Capitalizing R&D

Industrial companies reinvest in plant, property, and equipment to grow revenues. Technology companies grow by investing in research and development. However, on an income statement, these R&D investments are treated as an operating expense (instead of a capital investment).

One can reorganize financial statements of such companies by capitalizing R&D expenditures. You can check out this video, and this spreadsheet (look for R&DConv.xls) from Aswath Damodaran to understand the logic and mechanics of this reclassification.

The R&D asset you create goes to the asset side of the balance sheet.

Since assets = equity + liabilities, this increase in assets leads to an increase in the book value of equity, and hence the invested capital.

Similarly, we adjust the operating earnings by adding back the R&D expense and subtracting the depreciation of this asset in the current year.

When we capitalize R&D, we get a more authentic view of the earnings, reinvestment, and returns on capital. This alters the fundamental inputs that go into a discounted cash flow valuation, including earnings, growth and reinvestment rates (sales to capital ratio), and operating margins.

Process

To begin the process of capitalizing, we need the following inputs

  • An amortization period, N years, over which R&D is expected to deliver results (software ~ 2 years, hardware ~ 3-5 years, pharma ~ 10 years; Damodaran’s spreadsheet has some numbers for guidance)
  • Collect R&D expenses for the prior N (or N+1) years from the income statement. You can get these from company filings or a data service like Morningstar.
  • Create a table (see spreadsheet) to determine (i) the net value of the R&D asset on the balance sheet, and (ii) the current year amortization number.
  • Recompute operating earnings, net income, invested capital, and reinvestment rate.
  • Use these numbers to inform inputs in the DCF analysis

Example

I took Damodaran’s spreadsheet, and modified it slightly to account for partial year data. I did this to understand the spreadsheet better; not necessarily because I think such an adjustment is important. Of course, it uses the latest numbers, so it has that thing going for it.

For prior years, I assumed that R&D expenditures were distributed evenly throughout the year. Thus, if $100M were spent over an year, I assume $25M were spent each quarter. This spares me the burden of having to deal with quarterly filings.

I used this modified spreadsheet to analyze CSCO after two quarters of fiscal 2017.

1

Cells in yellow are inputs, while those in green are computed. This yields the following calculation for the value of the R&D asset and the amortization.

2

It also yields some summary statistics. In CSCO’s case, capitalizing R&D did not have a big effect on (i) operating income/margin and (ii) net income/margin. It had a modest effect on the reinvestment rate, which increased. The amount of capex, invested capital, and depreciation increased dramatically, while the return on capital went down modestly.

Valuing Financial Services Companies

Financial services companies are not quite amenable to a discounted cash flow analysis, because there is no wall separating operating and financial assets.

In such cases, it is difficult to figure out an ROIC, because “IC” in this context is a nebulous concept. This problem pops up in one form or another, when one attempts to do a firm level valuation (debt + equity). For example the sales/capital ratio which is handy in modeling reinvestment for growth is not particularly meaningful.

One way out of this quandary is to focus on equity (from Damodaran), and use a (potential) dividend discount model based on the Gordon growth model.

Let us define the relevant terms.

BV = book value of equity
ROE = return on equity
COE = cost of equity
nNI = normalized net income for next year = ROE * BV
g = stable earnings growth into perpetuity

The amount of free cash is determined by the g and ROE. This free cash can be distributed as a potential dividend. The payout ratio of this “free cash dividend” is

p = dividend payout ratio = (1 - g/ROE)

As a reality check, it is useful to compare the historical payout ratio to this value of “p”. One might have to account for all forms of cash return including dividends and buybacks while doing this.

Thus, the value of the equity is:

Equity IV = nNI * p/(COE - g)

Of course, if it makes sense, then the future can be split into explicit stages in which variables vary with time before settling into their terminal values.

On ROIC

ROIC and growth are the two central drivers of value.

ROIC = NOPAT/IC

The numerator comes from the income statement. The denominator comes from the balance sheet.

NOPAT is net operating profit after taxes. It is often modeled as EBIT (1 – tax rate). While depreciation is a real cost, the amortization of intangibles is often added to adjust EBIT. Thus,

NOPAT = EBITA (1 - tax rate)

There is a fair amount of subjectivity in defining the denominator IC (invested capital).

It can be obtained from the asset or operating side of the balance sheet (recommended), or the right or financing side. I found this position paper from Credit-Suisse very useful. The authors present practical tips on how to think about IC, instead of getting caught up in some particular formula.  An illustrative example (CSCO) is used to animate some of these ideas.

Screenshot from 2017-05-08 08-56-19Essentially, IC should include all the “capital” (inputs that generate revenue over long time frames).

To the first approximation,

IC = total assets - non-interest bearing current liabilities

It definitely should include current assets, net PPE, and other operating assets. One can make several commonsense adjustments:

  1. If a company carries excess cash or marketable securities (over that required to run the business), then that should be excluded from IC. A recommended rule of thumb is cash equal to 2% – 5% of sales (ranging from mature to growth companies) are required in the running of the business. Any excess should be excluded from IC.
  2. If M&A is part of the company’s modus operandus, then one should not exclude goodwill, as it represents a true cost of doing business.
  3. Capitalize leases and R&D, since they have characteristics of debt and long-term assets, respectively.

Here are some other resources on ROIC that I found useful.

This paper by Damodaran, “Return on Capital (ROC), Return on Invested Capital (ROIC) and Return on Equity (ROE): Measurement and Implications”, is quite readable, and presents some useful insights.

John Huber has some of the most well-articulated thoughts on ROIC. He has written several articles (compounding and high ROIC, and legacy versus reinvestment ROIC), which can be found on his website here.

This Bears-Stearns presentation on the role of ROIC in valuation has been floating around in a lot of different places.

Qualified Covered Calls

Early last year, I bought CFR around $53, valuing it at around $75. As it rose to $70, I wrote a $75 covered call expiring in 45 days. CFR promptly climbed to $80, and my call was ITM.

Since the stock had risen above my target price, I did not mind selling it.

However, I reckoned that if I could somehow hold on to the stock for couple more months, then I would avoid short-term capital gains tax on my profits. At that time, I naively thought that I would keep rolling over my $75 call, and only let it get exercised after I had owned it for 12 months.

It turns out that the rules governing deep in the money covered calls are somewhat complicated.

If you write a call sufficiently deep in the money, you are no longer considered an owner of the stock. This actually makes sense, since you have given away most of the pain, and the gain, associated with the stock’s gyrations that a true owner would feel.

The key phrase here is “qualified” or “unqualified” covered calls. If your covered call is qualified, then no problem.

If it is unqualified, the clock that determines the holding period for the stock stops. It resumes, once the call is closed or replaced by a qualified covered  call.

We need to know five numbers (two of which may not be needed, depending on the situation).

P = stock price at the end of the previous trading day
S = strike price of the covered call
d = number of days till expiry of the covered call
SP-1  = first strike below P 
SP-2  = second strike below P

For example, yesterday – May 5, 2017 – AAPL closed at $148.95. On the next working day (5/8/2017), suppose I consider writing covered calls expiring on May 19 (d = 12 days to expiry), at a strike price of S = $150. The two strike prices available below P are $148 and $149.

To determine if a covered call is qualified, I run the five numbers through the following flowchart.

qualifiedCC

Out of the money calls are always qualified. In the AAPL example above, the call is qualified because the call OTM.

If calls are ITM, they have to have sufficient duration (at least 30 days), and cannot be too deep in the money.

Averaging Down

In 2011, Radioshack was trading at a single digit P/E. It was participating in the “mobile” wave, and sported a superficially healthy ROE. I bought the stock at $15. As the stock price declined, I took a second bite at $6. I thought the market was overestimating the odds that RSH would be zero. Saj Karsan, a blogger who I respect, seemed to think so.

Not long after, I unloaded the entire position at prices between $3-$4. I lost about 2/3 of the originally invested capital.

In 2012, I bought Prosafe SE (PRSEY) at $7.50. Its historical stability, and high dividend seemed attractive. Adib Motiwala, a money manager I followed, had endorsed it. As oil prices declined, I took another bite at ~$4.50. By the time, I realized that oil prices might not snap back right away, and sold my position, I had taken another haircut worth 2/3 of the capital deployed in the position.

Finally, Weight Watchers. I bought it over 2013-2014 at an average price of around $24. It had fallen nearly 70% from a high of $80 in 2012. I did not know much about WTW, before Geoff Gannon who writes a superb blog, mentioned it. He and Quan put out a compelling writeup, explaining their case. In 2015, WTW slipped below $4/share – an 85% loss on my buy price. Since then it shot up to $26 (where somebody smarter would have sold), and retreated back to $12. I am currently sitting on an unrealized loss of  ~50%.

I thought about these three investment mistakes, as I read John Hempton of Bronte Capital share his thoughts on averaging down a stock. There are useful lessons to distill from his post, and my painful experiences.

In all three cases (RSH, PRSEY and WTW), leverage was a serious issue. Leverage can make a stock go to zero, if the underlying business falters, even temporarily.

Lesson #1: Don’t buy a levered business in decline

In all three cases, the endorsement of someone I admired made me more reckless than I would otherwise have been. No, I don’t blame these people for being wrong; on the contrary, I am thankful that they share their good and bad ideas so freely.

Lesson #2: Ask “am I buying this primarily because someone I admire likes it?”

In the case of RSH and WTW, pressure due to new technology (Amazon and MyFitnessPal, respectively) was a contributing factor.

Lesson #3: Don’t buy businesses that can be rendered obsolete by technology.

Many/most businesses are susceptible to technological obsolescence, even if we can’t see it at the time. For some, we can can see – in relatively high-resolution – how its core will be disrupted in a short period of time. Those are the bets to be cautious about.

Greenwald Formula

I’m nearly through Bruce Greenwald’s “Value Investing” book.

The valuation framework discussed in the book (net asset value, earnings power value, etc.) is both philosophically and quantitatively different from the DCF calculations that I’ve been practicing off of Aswath Damodaran‘s book and notes.

In his book, Greenwald lays out a simple relation between value, profitability, and growth. This is an extremely useful check on intuition.

Earnings Power Value

Earning power value is simply the adjusted after-tax earnings (E) divided by the cost of capital (COC).

EPV = \dfrac{E}{COC}

The earnings are typically adjusted for maintenance capex, non-recurring expenses etc. EPV is a no-growth estimate of the value of the firm. For an example, consider this recent valuation of National Oilwell Varco.

The return on capital (ROC) is defined as the ratio of the adjusted after-tax earnings (E) and the capital employed (C). Thus,

E = ROC \times C

Thus, we get:

EPV = \dfrac{E}{COC} = \dfrac{ROC}{COC} \ C

Consider a business capitalized with $100m. If the COC is 10%, then the EPV > $100m, if and only if ROC > 10%.

Growth

Now consider a special case of a firm growing indefinitely at a growth rate G. For simplicity, assume that revenues, capital employed and earnings are all growing at the same growth rate. That is, the ratios (net margin, ROC etc.) remain stable during this growth period.

Growth in earnings doesn’t come free. Part of the earnings (E) have to be reinvested (I) for growth, which results lowered cash flows (CF) that can be extracted from the business.

CF = E - I

For earnings to grow by G*E, the amount of capital employed has to be increased by G*C, assuming ROC remains stable. This is the reinvestment required (I = G \times C). Thus,

CF = E - I =  ROC \times C - G \times C = (ROC - G) \ C

The present value of a growing annuity is given by the Gordon Growth Formula:

PV =  \dfrac{CF}{COC - G} = \dfrac{ROC - G}{COC - G} \ C

Mathematically, if ROC > COC, growth enhances the value of the firm (PV > EPV).

Value of Growth

One way to understand the value of growth is to consider the ratio of the growth scenario to the no-growth scenario. It is straightforward to show that:

M = \dfrac{PV}{EPV} =   \dfrac{ROC - G}{COC - G} \ C \times \dfrac{COC}{ROC} \dfrac{1}{C}

If we represent g = G/COC , and r = ROC/COC, then we have:

M = \dfrac{1 - (g/r)}{1 -g}

If r > 1, this formula suggests reinvesting as much as possible (assuming incremental ROC = historic ROC) back into the business.

The figure below shows a contour plot of M(g, r). When r = 1 (ROC = COC), M = 1 regardless of growth. In this case, growth is neither good nor bad.

If r < 1, growth is actually bad. In this scenario, it might be best if the firm pays out all of its owner’s earnings. Divestment or -ve growth might also be shareholder friendly (see SPLS example).

growth

If r > 1, then growth is economically good.

Example:

Consider a bank like Wells Fargo, with a historic ROC = 15%, a historic COC ~ 10%, and a growth rate of nearly equal to the COC (say G = 9% to keep the Gordon growth formula from blowing up).

Here g = 0.9, and r = 1.5.

The value of growth to WFC is  nearly a factor of 4.

Outperforming Indices in Different Markets

Consider two scenarios in which the stock market index compounds at different rates:

  1. robust 10% annual return
  2. tepid 5% annual return

Now suppose you had access to a genuinely good asset manager (like Buffett) who guarantees a 1% out-performance of the indices, and charges no fees.

Clearly more is better than less in this case. So you should hire him or her!

But the purely academic question here is the following, “are the effects of outperforming the indices more valuable in robust or tepid markets?”

I was interested in this question, because I hear some murmur which suggests that active management might work better than passive indexing, when the market is expected to compound tepidly.

One can do a quick calculation to find out. Suppose that the holding period is n = 10 years.

At 10% market return, $1 grows to $2.59 in 10 years. At 11%, it grows to $2.84 – a nearly 9.5% out-performance.

At 5% and 6% returns, $1 grows to $1.63, and $1.79, respectively. This is a nearly 10% out-performance.

So technically it is true: out-performance is more valuable in tepid markets.

But in practice, no “alpha”-generating asset manager is going to work for you free. The extra value (note this 0.5% is not 0.5% compounded) is probably not enough to justify switching investing strategies from passive to active simply based on tepid expected forward returns.

Math Aside

To understand this apparent insensitivity to market return, lets bring out some low-powered calculus.

Let r be the market return, and let \Delta r be the manager out-performance.

After n years, the market compounds $1 to f(r) = (1 + r)^{n}. If \Delta r \ll r, we can approximate the out-performance f(r + \Delta r) by its Taylor series:

f(r + \Delta r) \approx f(r) +\Delta r\ df/dr

Taking the derivative, and rearranging terms, we get,

\dfrac{f(r + \Delta r)}{f(r)} \approx 1 + \dfrac{n \Delta r}{1+r}

The percentage outperformance is controlled by the last term on the RHS. The factor “n” just shows that any incremental annual outperformance is essentially additive. The remaining term \Delta r/1 + r does have a r in it, which means that technically it does matter whether the market is robust or tepid.

However, in practice, 1 + r \approx 1. It turns out that the percentage n \Delta r/1 + r \approx n \Delta r, and the state of the market effectively drops out of the equation.