To « self-insure » an organisation may simply create a fund internally from which transfers can be made to general reserves in the event of a loss. This is not necessarily  a good use of capital and may incur an important opportunity cost as the capital may be better used in developing the business itself. It does, however, avoid insurance premium taxes and other government levies applicable (assuming the circumstances in which losses are funded are appropriately structured).
A more formal way to self-insure is to create what is in effect a subsidiary insurance company, or “captive insurer” whose principal business is transacting insurance on the parent’s internal risk exposures. This can be positioned tax-effectively and bear all or part of the cost of risk exposures where these become claims. A captive can gain access into the wholesale (reinsurance) market, to protect itself and the parent against high-value losses. Through this medium the organisation can again smooth loss experience over time.
The captive must be capitalised adequately, that is, it must hold sufficient funds to meet all foreseeable losses. The decision to proceed with the formation of a captive will need to embrace the cost of that capital as well as the costs of setting up, resourcing, managing and maintaining the company. The captive represents a long-term investment for the parent, not least because it may be some years before the benefits return in profitability terms.
The captive must meet the standards of international financial regulators and local financial and corporate governance regulations, including the important changes in financial reporting recently introduced, when consolidated into the parent’s results .

Source : CII Knowledge Base — Extracts from: Alternative Risks Transfer  – www.cii.co.uk/knowledge