Credit risk and asset and liability management are two disciplines often approached via different silos. However, does any interconnection exist between them? If so, can we measure and adequately assess it? Do varying economic conditions affect firm balance sheets differently? We show that incorporating forward-looking behavioral defaults and prepayments into analyses of cashflows that underlie multiple ALM metrics effectively prevents overestimates.
To consider the impact of credit risk on asset and liability management (ALM), we incorporate forward-looking default and prepayment behaviors into ALM analytics. This approach provides more realistic insights into how cashflows will evolve under different economic conditions. Furthermore, neglecting to include both prepayment and default impacts when forecasting the earnings on a loan portfolio will cause an overestimation in potential earnings, regardless of the scenario.
Challenge
Credit risk and ALM are often considered and analyzed independently. It’s not unusual for risk analysts to apply prepayment models to their interest rate risk (IRR) and funding metrics, yet incorporating defaults remains less common, despite also having an impact on cashflows that underlie multiple ALM analytics. Although defaults are not as large an issue during more benign periods, we show that the current economic environment affects many such calculations, necessitating an integrated approach.
Insights
As COVID-19-related economic support measures come to an end, we expect to see many more defaults, prepayments, and, ultimately, ALM issues. An integrated and augmented view of cashflows and balance sheet dynamics enables banks to have more accurate views of these metrics. This vantage is particularly critical when navigating out of a recession, as lenders must make informed and integrated decisions regarding capital deployment.
Combining credit models and economic scenarios to produce credit-adjusted ALM enables more accurate analytics. And greater accuracy allows lenders to make more strategic and better-informed decisions. As economic assumptions influence expected cashflows, both in timing and amount, we show that they also affect multiple ALM metrics, such as net interest income (NII) and economic value (EV).
Analysis
To demonstrate the interaction of credit forecasts into ALM, we analyze UK residential mortgages. We employ loan-level credit models for the probability of default (PD), loss given default (LGD), and prepayment (PP) metrics. Using alternative economic scenarios, we forecast these risk metrics and show the impact on ALM by integrating the models and scenarios.
The credit models are driven by account-level and macroeconomic factors. The account-level factors consist of loan and customer information (for example, loan age and the loan-to-value ratio in the case of mortgages), while the key macroeconomic factors are housing prices and the unemployment rate. In the aftermath of COVID-19, we leverage Moody’s Analytics Baseline, Stagflation, and Severe Downside economic scenarios, which account for different levels of severity.1 To quantify the impact on NII and EV IRR metrics, we use Moody’s Analytics RiskConfidence™ ALM software, which includes both models and economic scenarios available out of the box. Figure 1 summarizes the key steps the tool follows to produce the analysis.
The Baseline scenario assumes a fragile recovery, where existing vaccines are effective against COVID-19, whereas the Severe Downside scenario assumes a new vaccine resistant strain emerges, leading to additional lockdowns and demand shocks. In the Stagflation scenario, signs of de-anchoring inflation expectations cause central banks, including the Bank of England, to hike rates amidst a second recession.
Figures 2 and 3 show the probability of default and prepayment for a UK mortgage portfolio over the next few years for Baseline, Severe Downside, and Stagflation scenarios. The more severe scenarios move defaults higher and prepayments lower. The PDs are rising as the economic environment is gradually deteriorating at the beginning of the forecast, with a rising unemployment rate projected for the three scenarios. The baseline PD forecast peaks at 0.3% in the second quarter of 2022, while Stagflation and Severe Downside peak in the third quarter at 5% and 6%, respectively. As economic conditions stabilize, PDs start declining. The PP displays a similar evolution in the opposite direction.
The PPs react one quarter earlier than PDs, since customers can adjust their repayment speed without restrictions, depending on the economic environment, while a borrower typically becomes delinquent after loan payment is 90 days overdue. The Baseline PP forecast indicates a minimum of 4% for the first quarter of 2022, whereas the Stagflation and Severe Downside reach 3% and 2% one quarter after.
Defaults and prepayments affect IRR metrics:
» Defaults on loans reduce the scheduled principal and interest income, while prepayments bring a reduction in expected interest income only.
» Both models affect the timing of the cashflows to which discount factors are applied when calculating EV.
Figures 4 and 5 display numerical examples of how defaults affect IRR analytics for a single £1 million mortgage.2 For interest rate risk, the main measures we use are changes in either earnings (NII) or EV due to a change in interest rates. If the starting point is estimated inaccurately, however, then it follows that the delta of each measure will also be faulty. Both NII and EV are reduced if prepayments and/or defaults are applied to the cashflows, which would be overestimated if these were not included. Under less severe scenarios, cashflows are mainly affected by prepayments, when the unemployment rate and house prices recover after the recessionary period. In contrast, under more severe economic conditions, PD plays a bigger role, since customers face more difficulties in making near-term payments. These two IRR metrics are also affected in terms of timing; as prepayments shift earlier, the repayment of the scheduled cashflows depends on both behavioral and macroeconomic projections.
As shown in the NII analysis below, neglecting to include both prepayment and default impacts when forecasting the earnings on a loan portfolio will cause an overestimation in potential earnings – whether considering a baseline or stress scenario. Similarly, defaults and prepayment will both affect EV and therefore delta EV. This effect is amplified when summing up the delta contribution coming from all the portfolio’s mortgages and when the portfolio is highly composed of risky customers. The implication for a portfolio being hedged to neutralize the impact of a change in rates on NII or EV is that these behavioral attributes should be taken into account in the hedging decision.
1 Forecast as of April 2021.
2 We assume that the mortgage has a 90% loan-to-value ratio, a 5% interest rate, and five-year linear amortizing.
