Posts

/by

Why Are My Income Simulation Results so Strong in a Shock?

,

In performing model validations for credit unions, we often see income simulation results that show significant improvement in net interest income (NII) and net income (NI) as rates rise, even for credit unions that have material positions in long-term, fixed-rate assets.  Why does this happen, and is it reasonable?

Income simulations are commonly run with a static balance sheet, in which the asset and liability mix remains constant regardless of the rate environment, and loan yields are often assumed to move in lock step with market rates.  However, this approach can lead to simulations calculating material returns on new business and increases in margins as rates rise.
The example below illustrates why this often happens.  The calculations assume an average industry loan-to-asset ratio; yield on loans, yield on investments, cost of funds and net operating expense continues into the future, shown for both the base rate environment and the up 300:
In the base rate environment, the new business calculation above is producing an ROA of 0.80%, roughly in line with the current ROA for the industry.  Said differently, if rates do not change the new business would be neither helping nor hurting NII or NI.  However, if you incorporate typical new business assumptions in an income simulation utilizing a static balance sheet, the new business ROA for the up 300 rate environment is 2.30%, a 150 basis point (bp) increase over the new business ROA for the base rate environment.
Combining a 2.30% new business ROA with the earnings on existing business helps explain why NII and NI results often improve as rates rise, even if the ROA on existing business is decreasing.  However, does a 2.30% new business ROA in a +300 bp rate environment pass the reasonableness test?
Consider the implication; this infers that if rates go up 300 bps, over time the entire industry would earn over 200 bps, more than double any peak the industry has experienced.  Following this logic, a +500 bp rate change would show an ROA exceeding 3.00%.
The purpose of ALM/interest rate risk analysis is to understand risks, not ideal situations.  What are risks that could cause earnings to not be this good?
  • What if our deposit mix becomes more heavily weighted toward member CDs as rates rise? In the largest credit union peer group, the last time rates went up, Reg Shares+Share Drafts dropped from 43% of funds to 33%; currently the ratio is 45%. When rates were higher, CDs were the largest deposit type, costing credit unions much more than non-maturity deposits. Does your base analysis incorporate members shifting funds like they did in the past?
  • What if loan yields don’t move at 100% of the market? Competition to generate loans can often lead to irrational pricing.  It is reasonable to expect that competition for loans may restrict the ability for institutions to move loan rates up 100% of the rate change.
  • What if our loan-to-asset ratio is pressured as rates rise?  Rates don’t only go up due to a thriving economy; the material drop in loan-to-asset ratios over the last five years has squeezed margins.  Lending has picked up over the last year, but there is no guarantee that loan-to-asset ratios can’t drop again.
The earnings on the new business would most likely still be positive in the 300 bps shock, but would be significantly lower than the default assumption.  These are real risks that can greatly impact earnings and need to be incorporated in the base risk analysis.  The exposure to risk should then be compared against policy limits for ALCO to understand the risks being taken.  Beyond the base case the stress of additional pressure should also be tested.
/by

Is Your Risk Methodology Giving You a False Sense of Security?

,

A recent front page article in the Wall Street Journal caused quite a stir by claiming that credit unions are “piling into longer-term assets, exposing the firms to potentially significant losses if interest rates rise…”

The objective of this blog is not to debate whether there is or is not too much interest rate risk (IRR) in the industry. The facts are the facts: credit unions have operated in a historically-low rate environment for the last six years; credit unions have, on average, extended their loan and/or investment portfolios; and there has been a flight-to-safety. The ingredients are there. There is potential IRR built into the system, that, if not managed properly could present issues when rates do rise – depending on what level they rise to and how quickly. The question, then, is “is IRR being managed properly?” or are some decision-makers getting a false sense of security by IRR quantification methodologies that miss risk or allow users to assume away the risk?

In our experience of doing numerous model validations, we have seen many examples of a false sense of security being created either through the methodology or the assumptions needed for the methodology. The most common concerns that may hide risk from decision-makers include:

Traditional income simulation

  • Optimistic assumptions about growth in new business
  • Optimistic assumptions about new volume yields as rates change
  • Seldom incorporate the risk of non-maturity deposit withdrawals or member CD early withdrawals
  • Time horizon is too short, often only one year

Net economic value

  • Optimistic deposit values in current and shocked rate environments (the never-ending argument about deposit length and valuation)
  • Optimistic loan values showing large gains that ignore the market perspective on both credit concerns and liquidity risk

Risk limits

  • Policy limits that focus only on net interest income and net economic value ignoring the entire financial structure, which may hide the potential of negative net income and resulting decline in net worth

Are you at risk of being blindsided? Now is the time to evaluate your modeling methodologies and assumptions.

/by

Policy Limits And The Proposed Regulation For IRR

The following is an excerpt from our response to NCUA’s proposed regulation on IRR.  While the excerpt focuses on a potential issue with the proposed regulation, it also highlights a concern we have with the way certain policy limits are written that don’t look at the actual risk of an institution.

Appropriate Policy Limits

Another concern is the evaluation of appropriate policy limits to ensure “compliance” with the regulation.  The proposed regulation states:

  • Set risk limits for IRR exposures based on selected measures (e.g. limits for changes [emphasis ours] in repricing or duration gaps, income simulation, asset valuation, or net economic value); (Federal Register, Page 16575).

While NCUA has stated in the proposed regulation that these are examples of the types of limits to set and how to set them, the concern is that these examples will become the rule.

Our question is:  Why the focus on percent change versus focusing on the actual risk?

If a line in the sand is never drawn, then as long as a credit union continues to be within the percent change they identified, it would be acceptable for their risk profile to continue to deteriorate.  Also, these types of limits don’t address if the credit union has an adequate net worth ratio.

Consider the following example if the guidance NCUA provides to examiners regarding this proposed regulation is similar to that in the below excerpt from the IRR Questionnaire (click on image to see Table A):


If a credit union has a 1.00% ROA, to maintain a “moderate” level of risk to earnings, the ROA can’t fall below 0.25% (maximum 75% decline) in a 300bp change.  Whereas, a credit union with a 0.40% ROA can have their earnings drop to 0.10%.   What if, at the time of the next simulation, the credit union with a 1.00% ROA is at 1.25%?  Then their ROA can’t fall below 0.31%.  If the credit union that was earning 0.40% now earns 0.30%, then their earnings can’t fall below 0.08%.

In essence, using the percent change methodology, if an institution’s earnings increase in the future, the bar is raised.  Conversely, if earnings drop in the future, the bar is lowered.  Is this really a good measure of safety and soundness?

Additionally, a percent decline approach applied to earnings would never allow a credit union with positive earnings to make the business decision to allow for negative earnings.  There are several cases where external forces or strategic plans make negative earnings in the short term a reality in order to balance the long-term viability of the organization.

Using these guidelines would put any credit union with negative earnings out of policy.  Does that mean that every credit union losing money would automatically be “out of compliance”?  Note that, in 2010, approximately 40% of all credit unions had negative earnings after factoring additional NCUSIF expense.  The potential ramification of this path could be detrimental to the industry.

To see our full response, please click here.

/by

Static Income Simulation – Things To Consider

One of the traditional approaches to income simulation is to use a static balance sheet, which assumes that a credit union’s mix of assets and liabilities remains constant, regardless of external forces.  In other words, static analysis assumes that every dollar of mortgages that matures is immediately put back into a new mortgage at the current market rate.  It also assumes that the deposit mix of the credit union never changes.

While it may sound simple and straightforward enough to just assume the balance sheet structure never changes, these assumptions about new business can be very misleading.  Would you expect mortgage volumes to remain the same if rates increase 300 bps?  Would you expect to maintain the same percentage of your funding in low-cost regular shares if rates increase 300 bps?  The answer to both these questions is, of course, “it’s not likely.”

If you are doing static analysis at your credit union, consider running alternate scenarios slowing mortgage volumes if rates increase.  Also consider increasing member CDs to the levels you experienced as rates were increasing from 2004 to 2006.  These simple-to-run scenarios will likely give you a better picture of your risk to earnings and net worth in a rising rate environment.

Finally, the yield curve we are experiencing today is extremely steep.  We recommend that you also test out scenarios where the yield curve flattens to a more historical level of about 150 bps.  Remember, in 2003-2004 we also experienced a very steep yield curve.  However, when the Fed started raising short-term rates in 2004, long-term rates did not increase by much, flattening the yield curve.