There are two measures of profitability common in the financial community, return on assets (ROA) and return on equity (ROE).
ROA = net income / total average assets
ROE = net income / total stockholders equity
Assets and equity, as used in these two common indexes, are both measured in terms of book value. Thus, if assets were acquired some time ago at a low price, the current performance of the organization may be overstated by the use of historically valued denominators. As a result, the accounting returns for any investment generally do not correlate well with the true economic internal rate of return for that investment.
Difficulties with using either ROA and ROE as a performance measure can be seen in merger transactions. Suppose we have an organization that has been earning a net income of $500 on assets with a book value of $1000, for a hefty ROA of 50 percent. That organization is now acquired by a second firm, which then moves the new assets onto its books at the acquisition price, assuming the acquisition is treated using the purchase method of accounting. Of course, the acquisition price will be considerably above the $1,000 book value of assets, for the potential acquirer will have to pay handsomely for the privilege of earning $500 on a regular basis. Suppose the acquirer pays $2,000 for the assets. After the acquisition, it will appear that the returns of the acquired firm have fallen. The firm continues to earn $500, but the asset base is now $2,000, so the ROA is reduced to 25 percent. Indeed, the ROA may be less as a result of other factors, such as increased depreciation of the newly acquired assets. Yet in fact nothing has happened to the earnings of the firm. All that has changed is its accounting, not its performance.
Another fundamental problem with ROA and ROE measures comes from the tendency of analysts to focus on performance in single years, years that may be idiosyncratic. At a minimum, one should examine these ratios averaging over a number of years to isolate idiosyncratic returns and try to find patterns in the data.