When Iceland’s National Audit Office (NAO) started its review of the IKr55.3 billion ($456 million) sale of a stake in Íslandsbanki by the Icelandic government in March, one of the things it wanted to look at was the book of demand at the point when the deal’s advisers determined the indicative pricing for the deal.
They still do some things old-school in Iceland – no blockchain nonsense on this deal, not even a proprietary book-building app – so in theory this was easy enough: just send them the Excel from the advisers’ meeting.
So Icelandic State Financial Investments (ISFI, the Icelandic body responsible for managing the government’s holdings in banks) asked Íslandsbanki for the file. And when it arrived, ISFI sent it on to the NAO, on May 27.
At some point it was noticed that the Excel had odd errors in it
Now the fun starts: at some point it was noticed that the Excel had odd errors in it. Before anyone gets all ‘wow those super-sleuths at the NAO are on it’, let’s be clear that these are not errors in some VBA code or badly referenced vlookups. No, we are talking cells formatted as text when they should be numbers and so about IKr20 billion of demand getting excluded from other calculations and stuff.
Simple, but obviously still 'oops'.
I like to imagine the sound made by whoever found that out – and by the next person to be told – might have been heard all through Reykjavik. Anyway, it turns out that this Excel was saved at 19:37 on March 22, pretty much at the start of the meeting between the advisory banks to determine the deal’s indicative price and size, meaning that it was, er, not final.
Presumably it was just where all the order information had been pasted in from emails and other places, ready to be sorted out over the next hour or so. A graph of demand at various prices in the sheet even had a note attached to it saying that not all bids were reflected as the data hadn’t been properly reviewed and cleaned at that point.
Of course, we’re not quite talking FTX-style 'all of these are rough values, and could be slightly off; there is also obviously a chance of typos etc', but you get the idea.
Anyway, sending the wrong file is… embarrassing. Bankers already get enough stick – with some justification in Iceland, given the history – and giving auditors an Excel with confusing things in it is not a good look, especially when those auditors have been tasked with taking a look at a deal that your country’s politicians already really hate.
But here’s the thing. The final Excel sheet used to decide indicative pricing, saved at the end of the meeting at 20:36, was sent to the NAO by Íslandsbanki on October 31. The NAO’s own report confirms this. But the report also still featured a graph of demand as per the first (pretty meaningless) Excel, followed by another graph of demand based on the second Excel.
Not surprisingly, there is quite a difference, since the first one didn’t have a bunch of the book in it, and there were formatting errors etc. It is not immediately obvious what point is being proved here.
So, to summarise, the wrong sheet was sent first, then the right one, the one on which the actual decisions were based, and so any formatting deficiencies of the first one did not affect the indicative pricing.
Yes, Íslandsbanki and then ISFI made an embarrassing mistake – and ISFI compounded it by continuing to refer to data presented in the first Excel in responses to some NAO queries after May. But the weird Excel formatting couldn’t have affected the eventual pricing of the deal because the data was quickly cleaned up on the night.
Inevitably, since the NAO report was published on November 13, there has been quite a bit of talk about the Íslandsbanki deal having been mispriced because of the Excel formatting. But that misunderstands what has gone on here. Even the report doesn’t actually make that claim, and with good reason.
Diatribes
It makes an awful lot of other claims though. The Excel mishap has got some attention because (a) it’s Excel so what’s not to like, and (b) oops. But the NAO report’s other 70 pages fairly drip with bitterness at ISFI and its advisers. And the favour is returned, arguably with interest, in a 45-page rebuttal published by ISFI. It’s all terrific fun.
Readers will doubtless remember the litany of complaints about the original sale from the first time we delved into it here. Largely, they were about how awful it is to sell shares at a discount to the market price and only to people who have somehow been designed special enough to take part.
If that sounds like pretty much every other accelerated bookbuild ever, then you’re not wrong. Except, as it turns out, this one was different to most: not only was it selling about 280 days of trading volume in the stock, but was also equivalent to about 10 days’ trading volume of the entire local stock exchange.
I’m no banker, but I have been writing about equity deals on and off for about 25 years and I think with those kinds of metrics it would be reasonable to see a 4.1% discount at a very volatile time in the market as not a bad result.
According to the auditors, the deal lacked transparency about all manner of things
But that’s not what some people thought – and some of those people were in a position to get the country’s audit office involved. The government also decided to put a hold on all privatizations, and proposed scrapping ISFI altogether.
That is still on the cards.
As well as its formal response, ISFI managed some impressive trolling by posting a summary of the UK’s National Audit Office report on the UK government’s sale of Lloyds shares – a case study that ISFI clearly thinks should have been the model for the Icelandic NAO’s report. After all, the UK concluded that accelerated bookbuilds that excluded retail were the appropriate way to go about things for a whole host of reasons.
And here’s the other surprising thing: amid its 71-page diatribe, the NAO manages to concede that the result of the sale “was generally favourable to the Treasury”. But it just didn’t think it was as good as it could have been.
That is an understatement, given the tone of the rest. According to the auditors, the deal lacked transparency about all manner of things, whether it was the sales method, the classification of different types of investor or the decisions around pricing. The NAO thought ISFI and its advisers lacked the necessary expertise.
In response, ISFI pointed out that it discussed the sales method in communications with the government – which were published in January. The classification of investors as either long-only or not was left to the financial institutions that were taking their orders and therefore knew them best. And the pricing was the result of extensive discussions with highly experienced advisers, including Citi, JPMorgan and STJ Advisors.
But there’s no pleasing some auditors. While the banks were debating between pricing at IKr117 or IKr118 (demand fell off a cliff after that point), the NAO would have preferred consideration of levels all the way up to the day’s close of IKr122. In particular, it noted that it would not have been necessary to cut long-onlys back at all up to IKr120.50. Even at IKr122 there was enough overall demand!
Well, as bankers know, there is demand and there is demand. The conclusion that the advisers reached was that to push pricing further would have meant nothing short of Armageddon on the domestic market, especially if there had been a huge sell-off of other stocks to accommodate higher pricing. And at IKr120.50, the deal would have been just 1.3 times covered. As someone familiar with all this mess tells me: “No one in their right mind would do that”.
It is hard to argue with that.
