang and Alex Levin
Agency MBS investors frequently ask: “Can your model match the market’s durations?” We computed empirical durations for a range of TBAs along with TBA hedge ratios and observed them for the past 12 months. We then demonstrate that our OAS model, operating with a risk-neutral version of the LoanDynamics Model (LDM, formerly “prepayment model”), resembles empirical measures of risk rather well.
According to the Federal Reserve of New York, outstanding auto loan balances have nearly doubled since 2010 growing from $0.7 trillion to over $1.2 trillion by Q1 2019. Auto loans are the second largest non-housing consumer debt behind only student loans.
At Andrew Davidson & Co, we are frequently asked about behavioral models for asset classes beyond single family residential mortgages. Our first area of expansion was an agency multifamily model released in 2017. Thanks to a new collaboration with Equifax, we are continuing to expand our asset class coverage by developing an auto loan model. Auto loans are the logical next step, as many financial companies carry them on their balance sheets or invest through the ABS market. These firms require accurate default and prepayment projections to reserve, stress test, and value their positions.
Growth in the auto loan industry along with interest from our clients has prompted us to further investigate auto loans.
This article is the first in a multi-part series designed to give an overview of auto loans. Future articles will investigate
delinquency, prepayment and default performance and discuss potential modeling techniques to predict these outcomes.
BOUNCING INTO AUTUMN
by Eknath Belbase
Over the last couple of months, the yield curve has fluctuated quite a bit. From mid-July to mid-August, it rallied hard, with the 10-year dropping a full 55 bps and the belly of the curve moving almost as much, pricing in a significant economic slowdown. Since then, on the heels of firmer growth and inflation numbers, coupled with slightly underwhelming Quantitative Easing in the European Union relative to expectations, it has backed up, reversing roughly 40% of the drop. After the Fed meeting this week, incoming wage, inflation and growth data promise to be highly scrutinized.
Against this backdrop, mortgage durations and how well model durations match up against empirical durations are a natural concern. In our first article, Andy, Feiqi, and Alex compared option-adjusted durations using both our base prepayment model and a prepay-risk neutralized model against empirical duration across a range of coupons. They find that in lower coupons, both duration measures perform reasonably well but that risk-neutral durations tend to do better across the board. This is not surprising since a risk-neutral view would tend to match widening in rallies due to increases in prepay model risk.
Our second article is the debut of Chris Widman, who recently joined our behavioral modeling group. Chris takes a look at the auto-loan market, which is the second largest asset-backed consumer loan sector, and behind only student loans in the non-mortgage consumer loan sector. He reviews the macro trends of the sector and provides an introduction to key features of individual loans worth knowing about. We look forward to more of his articles in this sector as we move towards the introduction of an auto loan model.
We look forward to your questions and comments.
Please login to view the full Pipeline article.EMPIRICAL DURATIONS AND HEDGE RATIOS: CAN AD&CO’S MODELS EXPLAIN THEM?
by Andrew Davidson, Feiqi Jiang and Alex Levin
Agency MBS investors frequently ask: “Can your model match the market’s durations?” We computed empirical durations for a range of TBAs along with TBA hedge ratios and observed them for the past 12 months. We then demonstrate that our OAS model, operating with a risk-neutral version of the LoanDynamics Model (LDM, formerly “prepayment model”), resembles empirical measures of risk rather well.
Please login to view the full Pipeline article.AN INTRODUCTION TO AUTO LOANS
by Chris Widman
According to the Federal Reserve of New York, outstanding auto loan balances have nearly doubled since 2010 growing from $0.7 trillion to over $1.2 trillion by Q1 2019. Auto loans are the second largest non-housing consumer debt behind only student loans.
At Andrew Davidson & Co, we are frequently asked about behavioral models for asset classes beyond single family residential mortgages. Our first area of expansion was an agency multifamily model released in 2017. Thanks to a new collaboration with Equifax, we are continuing to expand our asset class coverage by developing an auto loan model. Auto loans are the logical next step, as many financial companies carry them on their balance sheets or invest through the ABS market. These firms require accurate default and prepayment projections to reserve, stress test, and value their positions.
Growth in the auto loan industry along with interest from our clients has prompted us to further investigate auto loans.
Please login to view the full Pipeline article.This article is the first in a multi-part series designed to give an overview of auto loans. Future articles will investigate
delinquency, prepayment and default performance and discuss potential modeling techniques to predict these outcomes.