How to Spot Underinvestment Hiding Behind a High ROIC

Picture a screen sorted by return on invested capital and filtered for names whose ROIC has climbed for five straight years. Near the top sits a business with flat sales, no acquisitions, no new product line, and a return on capital that has gone from the high twenties to the high thirties. The chart slopes up and to the right. Nothing in the screen's columns tells you why.

Two very different businesses can draw that chart. One earns more because it is getting better — pricing power, scale, a widening moat. The other earns about the same while the capital underneath it shrinks, because it spends less on its plants and stores each year than they wear out. ROIC is a fraction. You can raise it by growing the top or by starving the bottom, and the ratio won't tell you which happened.

I want to walk through the method I use to tell them apart, then run it on two real sets of filings: a partnership that has been running down a chemical asset base and a restaurant chain that has been pouring cash into new locations. Both reported a higher ROIC in fiscal 2024 than in fiscal 2019. Only one of those increases says much about quality.

Where a Rising ROIC Actually Comes From

Start with the arithmetic, because that's where the trick hides. The version I'll use throughout is operating income over the capital the business needs to run:

Pre-tax ROIC = operating income ÷ invested capital, where invested capital = net PP&E + goodwill + intangibles + (current assets − cash − current liabilities).

I use pre-tax returns here on purpose. One example is a partnership and the other a corporation, and taxing them differently would blur the comparison. Our primer on ROIC covers the after-tax version; the logic below carries over unchanged.

The denominator is the part people skip. Net PP&E falls every year by the depreciation charge and rises by whatever the company spends on capex. If capex runs below depreciation year after year, the asset base shrinks by itself — no announcement, no decision anyone has to defend on a call. Hold operating income flat and ROIC climbs anyway. That isn't a moat. It's the balance sheet recording a business that is consuming its own plant.

And for a while it looks excellent. Free cash flow is high because capex is low. Margins hold because the depreciation charge is still based on assets bought years ago at older prices. The bill arrives later: as deferred maintenance, as lost share to a competitor that kept investing, or as a catch-up capex cycle that resets the denominator in a year or two.

This is a different case from the one in why high-ROIC businesses choose buybacks over reinvestment. There, a company keeps its existing assets fully maintained and hands back the surplus because it has run out of high-return places for new money. Harvesting means the existing assets themselves are left to run down. A company can do both at once, which is exactly why the check matters.

A Method in Three Ratios

None of this needs anything beyond the cash flow statement and the balance sheet in the annual report. I run three ratios over a five- or six-year window, in this order:

1. Capex to depreciation, summed over the window. Single years are noisy — plants have turnaround years, retailers have opening waves — so add up all the capex, add up all the depreciation and amortization, and divide. A ratio persistently below 1.0 means the asset base is shrinking in book terms. Below roughly 0.6 for a capital-intensive business is where I start treating it as a flag. A flag, not a verdict.
2. Invested capital trend against revenue trend. If revenue is flat or rising while invested capital falls, the business is producing more sales from less capital. That's either real efficiency or wear and tear, and the third ratio helps decide which.
3. Return on incremental capital: the change in operating income divided by the change in invested capital over the same window. When both rise, you get a number you can judge. When income rises and capital falls, the ratio breaks — and the broken ratio is itself the signal that the improvement came from the denominator.

Then one cruder test: recompute the latest year's ROIC using the starting year's operating income. If most of the improvement survives that swap, the shrinking base did most of the work. The return on incremental capital post goes deeper on that third ratio; here it serves as a diagnostic rather than the destination.

One practical warning before the numbers. The depreciation-and-amortization line often includes amortization of acquired intangibles, which no capex will ever replace. For an acquisitive company that drags the capex ratio down artificially. Both examples below carry almost no intangibles, so their line is close to pure depreciation.

Westlake Chemical Partners vs. Texas Roadhouse, Fiscal 2019–2024

Westlake Chemical Partners is a publicly traded partnership built around ethylene production assets, with a fiscal year that ends in December. Westlake Chemical Partners' 10-K for fiscal 2019 reported revenue of $1.09 billion and operating income of $350.2 million. Its 10-K for fiscal 2024 reported revenue of $1.14 billion and operating income of $390.4 million. Revenue up about 4% over five years; operating income up about 12%.

Now the capital side. Across fiscal 2019 through 2024, capex summed to $311.8 million while depreciation and amortization summed to $662.5 million — a ratio of 0.47. In no single year did capex reach 75% of the depreciation charge. Net PP&E fell from $1.10 billion at the end of 2019 to $903.6 million at the end of 2024, and invested capital on the formula above fell from $1.29 billion to $1.04 billion, down about 20%.

Put it together and pre-tax ROIC went from 27.2% to 37.7%. Impressive, until you run the swap test: fiscal 2019's operating income over fiscal 2024's invested capital gives 33.8%. Roughly six and a half of the ten and a half points of improvement came from the shrinking denominator alone. Return on incremental capital can't be computed in any useful sense, because operating income rose by $40.3 million while invested capital fell by $251.8 million.

To be fair to the partnership, this may be exactly what its owners signed up for. A vehicle designed to distribute cash isn't trying to look like the businesses in our anatomy of a long-term compounder. And a plant can stay well maintained at below-depreciation capex if its book depreciation schedule is more conservative than its real working life. I can't tell from the filings alone which of those is true here, and I'm less sure about the depreciation question than I'd like to be. What I can say is that the 37.7% is not evidence of a widening moat. It's evidence of a shrinking base.

Texas Roadhouse makes the useful contrast. Its fiscal year ends in late December. Texas Roadhouse's 10-K for fiscal 2019 reported revenue of $2.76 billion and operating income of $212.0 million; its 10-K for fiscal 2024 reported revenue of $5.37 billion and operating income of $516.5 million.

The capital side runs the other way. Capex across fiscal 2019 through 2024 summed to $1.52 billion against depreciation and amortization of $828.8 million, a ratio of 1.83. Net PP&E rose from $1.06 billion to $1.62 billion. Invested capital rose from $905 million to $1.23 billion — up about 36% — while revenue almost doubled.

Pre-tax ROIC went from 23.4% to 41.9%, a bigger jump than Westlake's, and it happened while the denominator grew. Return on incremental capital: $304.5 million of extra operating income on $326.5 million of extra invested capital, or about 93% pre-tax.

That 93% flatters the company a little, and it's worth being specific about how. Restaurants collect cash before they pay many of their bills, so working capital runs negative, and it got more negative as the chain grew — from minus $277 million to minus $557 million on my formula. Strip working capital out and count only PP&E, goodwill and intangibles: the extra capital becomes $606.1 million and return on incremental capital comes to about 50%. Still exceptional. A business that adds $304.5 million of operating income while spending 1.83 times its depreciation charge is the opposite of harvesting — it's a company still working down a long reinvestment runway.

• Capex ÷ D&A, fiscal 2019–2024: Westlake Chemical Partners 0.47; Texas Roadhouse 1.83.
• Invested capital change: Westlake Chemical Partners down about 20%; Texas Roadhouse up about 36%.
• Revenue change: Westlake Chemical Partners up about 4%; Texas Roadhouse up about 95%.
• Pre-tax ROIC, fiscal 2019 to 2024: Westlake Chemical Partners 27.2% to 37.7%; Texas Roadhouse 23.4% to 41.9%.
• Return on incremental capital: not meaningful for Westlake Chemical Partners (income up, capital down); about 50% to 93% for Texas Roadhouse, depending on how working capital is treated.

On a screen sorted by ROIC trend, both names look like winners. A rising trajectory earns credit in the returns pillar of MoatScope's Quality Score, and both trajectories rose. The direction is the same. What produced it isn't.

When a Low Capex Ratio Is Innocent

Not every business with capex below depreciation is starving itself. A few situations produce the same pattern for good reasons:

• A deliberate shift to an asset-light model — contract manufacturing, franchising, a sale-leaseback — moves assets off the PP&E line on purpose. The test is whether revenue and margins hold up afterward.
• The tail end of a big build. After a capacity wave, a company can run below depreciation for a few years while the new assets are young. The window has to be long enough to include the build, or the ratio misleads.
• Acquired-intangible amortization, as above. Back it out before computing the ratio, or use depreciation alone if the filing breaks it out.

Capex changes also get misread in real time. On its third-quarter earnings call on October 22, 2026, Tesla lowered its 2026 capital spending plan to about $20 billion from more than $25 billion, as Electrek reported that day. Some coverage read it as discipline, some as retreat. The checks above say neither yet: a cut in guidance only matters for this test once spending falls below what the existing assets consume, and one year's guidance can't show that. Run the multi-year sum before deciding.

My working rule is simple. A low capex ratio is innocent when you can point to the specific reason in the filing — a disclosed sale-leaseback, a plant closure, a move to outsourced production — and revenue holds up afterward. It's a warning when the ratio just drifts lower with no stated reason while management talks up its returns on capital.

Key Takeaways

• ROIC is a fraction, and a shrinking denominator raises it as surely as a growing numerator. Capex persistently below depreciation shrinks the denominator without anyone deciding to.
• Three ratios over five or six years separate harvesting from reinvesting: capex to depreciation, invested capital against revenue, and return on incremental capital. A broken incremental ratio — income up, capital down — is a signal in itself.
• Westlake Chemical Partners' pre-tax ROIC rose from 27.2% to 37.7% between fiscal 2019 and 2024, with about six and a half points of that coming from a smaller capital base. Texas Roadhouse's rose from 23.4% to 41.9% while its capital base grew 36%.
• A low capex ratio can be innocent, but only when the filing names the reason. If it doesn't, treat the high ROIC as a harvest until the numbers show otherwise.