Various methods used in different countries to estimate the overall magnitude of tax evasion have produced widely diverging results, both within and between countries. The idea of the present study is that random audits by experienced auditors will produc e a rather well-founded basis for estimating tax evasion in various industries. A well designed control strategy will also produce valuable information about the characteristics of firms that evade taxes and those which do not.
Because of limited auditi ng resources it is important not to "waste" auditors' time on firms that pay their taxes loyally. Therefore, the knowledge of auditors will be used to distinguish between presumably loyal and presumably not loyal firms. Tests based on random audits will s how if such a distinction is acceptable.
A two step audit will be carried out on a random sample of the firms that are assumed not to be loyal. A formal audit will be carried out in order to determine if a more thorough audit is warranted.
This proce dure requires a careful and detailed design of the audit strategy. In order to be able to use the findings in statistical estimates and tests, all audits will be carried out and registered in the same manner. The development and evaluation of audit strate gies, together with suitable principles of registration, will form a basis for estimating tax evasion and evaluationg control strategies both at present and in the future.