“Google plays the yield curve”

Here is Greg Mankiw’s very interesting post, but with an open comments section:

I was fascinated a story in today’s Wall Street Journal.  Apparently, Google is sitting on $37 billion in cash, but nonetheless decided to sell $3 billion worth of bonds.  Why?  To take advantage of low interest rates.

It is like reverse maturity transformation.  The banking system borrows short and lends long.  Google is borrowing long and lending short.  (Or maybe I should call it reverse quantitative easing, as Google is also doing exactly the opposite of what the Fed has been doing.)

Does this make sense for Google?  I have no idea, and I am ready to concede that those guys are a lot smarter (and financially successful) than I am.  But there is reason to be skeptical.

The chart above shows the spread between the ten-year Treasury bond and the three-month Treasury bill.  The yield spread is now high by historical standards.  The empirical literature on the expectations theory of the term structure (in which I have sometimes played) suggests that this is a good time to borrow short and lend long–the opposite of what Google is doing.

Maybe this time is different, and past empirical regularities will not hold going forward.  But ponder this question: If you had a friend with a paid-up house, would you suggest that he now take out a long-term mortgage in order to deposit the proceeds in a money-market fund?  If not, does it make sense for Google to be doing much the same thing?

The usual explanation for this kind of apparently strange financial behavior is that shareholders wish to force the managers into accepting the scrutiny of outside capital markets; see Easterbrook 1984.  That seems less plausible in the case of Google, where concentrated delegated monitoring by major shareholders remains strong.  An alternative explanation is that Google has a very high option value for the cash, which more or less implies they see a lot of acquisition and investment opportunities in their not so distant future.  A lot.


I think you hit the nail on the head in the end there. Google wants to buy things. A lot of things, and very expensive things at that. $40B for Twitter, a Groupon clone, maybe FourSquare or GroupMe and a bunch of other things? Wouldn't surprise me in the slightest.

Dan: You might be right about the acquisition tear, but Foursquare is almost certainly not selling to Google. Dennis Crowley has already been down that road with Dodgeball and he was not happy with the outcome.

Isn't selling bonds consistent with what the Fed wants Google to do? The Fed is pushing yields down to make spending money, rather than investing it, more attractive.

A large portion of Google's cash is tied up in its Irish, Cayman, and other offshore subsidiaries, where Google would be subject to a 35% US tax if they repatriated that cash back to the US or used it to acquire US property. Thus, it is cheaper for Google to pay 0-4% to borrow new cash to use in the US, in order to defer that 35% US tax on the offshore cash indefinitely (or until Congress passes a tax break for repatriating offshore cash).


Absolutely bingo. Or, rather, bingo with the US tax code.

Look at Cisco as well.

Tax is a very likely driver. Although global firms have many creative ways of avoiding the repatriation tax (see http://www.bloomberg.com/news/2010-12-29/dodging-repatriation-tax-lets-u-s-companies-bring-home-cash.html ), if Google was looking to buy a purely American startup then some of these tax "optimisation" strategies may be unviable.

It is somewhat ironic that Google was a major supporter for a "tax holiday" this year, which would have allowed reparation of overseas profits without any US tax liability. (See http://www.thefiscaltimes.com/Articles/2011/04/28/WP-Companies-Lobby-for-tax-holiday.aspx).

At the time I found it ironic to argue for such a proposal, since undertaken something similar under the American Job Creation Act of 2004 failed to achieve a boom of internal investment by corporations. See Michael Faulkender & Mitchell Petersen, "Investment and Capital Constraints: Repatriations Under the American Jobs Creation Act", NBER Working Paper No. 15248 (2009). But after reading this, I see their lobbying in a different light.

I worked in finance for a fortune 500 company and we had an enormous staff of accountants, tax lawyers, etc. in our treasury department. I would go as far to say that managing our repatriation tax liability was the driving force behind financing and even large investment decisions.

It makes me wonder if this sort of thing is an unappreciated influence on the effect of monetary policy.

good comment.

it blows my mind a guy like Mankiw can overlook a finance topic like taxes. having no debt on your balance sheet is a bad thing (WACC is higher).

Cash for tech firms is the nuclear option of the capital markets. MSFT held massive amounts of cash for years in order to tell competitors "look don't get a in bidding war over another company/OS platform because we will crush you."

If the yield curve steepens, then google would make money because the duration on bonds with longer terms to maturity is higher than those for short terms to maturity.

Basically they want to lock in low long-term interest rates today.

They plan to invest this money later, they want to lock in a low rate now.

I think L. Zhang has nailed it. Since the tax cost of repatriation would be prohibitive, it is a lot cheaper to borrow in the U.S. for U.S. acquisitions. Considering the tax implication (for argument sake 30% tax), there is virtually no interest rate that would not make the borrowing attractive.

Google's not the first one to do this. Quoting from the end of this Alphaville post http://ftalphaville.ft.com/blog/2010/10/29/386951/cash-hoarding-corps/ : "Moody’s notes that companies have more than enough cash to fund quite a lot of spending, including acquisitions, and to cover dividend payments without having to resort to debt markets. Yet many companies have done just that."

The answer is, as you note, optionality. Maintaining optionality through cash is a good strategy for an incumbent monopolist as I note here http://www.macroresilience.com/2010/11/30/the-great-recession-through-a-crony-capitalist-lens/ : "In an environment where incumbents are under limited threat of being superceded by exploratory new entrants, holding cash is an extremely effective way to retain optionality (a strategy that is much less effective if the pace of exploratory innovation is high as an extended period of standing on the sidelines of exploratory activity can degrade the ability of the incumbent to rejoin the fray)."

It could be argued that incumbents could follow this strategy even when new entrants threaten them. This strategy however has its limits – an extended period of standing on the sidelines of exploratory activity can degrade the ability of the incumbent to rejoin the fray. As Brian Loasby remarked in 'Equilibrium and Evolution' : “For many years, Arnold Weinberg chose to build up GEC’s reserves against an uncertain technological future in the form of cash rather than by investing in the creation of technological capabilities of unknown value. This policy, one might suggest, appears much more attractive in a financial environment where technology can often be bought by buying companies than in one where the market for corporate control is more tightly constrained; but it must be remembered that some, perhaps substantial, technological capability is likely to be needed in order to judge what companies are worth acquiring, and to make effective use of the acquisitions. As so often, substitutes are also in part complements.”

Incumbents choosing to preserve optionality rather than investing in new tech is a sign of stagnation and lack of exploratory innovation http://www.macroresilience.com/2010/11/24/the-cause-and-impact-of-crony-capitalism-the-great-stagnation-and-the-great-recession/ .

Intel used to do the same -- ie, have bonds as both assets and liabilities -- but swapped everything into 3m LIBOR, so it wasn't a yield curve play. There's been some work on just this issue in corp finance, I'll try to track down a ref.

How is Google lending money via a bond sale? I don't quite understand. I thought selling bonds was a way to borrow money?

Ok, I take that back, I assumed that Google was issuing short term bonds(since they obviously have the cash to pay them off in a year), but it looks like they're selling ten year bonds.

Also worth noting that the fact that the bond is a fixed-rate bond says nothing about Google's interest rate exposure. Most corporations issue fixed rate bonds because investors like them. But then they often swap it back into a floating rate exposure via an interest rate swap with a bank. So their 10y bond issued at 0.60% above treasuries will get swapped into essentially 3m Libor + (0.60% - 10y swap spread), probably around 0.75%-0.80% in the current environment. Which of course makes a lot more sense for the "maintaining optionality" story I made above.

Moral of the story: To make sense of corporate interest rate exposures, looking at bond issuance is not enough. You also need to look at the interest rate derivatives hedging program.

Interest rate derivatives also explain why the usual maturity transformation story is simply wrong http://www.macroresilience.com/2010/04/04/maturity-transformation-and-the-yield-curve/ .

If you doubt the government's willingness to maintain financial discipline, then you expect a period of high inflation. Borrowing money before such an event seems to me quite sensible.

Right Wing,

It's the tax code. If you believe in EMH, the price of inflation would be incorporated in the bond interest rate.

Either believe in EMH or believe in your "government willingness to maintain financial discipline hypothesis", but not both.

Unless you are bipolar.

First, I stand in awe of your ability to perform remote armchair psychoanalysis. Second, I thought that you were no believer in EMH. Third, all-to-common bubble phenomenon is a clear refuter of the strong EMH. So is the career of Peter Lynch.

I started warning about the housing market Q4 of 2005. I extended my worries to the general economy in the summer of 2006. I talked my parents out of dumping stock in 2009. NONE of these market-contrary, market-beating calls required anything more than paying attention, and recognizing irrational exuberance (or its inverse).

I'll say it again. The US bond market is in denial. I've been saying it since before Walmart announced it.

Wait a second Bill. If the tax code is modifying a behavior from an otherwise rational decision, that undermines EMH. It's not about EMH or not to EMH, it is about what affects EMH.

Andrew'--Nothing has changed in the tax code, in case you haven/'t noticed. If it is what is a future probability, it would be contained in EMH.

Or this may have a lot to do with the fact that Google just put together an internal trading desk and those guys are anxious and confident, so they want to get into debt market transactions...

My theory is that they aren't selling bonds, but shorting bonds, just like PIMCO. They may buy Gross' theory that the bond market will collapse at the end of June when the Fed stops buying. Thus there is no irony that they are selling bonds while sitting on so much cash, because they are gambling with that cash. This news temps me to sell my stock in Google, but I'll research further.... I'm gonna be shocked if Gross turns out to be correct and bonds yields rise in July; yields have been rising BECAUSE of QE2 despite everyone and their goddamn uncle claiming the Fed purchases are holding yields down: QE2 HAS RAISED BOND YIELDS!!!!!! Never mind. No one listens to me.

Mankiw is confusing two things. If your only way to make money is by making financial bets, then sure, borrow short and lend long and try to capture that spread to earn a couple of basis points (or more if you're leveraged). But if you're Google and think you can find long-term investment opportunities -- and simply want to make sure you have the cash to invest in them -- that return from 10-40% return on investment (like buying new companies, starting new products, launching new divisions), then locking in 1-4% interest rates now is a good deal. We'll see if they can actually find those opportunities, but it's a bit near sighted to think they should be playing the yield curve when the strategy is to get cheap capital to find long-term investment opportunities.

L. Zhang has the most likely to be correct theory- Google expects to need cash for purchases, but to use their cash for this, they will have to repatriate it first and face a tax bill that is larger than the interest cost for borrowing. Google is not run by idiots.

Remember Miller-Modigliani with taxes -- switching to debt financing increases their value since they can deduct interest on debt from corporate taxes. The additional interest cost is lower than the taxes they avoid paying.

Bingo. Plus, the market's not exactly efficient...MM doesn't hold in the real world. Surprised no one's pointed out that they could use all $3 billion to fund a share buyback if they wanted -- replacing expensive equity financing with cheap debt when you're still net cash doesn't make your (in this case, undervalued) equity appreciably riskier, IMO. Theory and practice...

Maybe this should be a regular feature of MR: reproducing Mankiw's posts with the opportunity for comments. It worked well in this case (thanks especially to L. Zhang).

companies do this all the time. probably every company in the DOW gets loans from time to time and most probably have outstanding loans right now, and probably most of the SP500. why? the same reason you use a credit card to buy groceries instead of selling your house. the fact that rates are low makes the use of debt for short-term financing even more attractive

Borrow now, keep cash around, buy short-term bonds only when short-term interest rates stop rising. And then become the master of the universe.
I doubt that you haven't noticed it yet, but banks are sitting on piles of cash too. Guess why.

Hmm.. This looks like another crisis forming. This time, it's going to be about companies overleveraging themselves in order to take advantage of low rates. Eerily similar to what house buyers did a few years ago..

Agree with Zhang, but would also add that Google must be thinking of using the money for an acquisition as well.

Google can use existing US MNC tax code provisions to fund US R&D, for example, by simply having the foreign sub contract with its US entity to do R&D and have the patents or the other IP remain in the foreign sub. In other words, current US domestic activities for a R&D intensive firm such as Google can be financed by foreign subs to avoid US taxes.

But, where Google would have a problem is if it wanted a whole bunch of cash to do an acquisition. It couldn't play the "contract with a foreign sub" routine.

So, I would guess an acquisition of some type is in the offing.

L Zhang's theory about the repatriation tax angle is intriguing, but Morningstar says:

The firm doesn't have the same issue with overseas cash that many of its large-cap technology peers do. Of Google's $37 billion in cash, only about $17 billion is sitting outside the U.S. Microsoft, by contrast, has no net cash remaining in the U.S., while nearly 90% of Cisco's cash is sitting outside of the country.


Those are big numbers, even if it is half and certainly dwarfs the borrowing..

I am willing to bet the major algorithms sit securely in a foreign tax haven sub pumping out revenue to other parts of the company to perform contract R&D for the foreign sub, which then again gets the IP rights on whatever was invented.

Why should their decision-making process with $20bn in US cash and $17bn in foreign cash be any different from whether they did not have any foreign cash at all? In either case you treat the foreign cash as if it cannot be accessed. I think the issue of the lock-in effect of permanently reinvested earnings is a red herring here.

Can there be any serious doubt as to whether Google can achieve better returns on its investment in equity than its cost of debt? Mankiw oddly assumes that Google can profit only from the debt spread. Google is not a bank. They will spend that money on startups.

Or maybe wide area wi-fi networks.

And Mankiw just posted an addendum to his blog from a reader who makes the same points that have been made above. The puzzling thing of all this is whether Mankiw is that naive about the tax code and the nature of corporate finance. Probably a good think that he is an econ professor and not a hedge fund manager.

I thought exactly the same.

Of course, the “anticipation of future acquisition or other cash need” is inconsistent with the “scrutiny of capital markets” theory. This juxtaposition argues that not only can Google gain more value from an acquisition than the previous owners could, they can gain more value than the normally gung-ho cheerleaders on Wall Street are willing to guess, so this pre-emptive borrowing prevents higher borrowing costs.

Either that, or, like virtually every corporate treasury, Google is making speculative interest rate bets, possibly encouraged by some sweet whisperings from their friendly investment bankers. The aggregate expense of the trillions of interest rate swaps and other corporate playthings is astonishing.

If Google is doing this as a financial firm, as you suggest, it is in the wrong business, and probably not suited for it.

Leave it to Goldman Sachs.

And, if you are a Google shareholder, please tell me where in their stock prospectus they disclosed they were embarking on an excursion into the financial services industry. Wait for the lawsuits if they are.

During the hyperinflationary period in Latin America some companies with privileged access to markets and not subject to banking regulations started making more money with financial management than with operating profits. This kind of distortion easily becomes the norm when the monetary and fiscal policies are chronically mismanaged.

It doesn't require hyperinflation. GMAC for quite some time was characterized as a money-losing manufacturing business feeding a money-making loan business. I saw quite a few stories about companies making more money off their retirement accounts than in their core businesses.

It might also be worth noting that there is a significant boom at the moment in small tech companies. Hiring is tight, wages are rising and there's aspirational desires to get in now.
Given the value created during the last bubble and Google's tax issues with repatriating money it makes perfect sense that they'd be borrowing from a generally-slow economy to invest in their area of expertise, which doesn't appear slow at all. Those returns aren't available to regular investors, since they are predominantly small, private companies that don't seem to be headed for IPOs at all and even then it requires technical expertise to sort the wheat from the chaff.

Some references:

* Dino Palazzo: http://people.bu.edu/bpalazzo/Cash_Holdings.pdf
* Viral Acharya: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=650997

Each has references to other work on the subject.

Another one: http://fisher.osu.edu/~zhang_1868/LivdanSaprizaZhang09JF.pdf

They desperately need to hire engineers, and the only way to hire huge numbers at once is to buy other companies.

They could buy Yahoo! - 20 billion gets you a bunch of very good software people. Yahoo is a lousy company, but great at building useful things.

If you're not in software, you may not have noticed how desperate companies are to hire right now. Go to any event and you have to push your way through throngs of recruiters. Google basically wants to hire _everyone_ who's good. They've got the projects to make it work.

A majority of Google's cash is available state side so I doubt they are borrowing to avoid the repatriation of cash.

They're simply locking in low interest rates now to as a hedge higher interest rates in 3-10 years. They clearly don't have enough investment opportunities today (they'd never have this much cash if that weren't the case) so this is simply a way for them to alter the economics of their decision making in the future. In 5 years, if interest rates spike, their hurdle rate for making a deal also spikes. This move, however, gives them flexibility to invest in 5 years at this historically low rate.

A good move if you ask me, rates will definitely need to be higher over that time frame.

It makes sense if their optimal capital structure would include some measure of debt. In which case they should borrow debt with a duration that matches their assets, and use that cash to buy back stock.

This shows which they last very much lengthier and thus saving you income which could otherwise are actually utilized to purchase new ones.

Comments for this post are closed