I don't think anyone, especially those outside the ALM and IRR-measurement world, would classify EVE-at-risk as a "light" subject. Indeed most discussions about EVE-at-risk quickly become multi-layered and mired in nuance. Small wonder that most bankers spend only a short amount of time paying attention to the measurement. When they do it's always a bit of struggle for them to understand the benefits of measuring EVE-at-risk.
Interestingly enough at the top of my inbox this new year was an email from a colleague with the subject, "Is EVE-at-risk a useful management tool?" Here's an excerpt:
"[I'm trying] to decide whether EVE-at-risk has any use as a management tool. I’m tired of reading that it’s a measure of long-term IRR...I am an EVE-at-risk agnostic...I hear a lot of experts telling us not to manage to it. If I were to venture a guess, I’d say 90% of banks have little to no confidence in EVE-at-risk as a management tool. So long as EVE-at-risk is within policy, we’re happy."
As blunt as his language is, I have to say I agree with this assessment. It matches with most of the deer-in-the-headlights discussions I've had with board members regarding IRR over the years.
But just because at times those discussions seem difficult doesn't mean EVE-at-risk isn't useful. I know that most board members will need to be convinced, but I think there are two basic steps to help them over the hurdle. First, show them how EVE-at-risk captures long-term IRR that is missed or covered-up in an earnings-at-risk simulation. And second, demonstrate how the bank's EVE-at-risk changes using a real-world relevant example.
Long-term IRR is real & earnings-at-risk simulations don't often capture it well
Several years ago I wrote an article entitled, "Earnings at risk is only a short-term view of interest rate risk." My goal was to paint a picture of what long-term IRR looks like when we look at a current balance sheet. To summarize, imagine a simple balance sheet with only long-term 10-year treasuries on the asset-side. On the funding side there is only three-year CD's and capital. I called this Simple Bank & Trust and its balance sheet generates a marginal amount of net interest income. Now, assuming there are no investment sales or calls and assuming there are no CD early withdrawals, we can project a stable amount of net interest income for this bank into the future. The simple static balance sheet will produce the same net interest income for the next three years (I'm ignoring retained earnings - again I'm trying to keep things simple.)
Is there any earnings-at-risk here? i.e. do we expect any negative change in margin given a change in market interest rates? No - not for the first three years anyway. Stress-test a one, two, or even a three-year forecast of earnings and we find little or no IRR using the earnings-at-risk measurement. I think most people would agree that while earnings-at-risk doesn't capture any exposure in this case, there is indeed IRR - it's just longer term IRR. It's the discrete time frame of the earnings simulation that limits its ability to capture longer-term risk.
Another problem with earnings-at-risk simulations is that they use future balance sheet volume projections. Regardless of whether we use a static or dynamic projection new dollar volumes are modeled. These new volumes produce income and expense cash flows that often mask or cover-up risk. Adding a little extra balance sheet growth to a forecast can go a long way toward minimizing a bank's exposure to rising rates. And let's face it - how good are we at forecasting the future anyway? Pretty-bad as it turns out.
The bottom-line is that many earnings-at-risk projections are based on future conditions we cannot predict, and future decisions we haven't yet made. The farther into the future our projections go the more impact the future assumptions have. It's hard enough to reasonably project one year into the future, let alone two. Evaluating a 5-year IRR projection is crystal-gazing in my opinion (a 5-year projection was hazardously suggested by the regulators in the 2010 IRR Advisory.)
Measuring EVE-at-risk in addition to using an earnings simulation will help us capture longer-term risk.
Real world example - how much is too much?
A few years ago I was asked this very question by one of our client banks. They were a $400MM bank located in a large metropolitan area in the mid-west. Most of their assets were short-term (loans and securities) with a duration of less than 1.5 years. Their funding was short-term as well, the bulk of which was money-market deposits that we modeled using a +100% beta factor and FDICIA305 decay (i.e. very short-term). Their overall IRR profile was very similar to other bank's with similar balance sheets. They had minimal short-term exposure to rising rates given their largely "variable-rate" only pricing policy on loans. In fact their earning-at-risk given a rate shock up showed a positive change in margin (+5.00%).
Their long-term IRR was low as well. Their EVE-at-risk was only -6.43% given a rate shock up. Given the short-term nature of their balance sheet the bank's board had previously set an EVE-at-risk policy limit of -15.00%. The limit was somewhat shorter than most banks. A more typical EVE-at-risk policy would have been between -20.00% and -25.00%, but the lower limits made sense for this bank.
Like many other banks over the past few years, this one was looking for ways to improve margin in the low rate environment. And, again like many other banks, they began to extend their investment portfolio to take advantage of higher long-term yields (the regulators warned about this "reaching for yield".) They essentially started taking their short-term on-balance-sheet liquidity and investing it long-term.
Here are the two relevant slides I pulled from my presentation to the board. The bank had $30MM invested in CDs less than one-year at roughly 0.75%. They wanted to start moving this into a much longer-term investment: 5-year/6-mo call FHLB Agencies @ 3.50%. Aside from the obvious change in yield, there was the question of extension risk and the board was asking how much money was too much to move?
Focusing solely on earnings, the CFO made a pretty compelling case. If they moved $5MM they'd improve margin by a few basis points to 3.59% in the base case. If they moved $30MM they'd improve margin considerably to 3.87% in the base case. In either scenario the bank's earnings-at-risk would be minimal. In fact, even at the $30MM investment level the projected change in margin was still positive i.e. even if rates rose margin would improve albeit only slightly (+0.48%) How compelling was the CFO's case? Convert the margin percentages into dollars and you'll see. The difference between 3.59% and 3.87% is 28 basis points...28 basis points on $400MM in assets is over $1MM, certainly enough to catch the attention of the directors.
So why not invest $30MM? Indeed why not invest even more? The short-term IRR measure couldn't adequately answer this question - it showed a positive change in margin in every scenario we ran, even at $30MM (see the slide at right.)
It was only by looking at the EVE-at-risk measurement that we could see the potential consequence. As the bank moved more and more investment long term they created more extension risk. This extension risk moved overall EVE-at-risk up to -16.82% which was outside the bank's policy limit.
So how much was too much? In this case $30MM was too much. The right amount depended on how much additional long-term IRR risk the bank was willing to take on. (On a different, but related note, it also depended on how much additional liquidity risk the bank was willing to take on.)
Balancing short-term vs. long-term risk
So is illustrating both the nature of long-term IRR and how it can be changed enough to convince you that EVE-at-risk can be a valuable management tool? I hope so - I certainly think it is. Although if you're a banker it really doesn't matter whether you buy into using EVE-at-risk or not...the regulators have bought into it. If you make enough short-term decisions to beef up current earnings it may eventually come back to bite you. The short-term versus long-term trade-off is no different for IRR than it is for any other risk. It all depends upon the board's level of understanding and willingness to take on increased risk.
.
Related blog posts:
Low Pressure
http://blogs.olsonresearch.com/brad/2012/10/low-pressure.html
Eventually when rates rise
http://blogs.olsonresearch.com/brad/2011/09/eventually-when-rates-rise.html
So you're not going to liquidate the bank...
http://blogs.olsonresearch.com/brad/2011/08/so-youre-not-going-to-liquidate-the-bank-that-doesnt-make-eve-at-risk-an-irrelevant-metric.html
Earnings simulation versus Equity simulation
http://blogs.olsonresearch.com/brad/2011/04/earnings-simulation-versus-equity-simulation.html
A two-year forecast is too long
http://blogs.olsonresearch.com/brad/2010/07/a-two-year-forecast-is-too-long.html
Unthinkable, of course it is...
http://blogs.olsonresearch.com/brad/2010/07/unthinkable-of-course-it-is-we-dont-know-what-will-happen-two-years-from-now.htm
Demystifying Economic Value of Equity
http://www.echopartners.com/our-blog/bid/21446/Demystifying-Economic-Value-of-Equity
Relationship between earnings-at-risk and EVE-at-risk
http://blogs.olsonresearch.com/brad/2007/08/relationship-be.html
Earnings is only a short-term view of IRR
http://blogs.olsonresearch.com/brad/2007/02/earnings_at_ris.html