Of course, self-insurance really means no insurance, but the latter sounds so raw and so exposed. The term self-insurance at least connotes the idea that the business enterprise recognizes the risks it faces and has prepared for their possibilities. Unfortunately, this spin might hide some creative accounting.

Accounting for self-insurance is rather straightforward. When accruing this amount, one charges insurance expense and reports a liability. Later, when some loss occurs, the bookkeeper debits the liability and credits cash or whatever other assets have lost value. The chief consequence of this accounting is to move the expense from one period to another. In other words, it promotes the management of earnings.

McDermott International conducts engineering, construction, and project management services in the energy sector. This company routinely reports on its balance sheet a self-insurance liability. For December 31, 2002, it discloses a self-insurance reserve of almost $72 million, while on December 31, 2003, it shows self-insurance liability of roughly $61 million. Unfortunately, it doesn't disclose the expense or the cash outflows related to this account.

One issue is whether the business enterprise correctly shows its debts. How do we know it has a liability at the end of 2003 of $61 million? Why not more? Why not less? Generally speaking, public auditors do not care about the overstatement because the risk to them and to shareholders generally is less than the problem of understating corporate debts. That provides small comfort, since we are still left wondering about the existence of the liability and the timing of expenses on the income statement.

But how do we measure self-insurance? Statement 5 helps a little, for it says that an entity must measure and report a liability when it becomes probable and can be reasonably estimated. This requires one to know with some confidence the probable cash outlays. This should imply that McDermott can reliably estimate its outlays related to this "self-insurance."

Paragraph 14 of Statement 5 goes on to say that "general or unspecified business risks do not meet the condition for accrual…." One reason is that balance sheet accounts do not deal with what might be or what could take place. Another reason is that it becomes impossible to estimate with any confidence the cash outflows for the what-may-be's of life.

Assuming that McDermott is not accruing losses for general or unspecified business risks, then perhaps it merely employs an unsuitable term for them. In scanning the reports of McDermott, it appears that these risks refer to lawsuits against the company and to workman compensation issues. If so, it would be more informative if McDermott termed this "liability litigation reserves" or "workman compensation liabilities."

Besides the question about the existence of a liability and its measurement, we also must entertain the question about the correlated expense. Does the expense exist and, more importantly, in what period should it be reported? For example, how does an investor know that the $11 million expense from self-insurance (the difference between the $72 million and the $61 million in self-insurance liabilities) belongs in a past period and not in the 2003 income statement? Might the 2003 earnings be overstated?

In short, the self-insurance disclosures by McDermott raise more questions than answers. If it is truly self-insurance, then likely the practice should be halted because it violates paragraph 14 of Statement 5. On the other hand, if it links to some specific risks that the corporation faces, then McDermott should rename the item and provide better insight into what is going on, including some details about the cash flows.

That self-insurance still exists at all borders on the incomprehensible. Auditors should completely eliminate its practice, in compliance with Statement 5, and they should require more informative labels to describe what business firms are actually doing and the risks that they face. While a small step, it would help reduce the gaming techniques that managers possess to mislead the investment community.