How to Think About Retained Earnings

Author's Avatar
May 31, 2012
Someone who reads my articles asked me this question:

Hi Geoff,

I'm assembling the financial statements for various companies that have done well over the past years to try and learn and actually see their characteristics for myself. Other than understanding the business economics and making the calculations, I'm trying to follow the money through the statements but I'm having trouble seeing the redeployed capital from retained earnings when other items cloud the picture.

In your article "How Does Warren Buffett Value Growth?" you approximated the reinvested capital, $1.35. Is there a specific way to determine this? I can calculate the various return formulas now, but I need to see how capital gets redeployed as well as the rates earned on that capital. If there is no capital stock purchased then the increase in retained earnings works out, but if there is, the numbers don't. I have attached the Fastenal Company financials (its 10-K has an error in the 2006 fiscal year data that the auditor did not pick out, but this is not important because the error resulted in an understatement). I also have the book "How to Read a Financial Report" by John Tracy to help with the flows, but at the moment, I can't see it and would be grateful for some guidance.



You ask a great question. It’s not easy to answer. I track receivables, inventory, PP&E, accounts payable, and accrued expenses for every year of the last 15 years (if I can get 15-year data). I use an Excel sheet to compare these numbers to sales, EBITDA, etc. So, if receivables have risen faster than sales – then that could be where the “reinvestment” is going.

If you have, say, 15 years of data – and you can get this on EDGAR if you’re willing to directly type stuff from EDGAR into Excel by hand – you can do a better comparison. But if you want it to be easy – you can use GuruFocus, Morningstar, etc., to do a shorter comparison.

Just look at:

· Net Income

· Free Cash Flow

· (Buybacks + Dividends)

Some companies – like Omnicom, Dun & Bradstreet, etc. – will say “we’ve returned 95% or 90% or 80%” or whatever of earnings to shareholders over the last 10 years. This is often an overstatement of sorts. Because they’ve increased debt. I can pay you back 500% of earnings if someone is willing to lend me $5 for every $1 I earned this year. Usually, we don’t want to count stuff like this.


Let’s look at Dun & Bradstreet (DNB, Financial). Share buybacks and dividends were $2.93 billion versus $2.43 billion in reported earnings over the last 10 years. Obviously, they increased debt. I’ll lump debt and pension shortfall together here. Net financial obligations (that is, debt and pension gap minus cash) was $637 million in 2002. And it was $1.68 billion today. (These are approximations – I’m not looking at EDGAR right now, I’m using the GuruFocus page. If you really do this yourself, check EDGAR. Because I’m using the “other long-term liabilities” line as pension gap here – and I don’t think all of it is really the pension fund. Anyway...) So debt increased by $1.04 billion. Let’s take that out of the $2.93 billion payback (because it was debt financed, not business financed). That leaves $1.89 billion that was paid out to shareholders without using debt. And reported earnings were $2.43 billion. So, $1.89 billion divided by $2.43 billion equals 78%. DNB pays out around 78 cents of each dollar.

They retained about 22 cents of each dollar. And these 22 cents that were retained – remember, they really weren’t retained at DNB, they just borrowed money – were able to grow net income by 6.9% a year (from $143 million to $260 million over nine years). Sales grew slower. Just 3.6%. The sustainable rate is probably closer to 3.6%.

But let’s think about earnings. Say, you had a dollar of earnings at DNB. Next year – if they grow net income by 6.9% a year – you’ll have about $1.07 in earnings next year. And DNB will pay you 78 cents in stock buybacks and dividends. So, that 7 cent increase came from 22 cents of retained earnings. That’s about a 32% return on retained earnings. If we assume the sales number is more accurate – we don’t give them credit for margin expansion in earnings growth – then return on retained earnings is just 18%. Either number is good.

For comparison, let’s look at Omnicom (OMC, Financial). Over the last 10 years, they’ve had cumulative dividends and buybacks of $7.94 billion. Cumulative net income was $8.25 billion. Net debt increased by something like $1.8 billion. Again, this probably isn’t right. Check EDGAR. I just treated all long-term liabilities as debt – which is probably wrong. Anyway, that means that we had 74 cents of each dollar paid out without the use of debt. Again, the actual payout was higher – but if they borrowed from the bank and paid me a dividend, I don’t count that. So, 74 cents a year is what Omnicom pays out to shareholders without increasing debt. That leaves 26 cents in retained earnings.

Net income rose by 4.4% a year. Sales rose by 7.2%. I think net income is a little depressed right now. And the real number is somewhere between 4.4% a year growth and 7.2% a year growth in normal earning power. But we’ll use 4.4%.

So, same idea, for every $1 of EPS OMC has today we assume they will have $1.04 in EPS next year. And they need to retain 26 cents to do that. Well, 4 cents divided by 26 cents is 15%. So they are earning about a 15% return on retained earnings. Again, this is a guess. But it gives you an idea of what they are earning over time.

Look at the change in net income and sales over 10 years and then the ratio of cumulative buybacks and dividends to cumulative reported earnings.

We’ll try this for a totally different company – Carbo Ceramics (CRR, Financial).

They had cumulative net income of $568 million over the last 10 years. And they paid out $152 million in stock buybacks and dividends. That means they are paying out 27 cents for every $1 they earn. Or looking at it the other way – they are retaining 73 cents for every $1 they earn.

Net income rose by 23% a year over the last nine years. Sales rose by 20% a year. We won’t assume any margin expansion – so let’s take the 20% sales number as normal.

This means that Carbo tends to earn $1.20 next year for every $1 it earned this year. And for every $1 Carbo earned this year, they tend to pay out 27 cents. On the part they retain, their return is 20 cents divided by 73 cents. Which is a 27% return on retained earnings.

Again, this is an estimate. Not a calculation. I’m not looking for a specific formula. I’m looking for what the central tendency of return on retained earnings has been. And I’m not worried about whether it is 23% or 21%. I’m worried about whether it is 5% or 15% or 30%. Is it a bad business, a good business or a great business.

Carbo (CRR) is good verging on great. Omnicom is good – definitely not verging on great. But Omnicom buys other agencies. Acquisitions like that give you more room to deploy capital – Omnicom has a lot more runway than DNB – but it also means you are unlikely to get spectacular returns on capital. At some price, the sellers simply won’t sell. One product companies have a much easier time of achieving very high returns on retained earnings – but they also have a tough time expanding indefinitely.