Evaluating Company Growth

For investors using either the absolute or heuristic fundamental analysis strategy (Big Picture Strategy page), everything comes down to expectations of future growth. This can be the

1. growth in take-over value (e.g.of a new drug),
2. growth in net asset values (e.g.of a resource company’s reserves),
3. growth in dividends (e.g.paid by a utility), or
4. growth in EPS (e.g.of an industry consolidator).

The first two types of growth are outside the discussions on this website. (1) When a company is not, and never will be profitable, its value lies in the possibility of a take-over business monetizing it. The retail investor has really no hope of analyzing that. (2) Commodity and real-estate prices go up and down according to their own rules. You can generate financial analysis only by first making a presumption about the commodity price. For example, Oil and Gas companies are valued according to their reserves in the ground. As the commodity price increases so does the company’s value.

(3) and (4): Both dividend and EPS growth derive from a business’s profits and their efficient reinvestment. Retail investors can get their teeth into this analysis. It is these ‘normal’ operations that are being discussed below.

………………………….

Investors need to be very clear that dividend growth is not something separate and distinct from generic ‘growth’. It is a mistake to simplify the stock picking process by considering only dividend yields and historical dividend growth rates. The dividend can be increased by paying out a greater portion of earnings, but that has its own obvious limits and comes at the expense of future growth. For sustainable dividend growth there needs to be profit growth. Investors look to the operations of the business, not at management’s decisions.

Future growth will not be constrained or determined by a company’s past growth, but ….. to evaluate the truth/probability of the promises, an understanding of how the company has grown in the past is critical. Anyone can spin a story about the rosy future. Will the past repeat, or what catalyst will change things? To quote the father of security analysis (Benjamin Graham’s “Security Analysis” 1934), “Some matters of vital significance, e.g., the determination of the future prospects of an enterprise, have received little space, because little of definite value can be said on the subject.” Most all the following focuses on understanding the past.

Most investors know and accept the back-testing that shows excess returns from buying cheap stocks (low price/book or price/earnings). But there are different understandings of the cause. Some claim cheapness is a risk. You would demand higher returns because you see higher risk, so you don’t get any kudos for those higher returns.

But there is another explanation for the higher returns from value stocks. They may result from market mispricing. Companies’ growth rates show mean-reversion to the economy’s overall growth rate. Excess sales or earnings growth in one period tends to be followed by deteriorating results in the next period. Investors over value stocks by extrapolating recent great results into the future. Actual realized growth is negatively correlated to past growth.

The point here is that history can explain what has happened in the past, but you presume it to continue at your peril.

Too many investors make no attempt to understand or evaluate the companies they buy. Following the market leaves you always one step behind. By the time an understanding becomes the received wisdom it is already priced into the markets. To play the stock-picking game you must be ahead of the curve. You must make decisions that contradict the market. If you cannot do that then you should buy index ETFs, not company stocks.

An example of this behavior is when investors use the widely published analysts’ earnings and 5-yr growth estimates. It may even be true that MOST retail investors do not make their own estimates. This (lack of) analysis will (by definition) justify current prices, because those estimates determined current prices. There is a tautology created by this reliance.

The value of these earnings forecasts is easily back-tested to prove how inaccurate they are. A study by the CXO Group shows how very wrong each quarters’ earnings estimate is at its inception, and how widely it swings in the interim before its factual reality is known after the fact. Take a minute to look at their graphs of individual quarters’ estimates.

Using S&P500; data and their reports on analysts’ EPS projections in March of each year (so the year end results are in) you can calculate their growth projections for the full year. Here are the results … that never predict years of losses, and widely over-estimate the recoveries from those losses. Most of the time analysts simply plug in a standard 15% to 20% … even though the historical earnings growth averages only 5%. Pretty much a joke.

Or look at this graph from Gerard Minack of Morgan Stanley, of original forecasts and their change as they approach maturity. The farther the colored line strays from 100, the more inaccurate the initial forecast.

ROE is often considered the next most important metric for analyzing stocks, after EPS, because it is a proxy for the company’s growth potential. But it is too simple to say that all high ROE companies make good investments. (See Canadian government’s list of ROE by industry). ROE is a measure of the company’s efficient use of assets and leverage. Some industries have high ROE because they require no assets – like consulting firms. Other industries require large infrastructure investments before they generate a penny of profit – like oil refiners. You cannot conclude that consulting firms are better investments than refiners just because of their ROE. It is generally the case that capital intensive businesses (low ROE) have high barriers to entry, while the high ROE firms are more risky because there is nothing preventing competitors taking away the customer.

There is no immediate benefit from a high ROE. Since stock prices are most strongly determined by EPS, you will be paying twice as much (in Price/Book terms) for a 20% ROE company as for a 10% ROE company. The benefit comes from the earnings REINVESTED in the company at a high ROE rate … giving the company a high growth rate. If the earnings are not reinvested, but paid out as dividends then the ROE is irrelevant. See the comparison of investors’ returns at Understand Equity

………………………….

ROE provides the basis for a theoretical estimate of the company’s growth rate. The most basic equation is:

Growth = ROE × (1 – payout ratio).

E.g. if the company pays 40% of its earnings as dividends and its ROE = 15%, then its growth will be 15% * (1-.4) = 9%. Of course this is very much theoretical. It presumes the existing assets continue to earn the same ROE and that the new investment earns the same ROE as well. It also ignores share buy-backs. Real life isn’t so predictable. But despite its limitations this growth rate is a strong predictor of future stock returns (Shapovalova, et al.).

The example following shows how the investor can predict his returns from what the company does with its profits (the left column). (This analysis ignores changes in the market’s idea of the valuation multiple.) Both dividends and buy-backs earn the investor a return equal to the earning yield (the flip of the P/E ratio). In this example the expected growth rate for the company’s earnings would be 12%. The investor’s growth in earnings PER SHARE would add another 2% return. His total return would be 15% when the 1% dividend yield is included.

You can use this model for comparing the expected total returns from different securities with different dividend yields and valuation multiples. Or you can compare the common stock to the preferred stock of the same company. The major unknown variable is the incremental ROE of the reinvested earnings. You might find it difficult to predict an exact input for ROE, but by reverse engineering the model to generate your necessary total return, you are presented with an assumption of ROE you can either accept as reasonable or reject. Use the Different Yields spreadsheet.

Historically the incremental ROE of the S&P; is 12%. This is calculated as the increase in earnings as a percent of earnings-less-dividends. So it includes the effects of share issues, buybacks and stock options. Many dividend believers will not accept these numbers. They are convinced that profits left inside a company gets wasted. But the evidence is the evidence. You check the data on Sheet 31 of the data spreadsheet.

………………………….

The most basic of valuation models that derive from discounting future dividends totally ignore the incremental return on reinvested earnings. Look at the spreadsheet for the Gordon Model. It generates a valuation and a resulting P/E value. You can see in the comparative section of the spreadsheet what ROE must be presumed in order to accept any of the P/Es resulting from using the model. The wide variance proves the models are really worthless.

………………………….

The published ROE requires you to to accept the numerator (earnings) and the denominator (equity) as stated in the reported Financials. Any revisions you make to those metrics will change the reported ROE. Companies with large depreciation expenses may have higher growth rates than predicted by ROE because higher operating cash flows are reinvested in the interim before long-life assets need to be replaced.

Also remember that ROE measures the combined effects of leverage AND the efficient use of assets to generate profits. Companies with different debt levels will show different ROE (and resulting growth) even though their operations are the same. The higher ROE may indicate higher risk from that leverage. Use the Return on Assets metric to separate these effects.

The reason investors pay so much attention to profits is because they presume those profits will be put to good use earning even more profits. That is a naive presumption. There are lots of ways management can fritter them away. Analysts must identify profits that are being wasted (vs well invested) and profits that don’t generate growth simply because of broad industry slowdowns (vs booming economies). That is one reason for always analyzing at least two competing businesses in the same sector before buying a stock.

Answering the question “where do the profits go?” is easier now that financial data is publicly available for download into spreadsheets. There are two answers. They can be used to pay dividends and buy back shares, or they can be stored on the Balance Sheet. The example provided in the Company Analysis spreadsheet includes a box itemizing the changes to Balance Sheet values. It is similar to the Statement of Cash Flow but tracks profits instead of cash. The top section includes all the sources of financing. The bottom section lists the operating assets.

Example:

The example above is of a power utility whose revenues are not increasing, but is still spending more money on Fixed Assets than it depreciates. (Thus disproving the common perception that utilities are ‘cash flow rich’ because their assets last forever). You can see clearly they are paying out 40% to 50% of profits as dividends – fine. They are increasing debt (ignore lumpiness) only in amounts equal to the reinvested profits (Income less Dividends) – fine. They are not relying on the proceeds from share issues for financing – fine.

The issue clearly is the large investment in fixed assets. Are those productive investments?. Profit margins have increased. Is that because they replaced manpower with machines – a onetime efficiency that will not continue. Or is the spending on Fixed Assets a result of capitalizing costs that should more correctly be expensed?

Revenues are not increasing, even though utility rates are determined by investments. Why? Looking further you find the regulated business does have increasing revenues, it is the non-regulated division that is the problem. Is that a cyclical economic issue or a business issue?

The management discussion says the expenditures were mostly in the utilities division where the ROE is regulated at only 9%. If 50% to 60% of profits are reinvested, that means growth would be expected at about 5%. Together with a 3% dividend, is an 8% return an OK investment for this level of stability?

This is just one example. The point here is that knowing where the profits are go, helps you determine where future growth will come from. It triggers questions you need to find answers for.

There is a large crowd of investors who have decided that nothing matters except dividends. Between themselves they indulge in endless debate grounded in numerous errors. One such debate asks “Is it better to buy the stock with the large dividend, but low growth rate, or to buy the stock with the smaller dividend that increases at a faster rate?”.

To answer that question they produce graphs comparing the yearly $dividend increasing over time. Or they plot the cumulative dividends over time. They identify the point where the lines of the two choices cross and conclude something like “Over 20 years you receive more $$ from high dividend-growth stocks than from high-yield dividend stocks, so it is better to buy high dividend-growth stocks.” Or they conclude the opposite when the time frame necessary for the lines to cross exceeds your investing horizon.

But, but, but …

• It is an obvious error to ignore the time value of money. Simply adding up the cumulative $$ received in each option ignores the fact that a $1 today is worth more than a $1 in ten years. You can correct for the TVM by assuming the reinvestment of dividends — which causes further problems deciding what the stock price would be at that time. To see a proper graphing of dividends correctly discounted by your required return, use Sheet 2 of this spreadsheet with your own variables. The lines will never cross. The area above the lines measures the return that must come from the proceeds of disposition. (See the last point on this list).
• The high rate of dividend growth confidently assumed is most often extrapolated from historical data. E.g. “In the 10 years since 2000 XYZ company has increased its dividend at a compounding 15% rate. So we have proof of management’s commitment and can be assured of that same rate of increase going forward.” But during that period companies increased the ratio of earning paid as dividends two or three fold. During the Tech Bubble companies paid out only 27% of earnings on average. By 2009 they paid out 66%. There is no room for a repeated increase of the payout ratio in the next decade. It would result in dividends 60% larger than earnings.
• The high rate of dividend growth confidently assumed is most often extrapolated from the period of the Great Moderation when corporate profits ballooned and all was well with the world. But you cannot ignore future economic projections.
• The investor’s assumption at time=0 must be that the dividend’s growth rate will equal both the earning’s growth and the growth of the stock price. See the historical charts on Sheet 30 of this spreadsheet. Yes, the dividends paid will be moderated by a changing payout ratio …. just like a stock’s capital gains will be moderated by a changing P/E multiple. But future changes to the payout ratio and P/E are unknown and can go either way. In the long run both metrics normalize, and dividend growth and capital gains track earnings growth. Unless you are making a specific argument that the ratios for the stock you are considering will change, your assumption at t=0 must be that dividend growth will equal both earnings growth and capital gains. But the dividend-true-believers always make presumptions of dividend growth that far exceed stock price growth. They don’t see the link and wrongly consider dividend growth as certain, but capital gains uncertain. It is not uncommon to see the assumption of zero capital gains used in their analysis of concurrent dividend growth. Nonsense. Some will pay lip-service to … “I have always said that dividend growth must come from earnings growth” … obviously without understanding the implications. When you assume zero stock price growth in a period when both earnings and dividends are increasing, then the P/E ratio is being assumed to decrease at a rate equal to the earnings growth rate, and the dividend yield is assumed to steadily increase. But this is invisible because they ignore these metrics. The static stock price assumes the reinvestment of dividends at unrealistic low prices, producing false growth in the share count, and false growth in the total $dividends. These metrics they DO follow. How would they feel if growth investors used the assumption that P/E ratios steady increase? It would certainly make the growth strategy model look good. Just like the assumption of decreasing P/E ratios makes the dividend strategy look good. The only neutral assumption is that both P/E ratio and Payout ratio continue unchanged – that earnings and dividends and stock prices grow equally.
• No matter whether you prefer no dividends, some dividends, or large dividends, as long as you presume the dividend growth rate equals the stock price’s growth (by extension the growth in earnings), you always end up with a portfolio of equal size. At that later point you can buy stocks with whichever dividend yield you like. Neither choice ends up with ‘more’ dividends than the other.
• Instead of modeling reality they try to justify generalized conclusions by tracking the actual historical results of a specific stock. The chosen stock will have experienced changes to its payout ratio and P/E that were particular to itself and not common to other stocks. Even less is it rational to consider its idiosyncratic experience to be repeated by generalized stocks in the future.
• You cannot choose between companies reinvesting at different rates without measuring both dividends and capital gains. The true-believers are comparing apples and oranges when they ignore capital gains. Try an analogy … You can choose a science stream in school or a multi-discipline stream. Would it be appropriate to test all the students at graduation with only science questions and conclude one stream is smarter than the other? For each dollar of profits NOT paid out as dividends very little gets rerouted into the increased future dividend. If ROE = 10% and the dividend payout ratio is 50% then for every $1 of earnings reinvested only $0.05 ( = $1 * 10% * 50%) get paid as an increase to dividends the next year. MUCH more of the profits not paid as dividends stay inside the company and show up as capital gains. The trade off between companies paying different dividend yields is really a choice between capital gains and income. It is NOT a choice between receiving dividends now vs receiving dividends later. To compare two stocks with different payout ratios (high-yield with low growth vs high-growth with low yield) you must measure the total return — the return from dividends as well as the return from growth. Use the Different Yields spreadsheet discussed above.

The growth referred to in the media is almost always the growth of Revenue – the top line of the Income Statement. But the equity investor benefits only from the very bottom of the Income Statement – after all the expenses are paid. Any business can grow their top line. Just throw money at advertising, or undercut the competitors’ price, and revenues will grow – but not the bottom line profits.

The metrics you look at are Gross Margin, Operating Margin and Net Margin. These are calculated as a percentage of the top-line revenue. A company that is gaining benefits of scale from its growth will show improving margins. A company that is sacrificing profits for the sake of bigger market share will show declining margins.

The chart below from Deutsche Bank helps illustrate what this page is trying to say about the various components of growth. Right-click it to enlarge. It shows the breakdown of net EPS growth into it’s revenue, margin and share buy back components.

Management is allowed to fudge these metrics by removing all the revenue/expenses of certain business lines from the reported Revenue and Expense. All they have to do is decide the product line is for sale. At that point all that is reported on the Income Statement is one number – the net profit. In the notes you will find the relevant breakdown. It is up to you to restate the Income Statement with the additional revenues and expenses from these ‘discontinued operations’. Since most businesses are sold because they are doing poorly, restating the financials will result in lower margins being calculated.

Revenue growth rates are considered very important by most players in the market. Rightly or wrongly, is immaterial. Remember the Beauty Pageant Analogy. But consider:

• Don’t be suckered by short growth spurts. One year does not a trend make.
• Revenues can grow from translating foreign operations into the reporting currency at better exchange rates. The costs of sales may be exposed to a different currency. FX rates go up and down. They do not continue in one direction indefinitely.
• Profits can increase because of cost savings. Yes, companies can become more efficient with new equipment, or lower costs by outsourcing, but this should not be expected to continue. You need revenue growth as well.
• Was the growth organic – or from new business purchased? The purchase of a new business allows for many accounting problems.
  • The costs of the transaction and transition can be pushed back into the final reporting period of the old company, so that only the benefits show up in the acquirer.
  • The purchase can be paid for by issuing new shares. When they are priced at multiples of Book Value, the share sale benefit that should be calculated separately (see Comprehensive Earnings) is muddled into the reported growth.
  • The purchase price can be allocated to goodwill and intangibles that are never expensed.
• Did the business acquire its competitor because it didn’t have the intellectual property to develop its own better widget? This would not indicate future growth.

Management are all empire builders. Their salary and status increases when they increase the size of their empire. Be clear that your own ownership interest in the company (attributable to each share) may not be growing at the same rate. The company’s overall growth may be financed by additional equity. The larger pie is divided into an equally larger number of pieces. It may even be that the growth in shares is greater than the growth in profits, so that your share of the profits (EPS) falls.

Are profit margins falling even though the EPS are increasing? In this situation you are still benefiting from the whole-company growth, but the company is finding it harder and harder to grow. This problem may be due to overall industry economics, and expected to reverse in time. Or it may be because growth is peaking.

A definition of goodwill is simple. Technically, it is the portion of a purchase price paid for whole businesses that exceeds the market value of its individual assets and liabilities. But an understanding is more complicated. It is unlike, and exactly opposite to, all other assets of a company. .

Business profits reinvested (retained) in the business but spent on goodwill, are not reinvested in any productive asset. It is money given to the old owners of the business purchased. The old owners (not you) are rewarded for growing a great business. It is very common now for companies to give ALL their profits, every year, as well as the original share capital, to these external owners. The recorded goodwill can be as great, or greater, than the Shareholder Equity.

Sure the business acquired may be a great business. But not because of your own management’s work. Your company deserves no credit. Your company did not out sell, or out produce, or out smart the competition. It bought the competition – and paid full price doing so. Maybe there is an additional value to the business beyond the assets it purchased, but that is not proven just because your management paid extra. Academic studies have mostly shown that business combinations do not live up to the values claimed at their start.

Booking goodwill has two effects.

• Most importantly it prevents the cost of the purchase from being expensed and reducing Net Income. All other assets and liabilities purchased eventually work through the Income Statement. Rarely is goodwill expensed until a major disaster hits – too late. The market crash of 2008 saw dozens of companies take a big bath and write off large chunks of goodwill … because they had an excuse and knew investors would discount the expense. In reality the goodwill probably had had no value since soon after the business was bought.
• Booking goodwill on the Balance Sheet increases the value of Equity. This has the effect of making the debt seem relatively smaller. It also prevents Equity from becoming a negative value. The computers that amass company data for databases have a hard time integrating negative equity. It throws out of whack the ratios for debt-to-equity, price-to-book, return-on-equity, etc. All the computers and investors sorting stocks electronically will bypass negative equity companies.The two effects (Balance Sheet and Net Income) must be assessed together. For many financial metrics removing goodwill from the Balance Sheet reduces the divisor, but writing off goodwill from Net Income reduces the numerator also.

………………………….

How should YOU treat a goodwill write-off? The old fashioned ‘investing bibles’ are helpful here. They recommended that investors ignore the yearly fluctuation of earnings and take five year averages of the earnings and the growth in earnings. You would treat the odd ‘restructuring cost’ the same way. Average it out over the past few years.

Management always argues that any write-off is a one-time item that should be ignored because it is a past event that will not repeat in the future. But if management overpays for one purchase, chances are they will over pay the next time as well. For a more proactive approach to valuation, do not wait for the big write-off.

• Decide that the only goodwill that has any value would have been added within the last 5 years.
• Remove all the other goodwill from the Balance Sheet.
• Reduce each reported Net Income for the next 5 years by 1/5th of the remaining amount.
• And reduce the remaining Balance Sheet $$ by that amount expensed.

………………………….

One argument for leaving Goodwill on the Balance Sheet rests on the presumption (e.g. Buffet) that the Balance Sheet should measure the market value of the business (in total). It then seems reasonable that a value derived from a market transaction should be left as is, to reflect a real value. But there are problems with that presumption.

a) Companies that grow their business by sweat and tears will be expensing the costs of promotion, price discounting, etc ; all the expenses incurred to grow. None of those outlays will be stuck onto the Balance Sheet permanently as assets. Accounting should not create superficial end-runs around booking expenses for those companies who choose the easy way to grow. The sweat and tears company should not be penalized with a weaker Balance Sheet than the competitor with Goodwill.

b) The Balance Sheet does not, and never has been, meant to measure the market value of the business. The Balance Sheet measures assets and liabilities. Investors can chose to pay many multiples of that book value for the stock, so the stock price DOES measure the market value of the business. It serves no purpose to produce a Balance Sheet that is a duplicate of the stock market capitalization value.

c) Buffet argues that a high ROE proves Goodwill has value. It must have value because of the extra profits resulting from the purchase . But that argument ignores the changes to reported Net Income if Goodwill were expensed. If Goodwill has no value then it must be expensed. The reported Net Income would be severely reduced – if not zero. E.g. if Goodwill equals Equity then writing it off will restate all historical profits to zero in total. So Buffet’s high ROE disappears.

The media always represents new share issues as dilutive – the opposite of share buy-backs which are supposed to be a good thing. The simplistic idea is that the pre-existing earnings would now be split between more shares – leading to lower EPS. This reasoning is wrong because it completely ignores the proceeds received from the sale of those new shares, and the additional profits those proceeds would earn. If you don’t understand that, now might be a good time to read ahead to this section from the Understand Equity page.

What happens when new shares are issued at a premium above book value, and the proceeds are kept within the business? The reinvested premium works to exaggerate the growth of both EPS and book-value/share. The following example shows the difference between (A) a business that does not issue additional shares and (B) one that issues only 5% more. Otherwise the companies

• operate at the same 20% ROE,
• start out with 1,000 shares outstanding and
• $12,000 Equity (that is $12 bk/sh).
• Their market cap is $24,000 (twice book value) because their ROE is double the market return.

This effect can be summed up as “To the winner go the spoils”. Market leaders with the strongest stock price can easily stay leaders. By this simple mechanism a company can exaggerate its growth and earn itself a higher P/E multiple if the market is valuing their shares on that metric, or it can hide negative EPS growth if the market is pricing their shares on the Price/Bk metric. It is this mechanism that makes the stock of ‘industry consolidators’ such high flyers. Each acquisition is an excuse to issue more shares. The reinvested premiums magnify the EPS growth, leading to higher P/E multiples, leading to higher premiums on share issues – a virtuous cycle.

It works until it doesn’t. The cycle can fall apart:

• When the share proceeds are spent on goodwill (which is not a productive asset), the EPS don’t grow the necessary amount to keep the cycle alive.
• When the company has no ‘excuse’ for issuing more shares.
• When there is an industry-wide stock price correction the P/bk premium will shrink. The smaller premium leads to less of a boost to Bk/sh and EPS. The smaller growth in Bk/sh and EPS leads to a lower growth expectation and lower stock price, which leads to even lower premiums from share issues, etc, a destructive cycle.
• When the shares are issued for compensation less than market value, i.e. the exercise of stock options.

You can get a handle on the extent to which share-issue-premiums play a role in growth by reconciling book-value-per-share beginning of year, to end of year. You can see this in the spreadsheet for Company Analysis. There is a box called ‘Book value Reconciliation’.

A country’s Gross Domestic Product growth (GDP) will always reflect the economy in which business operates. Profits will always come easier in a booming economy and dry up in a recession. But it is wrong to conclude that there is a absolute link between the growth of earnings and the economy. There are a number or reasons.

• GDP measures the income earned by all the sectors of the economy – the government taxes, the employee paycheques, the debt holder’s interest, as well as the profits to equity owners. The portion accruing to each sector is not static. E.g. at some times labour competition from Asia drives down paycheques – at other times union bargaining drives up their share. E.g. in an economy with excess industrial capacity interest rates will be low – in a booming economy the high demand for borrowing raises rates. E.g. in periods of unexpected high inflation, the corporation may not be able to raise prices at the same rate its wages increase. The common share owner only gets what is left after wages, financing costs and taxes are paid.
• The corporate profits’ allocation of GDP is shared between both large and small companies – both public and private. Growth is not necessarily shared between them is a static ratio over time.
• Inflation impacts GDP and earnings differently. The GDP deflator is not the same measure of inflation as the Consumer Price Index CPI. For example, if raw material prices are rising but low wages prevent retailers from raising prices (because of low consumer demand), there can be a mismatch. Also, profits from foreign countries are impacted by that foreign inflation rate. Investors measure their returns relative to their home country’s CPI. Companies adjust for unexpected inflation in the long run, but not the short run. High nominal GDP growth because of inflation may cause low company profits.
• GDP measures the economy within a country’s borders, but multinational corporations’ profits are generated worldwide.
• Growth in GDP or company profits can be financed by re-investments from foreigners as easily as from natives. Some countries have restraints on capital flows, but in total money flows around the world as it wishes.
• Company earnings (and their growth) are measured according to accountants’ rules. These differ from the methodologies of economists measuring GDP. One example would be compensation with stock options. Another would be amortization of long-life assets.
• Growth in stock index earnings are essentially a ‘per-share’ measure. But companies also grow in total size by issuing additional shares and debt. GDP more closely measures the latter. This is the same distinction addressed in the page Comprehensive EPS. There is no ‘dilution’ of growth as some authors claim. Be clear what you want measured in different circumstances.
• Growth in company earnings per share can be manufactured by stock buy-backs. The cash movement from within the operating business to the secondary market shrinks the business and earnings company-wide. When comparing growth rates both dividends and share-buybacks must be removed in the calculation of the assumed earnings reinvested.
• Mergers and acquisitions also move cash from inside businesses to the secondary markets. This may or may not be offset by Initial Public Offerings (IPO’s) drawing in new investment.

The following graphs come from Sheets 11 and 16 of the data spreadsheet.

Another false comparison is between GDP growth and stock total-returns. Even Pimco’s Bill Gross makes this doosie of an error in this piece. There is a difference equal to the portion each year that is ‘consumed’ (i.e. not reinvested to generate future growth).

Part of a stock’s total return come in the form of distributions that either fund the owners’ consumption or get reinvested in the secondary market. This portion does not create future growth in the productive business. Similarly, part of each years’ Gross Domestic Product gets consumed when food is eaten, clothes wear out, health services still result in death, and leaky houses rot, etc. You must only compare apples with apples. Only a comparison of stocks’ price-index (with dividends removed) vs. GDP is appropriate. Or even more appropriately compare stock-earnings vs. GDP.

A normal person would think that companies buying long-term assets are more likely to subsequently realize higher stock prices because the assets work to produce larger profits. They would be wrong according to backtesting. High rates of asset growth (ΔA) are negatively correlated with future stock price returns. This Investment metric has now been included in the most recent Fama-French 5-factor asset-pricing model. Although tempting to use the results to conclude that management should pay dividends instead of reinvesting, or that growth stocks are poor investments, or that capital-intensive businesses are bad investments, etc, better to be clear exactly how little is actually proved.

In 2014, after much of the academic work was done, Fangjian Fu (Dissecting the Asset Growth Anomaly, 2014) has found two explanations for this anomaly.

1. The first is an error in the dataset. He found that the CRSP dataset does not include the monthly returns of stocks delisted in the month, 90% of the time. Since the returns of delisted stocks tend to be very negative (-38% in his sample), omitting them introduces significant survivorship bias. The companies shrinking their assets were often in this group so their returns were actually much lower than reported.
2. He found the remaining anomaly was explained by the large external financings of new debt and equity assumed by those companies growing assets the fastest. Their poor returns were not so much due to the increase in assets, but due to the increase in financing.

In 2015, Hongtao Li (A Closer Look) looks harder at that second point. He finds that the negative correlation between stock returns and Investment is much stronger when only Investment funded by external financing is considered. This factor co-varies strongly with the Fama-French value factor and even subsumes its explanatory power, so he modifies his factor by scaling the Investment to the companies’ size to get rid of that effect. His Externally-Financed Investment metric has a much stronger predictive power than the traditional ΔA. He offers the logic that companies know when they are over-valued by the market, and time their external financing to exploit the mis-valuation. So future poor stock performance is simply the market value returning to its mean, and not a result of Investment.

He found that Internally-Financed Investment (from retained earnings) has no predictive power. Internally-Financed Investment is driven by companies’ profitability, which vary over time, per common sense. Its correlation with stock returns is positive, not negative, moving in the opposite direction to Externally-Financed Investment.

Other problems with the historical studies of ΔA ……

• Stock returns over what period of time? It would seem reasonable to expect poor earnings and resulting poor stock returns shortly after big asset purchases. There are the costs of integrating different cultures, start-up and training costs of new equipment, write-offs found once the books of businesses bought are fully examined, increased depreciation charges, increased financing costs and stock dilution. But you would expect the earnings and stock returns to recover quickly in later years and reflect the growth. Studies don’t agree with each other about how long the negative correlation of ΔA lasts. Cooper et al. vs. Watanabe et al.
• What stocks are significant in producing this generalization? Jiang and Zhang (2011) show that the ΔA effect is produced only by the most extreme 10% of growth companies – not your run-of-the-mill growing company and not by low-growth companies. In contrast, Cooper et al. (2009) found excess returns growing proportionally between the deciles. Watanabe et al. (2011) showed the ΔA effect can also be generalized to show up more in smaller companies, more in the US than in other countries, more in countries with established, liquid, efficient stock markets.

………………………….

Reasons for the existence of this negative correlation. There is disagreement among academics.

1. There are two lines of reasoning favoured by academics. One reasons that more and more investment, means that projects with lower and lower returns are undertaken. This is called a ‘decreasing return to scale’. Since stock returns will track investment returns, you would expect lower stock returns when projects of lower profitability are undertaken.
2. The other line of reasoning argues that companies with a low cost of capital (ie. high stock price) can/will undertake less profitable projects than companies with a high cost of capital. The undertaking of those low-profit projects leads to the first point … lower investment returns and lower stock prices.
3. The poor returns from high-asset-growth stocks may be a return-reversal caused by systematic over-pricing of high-growth stocks. This was found by Cooper, et al. (2009).
4. It may be that the poor stock returns after high asset growth are because risk has increased. Nyberg and Poyry (2011) provide strong evidence that ΔA is one of the key drivers of stock price momentum. They first sort their database by ΔA and then measure the excess return from momentum for each. They find momentum returns much higher in the high-asset-growth stocks.
5. Another evidence of the ‘increased risk’ explanation is found in Zhou and Ruland (2006). They look for factors explaining future Earnings Per Share growth. They find that Asset Growth is a stronger factor and more statistically significant than the Dividend Payout Ratio. If ΔA goes hand in hand with growing EPS and lower stock returns, then the market valuation of those Earnings must have fallen. The market lowers P/E when it perceives increased risk.

………………………….

With so much contradictory evidence and explanation, you should ignore all this. You can also find studies showing no statistical effect Hwang, Lui (2012). And most certainly remember that ‘correlation is not a proof of causation’. If poor stock returns are found after large capital investments, that does not prove they are caused by those investments. Both effects could be caused by over-valuation of the stock at the beginning. You cannot conclude from these finding anything about ‘what management should do’.

The major problem with the stylized model of ROE predicting growth is the assumed return on reinvested profits, and the presumption that earnings will grow faster when more profits are retained for reinvestment. Arnott and Asness (2003 “Surprise! Higher Dividend (Payout Ratios) = Higher Earnings Growth”) correlated the US stock index’s dividend payout ratio with the subsequent 10-year growth rate of the index’s earnings. They found that periods of large payout ratios (in general by all companies) presage large increases in earnings growth, and vice versa. Their Figure 1 charting the similarity over time is very convincing.

Do their findings prove that profits are better distributed to shareholders as dividends than reinvested for growth? That is the conclusion reached by pundits promoting dividend investing. That is the conclusion the authors clearly had in mind from the start. But NO. This study looked at how the general market changed over time. It did not distinguish between companies paying different payout levels. The same companies behaved differently at different points in time. The same companies realized different growth rates at different points in time. There is a big difference between cross-sectional studies that compare different stocks in the same period, and inter-temporal, or time-varying studies that compare the same stocks (or indexes) in different periods.

There are several interpretations of their results possible.

One explanation is the most obvious – while dividends are sticky, earnings are volatile. A temporary low earnings will create a high payout-ratio metric that self-corrects when earnings revert to normal. The authors claim to disprove this explanation using regression analysis. But earnings have been shown to revert as discussed above in “History As A Guide?”. And the results of regression analysis can be driven by outliers.

S&P; data for which both future-5 and past-5-year-earnings are available (from 1926 to 2006) can be charted against payout-ratio. When the charts are compared it is clear that it is the outliers that generate the statistical conclusion. Those outliers of high payout-ratios correlate with past negative earnings and those reverse to positive future earnings. The explanatory value of payout-ratio vs. past earnings (R-squared 37 percent) was far higher than of future earnings (R-squared 7 percent). An R-squared of 7% is not a meaningful relationship. You have to question what ‘adjustments’ the authors needed to make to their data to generate their claim of an R-squared = 54%.

A second possible explanation is that management are good predictors of future earnings. It can be argued that they should be because of their inside information. And there are arguments that the astute timing of management options proves this. The authors conclude that this explanation is possible and has not been disproved by their data.

A third possibility is that the economy and public companies go through cycles of high animal spirits and high investment. The authors’ figure 4 shows a very tight correlation between high GDP level of investment and high investment by public companies (low payout-ratio). Expansion for future growth requires a leap of faith. This feeling of assurance is shared throughout society at certain times, leading to all participants investing at the same time. To call this ’empire building’ is to insult their gumption. Management cannot accomplish the investment by themselves. They must convince the lenders and Boards of Directors. They must feel their stock will not be punished as a result of lower free-cash-flow. They need the animal spirits of society. This interpretation is validated by the 2013 work of Arif and Lee

Because everyone is doing the same at the same time, the expected result is excess capacity, leading to the classic business cycle recession. That reality is not a reason to NOT invest in the first place. It is not a justification for concluding the money is put to better use by investors in the secondary market. That is not a reason to characterize retention of profits as “inefficient empire building”

………………………….

Instead of looking at aggregate metrics changing over time, the question is better addressed by comparing different companies’ earnings growth in the same specified time. Zhou and Ruland did this in 2006 (“Dividend Payout and Future Earnings Growth”). Their results also find a positive correlation between the dividend payout ratio and earnings growth over the subsequent 1, 3 and 5 years. They conclude, essentially, that this proves profits should be paid out to owners because management just wastes it.

But their results can be concluded differently. Their Table 8 summarizes the findings. Yes, the Dividend Payout Ratio has a large t-stat showing it significant in explaining future Earnings Growth. But the most important factor is concurrent Asset Growth. The authors make no comment but it is just common sense. You need assets to generate profits – more assets will generate more profits. The strength and direction of the t-stat puts the lie to the author’s conclusion. Are they trying to say that “yes, management can effectively put assets to work …. as long as the assets are not funded by retained earnings? Ridiculous.

The second issue in their results that they fail to address is the surprising strength of Return On Assets to predict poor Earnings Growth. Why on earth would this be so? ROA measures how effectively assets generate profits. Since the metric Asset Growth correlated with good Earnings Growth, why would ROA have the opposite effect? and why so statistically strongly?

There is only one reasonable explanation for these two un-explained results …. that earnings are cyclical, even in the short term. Earnings are within the metric ROA and the Dividend Payout Ratio and the Past Earning Growth metric. If earning cycle (not necessarily revert-to-mean) then the direction of the t-stats are fully explained and agree with the ROA t-stat.

The authors claim to discount the possibility of earnings’ reversal-to-the-norm. But their methodology is not persuasive. E.g. they think that including a factor for V/A removes any taint from Earnings cyclicality from the Dividend Payout Ratio’s metric.

This paper found that the Dividend Payout Ratio was just one of a broad selection of factors of greater and equal significance. Correlation does not prove causation. Their conclusion regarding management’s waste of assets is not justified by their finding. Instead the data include totally unexplainable findings and tantalizing findings that were ignored.