I recently had a constructive debate with my friend about the return on invested capital (ROIC). I said that we don’t calculate ROIC for fun; we calculate it to know what return retained earnings can make. High return on retained earnings means good business. He shot back that See’s Candies has little volume growth and it’s still a good business. His point led me to the broader topic of what a good business is.

In a nutshell, a good business can create value. In other words, it can generate more than 10 cents for each $ of earnings it retains – assuming a 10% hurdle rate. But there are special cases in which a company can make more profits by retaining zero or negative earnings.

**Exceptional Businesses Have Negative Invested Capital or Pricing Power**

One special case is negative invested capital. Omnicom is a good example. It pays for advertising spaces slower than it bills clients. Working capital is about -20% of sales. The negative sign means that Omnicom gets 20 cents pre-funding from clients for each additional $ of sales. If growth is stable, a business with negative invested capital deserves a higher than average multiple of EBIT.

Another special case is exceptional pricing power. See’s Candies has exceptional pricing power. From 1972 to 1998, See’s Candies raised price per pound by about 6.9% annually. Inflation over this period was about 5.4%. So, pricing power generates about 1.5% real growth each year. That leads to margin expansion. This magnitude of pricing power is rare because the product becomes more expensive relative to a customer’s purchasing power overtime. That’s not sustainable in most cases. But See’s Candies has been able to do so for many years.

Another good example of pricing power is luxury Swiss watches. Swiss watchmakers managed to reposition mechanical watches from a utility product to an emotional product. But after that repositioning, it’s difficult to raise price faster than inflation. To do so, a brand must move upmarket and become more exclusive. Omega, on the path to regain its past prestige, has raised price from Longines’s price range closer to Rolex’s price range.

Without exceptional pricing power, value is normally created through volume growth. Volume growth normally requires additional investment in production/service capacity and working capital. Value is created only if return on investment is high.

**How to Calculate ROIC**

A practice that many analysts use is to say that a business is good if it consistently make a high ROIC. Joel Greenblatt’s formula for ROIC is EBIT/NTA. NTA is **N**et **T**angible **A**ssets, which is the sum of net fixed asset and net working capital. There are several versions of ROIC. But all version use net fixed assets in calculating the denominator. And that creates some controversies.

Joel Greenblatt explained why he uses net fixed asset:

“*Why are we taking Net Fixed Assets (NFA)? It is not always right. Say we buy a hotel for $10 and it is going to last 10 years and we write it down over 5 years and now it is at $5. But if this goes down to zero, I might have to invest another $10. This would give me ($5) a skewed return (being too high) because of not considering replacement and reinvestment into the fixed assets.*

*Say you have 100 hotels and they are all on different cycles, then on average, you will be correct in using NFA. 10% of your hotels will be refurbished each year over a 10 year normal cycle. That is my quick and dirty for an ongoing business.*”

And,

“*Denominator is NWC + NFA--why using net and not gross fixed assets? On average that is the right thing to do. Because in general what happens to your fixed assets, you buy something and you depreciate the assets so the value of your asset goes down, but to maintain your asset, there has to be on-going capex. Depreciation and Capex cancel out (assume Deprec = Maint. Capex). If capex is more than depreciation, then FA will increase accordingly and you will be updated. If you are in expansion mode, you build new stores and the FA balloon before you earn on those assets, so your ROC will decline--so you must normalize or adjust for that. Fixed Assets minus depreciation plus Maint. Capex is why I use a Net number.*”

There’s some logic in his argument. But he didn’t examine how accurate EBIT/NTA is as a measure of ROIC for an ongoing business. If we own the 100 hotels in his example, we get cash flow roughly equal to EBITDA each year (assuming no tax). 10 hotels are totally depreciated each year. We can choose not to make any refurbishment at all and let EBITDA decline by 10% next year. We can refurbish 10 hotels and maintain EBITDA. Or we can refurbish and build 10 more hotels to grow EBITDA by 10%. In either case, ROIC of each new build or refurbishment will be based on the $10 **gross** investment in each of these projects because that’s what we have to spend upfront.

Let’s take another example. The Fresh Market (TFM) spends about $4 million in a new store, which generates about $10 million sales and $1 million EBITDA. TFM remodels its stores every 10 years. The remodel cost is lower than $4 million in real term but let’s assume the remodel cost to be $4 million. So, annual depreciation is $0.4 million and EBIT is $0.6 million.

A very optimistic assumption is that the store requires no remodel. So, the $4 million upfront investment results in $1 million annual cash flow forever. That translates into 25% annual return (25% = ¼). Realistically, there’s remodel cost every 10 years. So, 25% is the ceiling of ROIC. Generally, ROIC is always lower than EBITDA/Gross NTA. (Gross NTA = Gross fixed assets + Net working capital.)

A very conservative assumption is that we set aside “DA” each year. In the TFM example, we set aside $0.4 million each year so that after 10 years we have $4 million to spend on remodeling. That way, we’ll have $0.6 million free earnings each year (the “free” part is borrowed from the term free cash flow). So, the $4 million upfront investment results in 15% annual return (15% = 0.6/4). Realistically, we don’t set aside $0.4 million each year but use that money to fund new store openings. So, 15% is the floor of ROIC. Generally, ROIC is always higher than EBIT/Gross NTA.

If we open Excel and calculate IRR for various scenarios, we can see that IRR tends to be in the upper end of the range between EBIT/Gross NTA and EBITDA/Gross NTA. The midpoint of the range is quite a good estimate of ROIC.

Using EBIT/NTA is dangerous when fixed assets are a big part of NTA. I made that mistake when I first looked at Town Sports International (CLUB). Median EBIT/NTA was 20%, which looks good. But median EBIT/Gross NTA was 9% and median EBITDA/Gross NTA was 19%. So, pre-tax ROIC is around 14% instead of 20%. That’s a mediocre return.

We must be flexible when estimating return. We have to look at composition of NTA. It’s okay to use EBIT/NTA when PPE is a tiny part of NTA because the error is small. If PPE is a big part of NTA, using the midpoint of EBIT/Gross NTA and EBITDA/Gross NTA is preferable. If receivables are a big component, we should make adjustments. For example, America’s Car-Mart (CRMT) has $324 million receivables, $34 million inventories and $34 million PPE. However, Car-Mart doesn’t really lend money. Car-Mart lends cars. So we should adjust receivables to (1-gross margin) * receivables to estimate the total value of the cars it lends and use that number to calculate NTA. A better method to estimate ROIC is to look at the economics of each loan.

**Return on Incremental Invested Capital (ROIIC)**

What we really want to know is ROIIC rather than ROIC. We can calculate ROIIC by taking incremental EBIT or EBITDA over incremental invested capital over a 1- to 3-year period. That’s not a good approach. Sometimes a company has excess capacity so growth doesn’t require fixed investment for a while. Or sometimes a company has excess working capital and it can take capital out. But these examples are short-term adjustments. In the long run, volume growth requires investment in new production/service capacity and in working capital. So, ROIC is a good starting point to estimate long-term ROIIC.

Reinvestment in the same business tend to achieve returns similar to past ROIC. That’s why many businesses have ROIC within a certain range. However, we need to make some adjustments to ROIC to have a fair expectation of ROIIC.

Margin expansion can make ROIIC higher than ROIC. Margin expansion is usually a result of volume growth that drives down unit cost. For example, when gross margin is high and SG&A is relatively fixed, volume growth will significantly increase EBIT margin. Tom Russo usually uses Brown-Forman to illustrate the concept of the capacity to suffer. Brown-Forman is willing to incur expenses today to build infrastructure for international growth tomorrow. And the next 50,000 bottles it sells will have better margin than the last 50,000 bottle.

Frost (CFR) is another good example. Frost’s branches grow deposits faster than inflation. So, operating expenses per $ of deposit declines overtime. Gross margin in the banking industry is net interest spread. Net interest spread is influenced by interest rates and demand for loans. It’s cyclical but very stable over a long period of time. So, lower operating expenses per $ of deposits improve ROA. Today, Frost makes lower ROA than it did in the past. But without the impact of low interest rates, Frost should be able to make much better ROA.

We must be careful when volume growth is outside of current goodwill. In such case, high ROIC in the past doesn’t guarantee a high return on reinvestment. See’s Candies wasn’t able to grow profitably in other states because it failed to replicate the mindshare it had in California. TFM is a current example. TFM is a gourmet food chain. Consumers shop at traditional grocers most of the time. But in some special occasions, they may go to TFM for very good foods. Consumers on average go to TFM only once a month. TFM wants to be the first choice retailer for “special.” So, unlike other grocers, TFM relies on mindshare instead of habit. TFM is very strong in the Southeast. It got into trouble in recent years when it expanded into new markets. It’s very difficult to create mindshare in a totally new state. But perhaps it’s easier to open the next store in that state because the first store helped build some awareness and word of mouth.

**Conclusions**

The term “good business” is perhaps too broad. A firm that achieved high growth and great return but have little growth potential in the future isn’t as good as its past success suggests. Firms that barely made profit in the past might now be done with the investment phase and will enjoy great profitability in the future. What investors care about is perhaps more specific: a good business to buy. I propose 3 indicators of a good business to buy. The first is negative invested capital. The second is exceptional pricing power. The third is high ROIC. Past ROIC is a good benchmark for ROIIC. But to have a fair expectation, we need to consider other factors like whether margin of additional units will be higher and whether volume growth is inside current goodwill.