I always hold my breath when a bank official speaks in public about some nerdy accounting topic. Last week was no exception: Bank of England’s Governor Andrew Bailey spoke about the prudential treatment of software assets, see the speech here. The short story is that the Bank of England doesn’t like the way Europe treats software assets for prudential purposes. The EU recently decided to include them in bank capital. But, according to Andrew Bailey this would give a false picture of a bank’s loss absorbing capacity. And by the way, it is also not “in line with global standards.” Consequently, the BoE wants to exclude software assets from capital. Drama, drama ensued, as well as tweetstorms and press coverage. The Financial Times seemed alarmed: “Bank of England plans break from EU with tougher bank capital rule.”
Of course, it is the Governor’s prerogative to talk about software assets and whether they count or don’t count as capital. However, speaking about this topic is tricky. The prudential treatment of software assets looks deceptively simple, but it actually is complex accounting stuff. The risk is that some Bank of England policy officer inadvertently misinformed Andrew, or that Andrew’s speechwriter misunderstood the policy officer. I mean, this is about accounting, and things get lost in translation.
There is also the risk that Andrew wanted to use the prudential treatment of software to make a point. He may have encouraged speechwriters and policy officers to deliberately find a stick to slap EU regulators on the wrist.
Whatever motivated Andrew, for someone who follows the discussion on the prudential treatment of software assets, the speech would curl a toe here and another toe there – mainly because of the way the Bank of England misrepresents the EU treatment of software assets, but also because of the limited effects of the differences between the BoE and the EU approach of the treatment of software assets.
The new EU rule on software assets
The EU changed the prudential treatment of bank software assets late last year, see the publication in the Official Journal here. This after an EBA consultation that followed Europe’s pandemic response or Quick Fix. One reason for the change was to mitigate the risk of regulatory arbitrage. Apparently the treatment of software assets across EU banks is patchwork.
The new rule requires banks to amortize intangible software assets in three years time. For example, if a bank balance sheet shows a value of €90 for an intangible software asset in year one, then the year after that value will be €60, and another year later it will be €30. The year after it is fully amortized.
Before the rule change, intangible software assets were fully deducted from Common Equity Tier 1.
Flaws in the Bank of England’s narrative
Part 1: actually, the software assets will be deducted in full. On the face of it, the new rule constitutes a clear and significant departure from Basel III. Apparently this is how the Bank of England wants to present the EU rule change: in the past software assets were fully deducted, and now the EU wants to count them as capital, which, according to the BoE, is kind of bad.
However, the Bank of England narrative is flawed because it ignores the fact that software intangibles are fully deducted from capital after initial recognition. The ‘addition’ to capital is temporary: every addition will be offset by a deduction. To permanently enjoy the benefits of the new EU rule, banks should keep investing in software assets. It is very much like your gym’s treadmill: you should keep running, or else … .
A bank that keeps investing in software assets will see a permanent increase in asset values of two times the value of annual investments; assuming that the bank invests the same amount every year.* The blue line in the graph below shows the permanent increase in asset values for a bank that invests €100 in software assets annually.
The graph also shows that under the new EU rules, assuming no growth, investments in software assets are fully deducted from common equity from year 3 onward. After year 3, investments in software assets match the amortization expense: the result it that the sum of assets, the blue line, plateaus at €200.
Of course, the permanent increase will be larger if the bank wants to grow software assets. This is an important point, because ignoring goodwill has known real effects on investments. See this recent paper by the FED Board: Bank Lending in the Knowledge Economy.
Cutting banks some slack with respect to software investments may actually be a good thing.
Part 2: the effects of the decision are small. Though the Financial Times article sounds alarmist, it also mentions that the effects on bank capital are limited to about 30 basis points. Louie Woodall from Risk.net noticed that Deutsche Bank was the greatest beneficiary, reporting a gross 43 basis point benefit.
Part 3: tinkering with prudential capital adjustments will not move the needle. Will 30 basis points affect the markets? According to my research with Roger Willet, see here, the effects of adjustments to bank capital are minimal. The graph below shows, for U.S. banks, how share prices respond to the deduction of Goodwill and Intangibles.
See that dashed line close to the horizontal axis? It shows the effect on share prices of a one percent change in the value of Goodwill and Intangibles. It is limited to single-digit basis points. And note: the deduction of Goodwill and Intangibles lowers capital by 215 basis points on average, see Table 1 in the paper. It is therefore unlikely that a decision of either the BoE or the EU on software intangibles worth 30bp has an economically significant impact, if at all.
In all, using the treatment of software assets to signal a toughening-up of prudential rules looks a bit over the top. The prudential effects are negligible. It will probably not affect a future bank resolution outcome. However, on the margin and in going concern it may present EU banks an incentive to increase investments in software – what is wrong about that?
* The permanent increase in asset values will be two times the value of the annual investment, assuming constant investments and a three-year amortization period, with linear amortization as required under the new EU rule. The maths behind this: A = ½ × (n+1) × (I+R), where n is the amortization period, I is the annual investment, and R is the residual value of the asset. For n=3, I =100, and R=0 it is 200. [½ × (3-1) × (100+0)], see graph above. The choice of the amortization period can increase the value of software assets significantly. So, yes, I can understand why the EBA wanted a short period, i.e. not exceeding three years.