This Bloomberg item on the best way to measure corporate performance, which is based on a white paper by Deloitte, seems a bit naïve to me. 

Yes, return on equity is easily gamed by leverage and stock buybacks. But return on assets is hardly immune to financial engineering. 

And you needn't go as far as Enron did to inflate your results through off-balance-sheet financing. All it takes to reduce your assets and goose your returns is receivables securitization, which is common enough, not to mention capital leasing and some artful reductions in ownership of heavily indebted subsidiaries. 

Think of what I call Coca-Cola's 49 percent solution, wherein the syrup maker doesn't include the debt of its biggest, asset-heavy bottlers on its balance sheet because it doesn't own a majority stake in them, even though one can argue Coke effectively controls its bottlers through board influence and marketing agreements. That does wonders for its ROA without really changing anything else. 

In fact, I don't think there's any single best way to measure performance, though one employed by our editorial advisor Chuck Mulford of Georgia Tech may come closest, thanks to his focus on sustainable free cash flow from operations. 

Mulford's methodology adjusts for all such financial engineering as well as one-time boosts from financing and investing activities. 

Mulford himself doesn't dismiss ROE, though. In fact, he told me in an email yesterday that he thinks both ROE and ROA are useful, and that ROE is probably a better measure of shareholder returns. And he points out that ROA can also be misleading in so far as it doesn't include assets created through research and development or through advertising, sales and marketing efforts, since those are expensed rather than capitalized under GAAP.  

In addition, Mulford points out that while software development is capitalized, companies account for it different ways. Plus, he says, many companies carry assets at old historical cost, which understates those assets' value and thus inflates those companies' ROA. 

As he put it, "ROA and ROE measure different things and carry a different point of view.  They are both important and neither should be used at the exclusion of the other."   

I'd only add that if the Financial Accounting Standards Board ever gets its act together on its financial statement presentation project, I could easily imagine Mulford's type of analysis becoming more popular. 

That project would create a matrix in place of the current income statement to show where earnings are coming from, essentially incorporating the elements of the cash flow statement and balance sheet to illustrate and distinguish among profits earned from operations, financing and investing. 

Alas, the project has been in the works for almost all of the past decade, and while it's getting closer to completion, FASB won't even make a decision on implementation dates until next year at the earliest.

For now, that leaves ROE, ROA and additional analysis like Mulford's.