Thursday 21 February 2019

Flawed thinking on buybacks, and why buybacks are still underutilized

Something I encounter repeatedly in the world of investment commentary/analysis is a lot of flawed thinking and fundamental confusion about the issue of dividends vs. corporate buybacks. You'll often hear people say things like "I would prefer the company reduce the dividend so it can increase buy backs, as the shares are undervalued", or "management shouldn't be buying back stock at current elevated prices; they should instead be returning excess capital to shareholders via dividends".

A lot of this confusion has been caused by Warren Buffett's past discussions of buy-backs, which have muddied many investors' thinking. Buffett has simultaneously noted that companies should only buy back shares when they are trading at an appreciable discount to intrinsic value, while he has also said, in differing contexts, that companies that cannot find productive ways to reinvest capital into their businesses (where at least one dollar of intrinsic value is created for every dollar retained), should instead return excess cash to shareholders. This seems to imply companies should return capital to shareholders via dividends, unless a company's shares are undervalued.

I will argue that the latter is a flawed perspective, because the truth of the matter is that - tax efficiency considerations aside (discussed later) - when it comes to returning excess cash to shareholders, dividends and buybacks are essentially identical in nature. As is often the case, the issue lies not with Buffett having misunderstood the issues, but instead with investors having taken what he said out of context. Buffett's buyback remarks were likely intended for a context where a company does have productive means to reinvest capital internally, and therefore buybacks do not represent the return of excess cash to shareholders, but rather an active decision about where to best allocate scarce capital. In this situation, the prospective returns on buybacks need to compete with the prospective returns on corporate reinvestment projects, and assuming a company is only reinvesting in projects with a decent return on capital, such buybacks should therefore only be undertaken when the stock is meaningfully undervalued. I will describe these as opportunistic buybacks.

However, the situation is completely different where a mature company has limited reinvestment prospects and has excess capital that needs to be returned - one way or another - to shareholders, and I do not believe Buffett's buyback comments should be interpreted as suggesting Buffett favours dividends over buybacks, regardless of factors such as tax efficiency. When it comes to returning excess capital, dividends and buybacks are actually functionally equivalent, and after allowing for tax considerations, buybacks are usually the vastly superior option. I will refer to these types of buybacks as capital return buybacks.

Here is the proof. Let's scrutinize the claim that a company should 'cut the dividend to increase buybacks'. This statement overlooks the fact that if you are a shareholder and are receiving those dividends, you can use those dividends to purchase more shares, and in the process, create your own 'synthetic buyback'.

Consider a situation where a company is generating stable distributable earnings of $10 per share; the stock trades at $100 per share; and the full $10 is paid out as a dividend (10.0% dividend yield). We will ignore taxes for the moment. If I own 1,000 shares (worth $100k), I will receive $10k in dividends, and assuming the stock continues to trade at $100, I will be able to purchase another 100 shares with the proceeds. One year hence, I will own 1,100 shares worth $100 per share, with a market value of $110k.

Now, let's say at the end of year one, the company decided to undertake a 10:11 share consolidation, or cancel 1 out of every 11 shares. Everyone knows that share splits or consolidations (reverse splits) do not impact underlying value, so that point need not be argued. After the share cancellation, I would now own 1.0k shares instead of 1.1k, and they would be worth $110 instead of $100. In both cases (before and after), my second-year dividend income would be $11k. Before the reverse split, I would receive $10 per share on 1.1k shares, and after, I would receive $11 per share on 1.0k shares.

Importantly, this is exactly the situation I would be in if the company had bought back shares instead of paid a dividend. If instead of paying a dividend at year-end and then conducting a reverse split, the company had instead used the $10 of earnings to buy back shares at $110 at year-end ($110 being $100 plus the $10 of undistributed earnings), it could buy back 1/11th of the company's stock. Post buyback, the shares would trade at $110 with $11 in distributable earnings per share the following year, due to the reduced share count. I would still own my 1.0k shares, and would be entitled to $11k of distributable earnings the next year. In essence, capital return buybacks are actually little more than a form of automatic dividend reinvestment.

Now let's consider the inverse situation, where it is claimed a company should not buy back shares, because they are not undervalued, and should instead return excess capital to shareholders via dividends. This is also an equally flawed approach, because just as investors can create a 'synthetic buyback' by reinvesting their dividends, they can also create a 'synthetic dividend' by selling shares on market in proportion to the company's buyback, yielding cash while maintaining a constant percentage ownership in the company. Indeed, this is exactly what Buffett has been doing in recent times with his shareholding in Wells Fargo, where he wishes to keep his shareholding below 10% of the total company for regulatory reasons. He sells shares on market in proportion to the amount of shares WFC buys back, thereby preserving his shareholding at about 10%.

Let's continue our example, where after one year, the company has bought back 1/11th of its stock, and the shares are trading at $110. I have the same number of shares (1.0k), but they now represent a larger percentage ownership of the company, as the share count has shrunk. That is why they are now worth $110 (US$110k). If I am in need of income, and want to receive $10k in cash, I can create a synthetic dividend by selling 1/11th of my shareholding at $110, yielding $10k. I will now own 909 shares worth $100k (at $110 per share), and will have $10k in cash. My percentage shareholding will now be the same as it was at T=0, and my dividend the next year will be 909 x $11.00, or $10k - exactly as before (the DPS has grown due to the reduced share count). I have created a synthetic dividend.*

Furthermore, the company could at this point conduct a 11:10 split, or a 1:10 bonus issue, to return the share count to its pre-buyback level. Post split, I would end up with 1,000 shares worth $100, instead of 909 shares worth $110, and be in exactly the situation I started (with my $10k effective dividend).

For these reasons, the debate which often happens as to whether a company should return capital to shareholders via dividends or capital return buybacks often makes little sense. By contrast, what does make sense in the real world - but which ironically is seldom discussed - is the tax consequences of one choice over the other, which can be material, and in the majority of cases, favours buybacks. Unless there are practical liquidity constraints limiting the ability of a company to implement a sufficiently-large buyback, tax considerations ought to be the primary determinate of the choice between dividends and buybacks.

Buybacks convert dividend income into capital gains, and in most jurisdictions, capital gains - particularly long term capital gains - are taxed at a lower rates than dividend income. Furthermore, capital gains taxes are deferable, as they are typically only payable when a position is sold, and this also carries very meaningful advantages for long term holders.

As Warren Buffett has explained in the past, when deferred capital gains taxes are accrued but not yet payable, they function like a perpetual, unsecured, interest-free loan from the IRS, where the principal amount of the loan also falls if the capital value of the shares fall.

For instance, let's say I had bought and held the said company in our initial example at $100, and through many years of compounding buybacks, the stock price had risen to $300. Let's say long term capital gains taxes were 25%, which have been accrued but are not yet payable. I now have a gross position of $300, and a deferred tax liability of $50 (25% x the $200 gain), for a net carrying value of the position of $250.

In economic terms, the gross $300 position is leveraged by 16.7%. If the stock rises further to $400, the gross asset return is 33.3% (or 24.9% net of 25% taxes). However, the net value of my $250 position would rise to $325 (now net of $75 of deferred taxes), or by 30.0% after tax, which is 5ppts higher than it would be for someone initiating a position at $300 without this deferred tax leverage.

Furthermore, on the downside, this leverage does not function like a normal margin loan. Not only does it not accrue interest, or face any risk of being withdrawn and hence forcing liquidation, but its size also automatically shrinks if the stock price falls. If the shares fall to say $200, the deferred tax liability will shrink to $25. It is therefore a wonderful way to leverage your returns in a low risk way.**

What this means in practice is that most of the time, buybacks are the far more tax efficient means or returning excess capital to shareholders than ordinary dividends (which are often subject to immediate taxation at shareholders' marginal tax rates, which are usually higher than long term capital gains rates). There are some exceptions (such as Australia/NZ where dividend imputation exists), but they are in the minority.

Consequently, buybacks should be vastly more prevalent than dividends, and in the US at least, they increasingly are. Outside of the US, however - particularly in Europe - many companies continue to pay large dividends subject to high taxes, while returning very little by way of buybacks. Outside the US, opportunistic buybacks are really the only sort of buybacks undertaken, and even then, are undertaken far less often than they should be. Indonesia, for example, has 10-20% dividend taxes, but no capital gains taxes on domestic equities, and yet buybacks in Indonesia are vanishingly rare. It makes no economic sense. Major holders could create tax-free synthetic dividends by buying back stock and then selling shares into the market in proportion, and yet they don't.

There has been a lot of hand-wringing in recent years about the high pace of buybacks in the US, but high buybacks and low dividends is actually the economically rational course of action from the perspective of shareholder value maximisation, and companies should be applauded by analysts for prioritising buybacks, not condemned. And the truth is, even today, US buybacks are still far rarer than they should be, given current tax rates.

If there was more clear-headed thinking about the issue of buybacks, some of this misguided criticism might subside, and better yet, more companies outside of the US might consider following the US example. Here are the rules of buybacks from LT3000:

LT3000's buyback rules

*There are two types of buybacks - opportunistic buybacks and capital-return buybacks. They should not be conflated or confused.

*Opportunistic buybacks should be undertaken only when the shares are undervalued, and the rate of return being offered on capital committed to buybacks exceeds the risk-adjusted return that can be achieved from reinvestment into the business.

*Capital return buybacks are fundamentally different to opportunistic buybacks, and tax considerations aside, are functionally equivalent to dividends. However, the tax treatment of dividends and buybacks can vary significantly. In the absence of share trading liquidity constraints, whether dividends or buybacks are to be preferred should be purely a function of tax efficiency considerations.

*In most (but not all) cases, buybacks are more tax efficient than dividends. The tax benefits of buybacks should consider not only the differing rates of long term capital gains taxes and dividends, but also the value of the deferability of CGTs. If buybacks are more tax efficient, all capital returns should take the form of buybacks, and if dividends are more tax efficient, then all capital returns should take the form of dividends.


*Some might argue at this point, 'but what if the company shares are overvalued'? If the shares are overvalued, investors will receive either a subpar dividend yield or a subpar buyback yield, and it makes no difference which is incurred. The problem is the stock's overvaluation, not the means by which excess cash is returned to shareholders. Furthermore, presumably, at any given time, all existing shareholders do not believe the shares to be overvalued, or else they would not hold them in the first place. 

**This is one reason why Buffett holds on to fully-valued positions like Coke after they have risen many many multiples from his original purchase price - the value of the interest-free deferred tax leverage is too significant to relinquish, and significantly reduces the effective economic P/E multiple. Note that Berkshire's book value is net of accrued but unpaid deferred tax liabilities, and this significant deferred tax leverage (along with float leverage) has contributed meaningfully to Buffett's performance in recent decades.