A few years ago I wrote a series of posts about how you might set interest rate risk policy limits.  In one particular post I echoed this quote from a colleague, “ALCO Policy Limits should be wide enough to be functional, but narrow enough to be meaningful…”  Sharing this little observation with anybody involved in bank management usually draws a nod of understanding.  However, I think that while many agree with the “narrow enough to be meaningful” part, they don’t give the “wide enough to be functional” part much thought.  Or, if they do, they think, “yea I know a bank or two that sets their policy limits really wide so they’ll never violate them.”  To be fair, that’s what I usually think too.  And let’s be honest, we all know of bankers that would ditch their policies altogether if the regulators didn’t care.  Ignoring ALCO policy is a bad idea, but you do need to consider that “wide enough to be functional” part.

A great example of this came up with a client recently.  Their bank is receiving a lot of heat from examiners for 1) having a policy limit that’s too wide and 2) violating that policy and not having a plan to correct it.  The ALCO policy in question? A Net Earnings-at-risk limit of –30.0% given a +/-200bp shock.  Before you get too alarmed – lets be clear – that’s a net income-at-risk policy, not a net interest income-at-risk policy.  One measures the change in net-income (NI), the other measures the change in net-interest-income (NII).  

Confusing the two is an easy mistake to make.  People commonly say "Earnings-at-risk" when they are actually referring to is "net-interest-income at risk" or "margin-at-risk".  I'm not trying to be pedantic here, I just want to make sure I highlight the distinction.  The bank in question here has two policy limits in place:

• -15% Net Interest Earnings-at-risk (this may be considered a little wide, but the bank is growing)

• -30% Net Income-at-risk

To address the regulator’s concerns I offered several possible solutions:

A) Give it time
Leave the policies in place, but each quarter when reviewing IRR make note of the policy violation.  It’s a temporary thing.  The bank is still growing and eventually net income will “catch-up”.  The bank is also unlikely to continue it’s high level of provision expense indefinitely, currently they are projecting $1,500.  Using the back-of-the-envelope if you reduce to projected provision expense back down to a more normal level, say around $300 the IRR measurement quickly comes into line.  Essentially add $1,200 back into net income.  $319 + $1,200 = $1,519.  A change of $-312 applied to $1,519 is –20.53% which is within the policy limit.  Be sure to include an explanation of this line of thinking in the ALCO meeting minutes.

B) Change your margin-at-risk limit
If the regulators still feel the –30% policy limit is too wide, then all things being equal, they have to believe that the –15% net interest earnings-at-risk is too wide also.  Remember, normally in a shock-test most items below the margin don’t change when rates change, so we’re applying the same dollar change to both margin and net income.  In this case if we’re only going to tolerate a –30% change in net income that translates into $-96 ($319 x –30% = $-96).  That means that, at the policy limit, we’ll have to constrain or our margin change by the same amount.  So a change of $-96 applied to a projected margin of $6,479 is roughly –1.50%.  To be consistent the bank would have to change its margin-at-risk limit to -1.50%.  That’s quite narrow and basically impossible to manage.  Again take a look at the peer data.  The average margin-at-risk is –7.8% and the lowest in the typical range is –4.0%; A limit of –1.5% would be way too tight.

C) Eliminate the Net Income-at-Risk policy altogether
If the bank doesn’t have any significant interest-sensitive items that it reports below the margin then what’s the point of the net income-at-risk policy?  Fundamentally rate changes impact margin so set your policy limits using net interest margin.  There are too many other factors that will skew the net income-at-risk numbers.  In this case it’s a high provision expense, but it could have easily been a high or low security gain/loss, or an unusual expense item.  Net income is already quite small compared to margin for most banks.  Anything that significantly reduces the ratio of net income to margin will wreak-havoc on the net income-at-risk measurement.

After writing this out I think I’d prefer option C.  If the regulators don’t like that, then choose option A.  If option A won’t fly then option B, but option B would be a real pain.

(Additional note as of 3/17/2024 - I'll point out that semi-annually the OCC has been  publishing its industry peer data for IRR.  They do not include statistics for net income at risk.  While I don't know the official reason for this, I suspect the volatility of the measurement had something to do with it.)