The discipline of political risk analysis has been criticised as being a 'soft science'. Therefore, a major challenge is to provide an empirical analysis of political risk and to prove that political risk can indeed be measured. This article provides an empirical analysis of political risk by testing the reliability of current risk assessment approaches to accurately forecast political risk. Although only a few attempts have been made to test the reliability of specific political risk assessment models, that of Howell and Chaddick 1994 in respect of the Economist Intelligence Unit, Political Risk Services and Business Environment Risk Intelligence political risk assessment stands out. This article revisits the Howell and Chaddick study.
In order for forecasts to be reliable, they must be justified and defended by applying practical logic, implying that theory be tested against real world experience. Hence, a reliable analysis requires that actual losses be tested against theory. Therefore, in addressing the connection between theory and actual losses, this article correlates losses incurred in the period 1994-2004 with theory. The findings suggest that political risk can be empirically tested and measured and that the analysis of political risk is essential for political risk management.
Few empirical attempts have been made to test the reliability of political risk models used to assess political risk faced by enterprises. This article attempts to test three political risk forecast models against actual losses incurred by firms, thereby addressing the question of the reliability of the models concerned. The article draws on the work of Howell and Chaddick which showed conclusively that there is a variation between the Economist Intelligence Unit (EUI), Political Risk Services (PRS) and Business Environment Risk Intelligence (BERI) political risk models used to forecast political risk. (1) In this article, the assessment of Howell and Chaddick is repeated by testing the predictive reliability of the same models ten years on.
2. THE 1994 HOWELL AND CHADDICK ASSESSMENT
Political risk analysis has faced a lack of credibility since its inception. (2) The most prominent reason for this, amongst others, is the perception that the field of political risk analysis is a 'soft science'. (3) Although many political risk assessment models have an ostensible quantitative base, few efforts have been made to demonstrate the reliability of political risk forecasts. A noted example is Howell and Chaddick's study, which addressed the question of reliability of political risk projections. Their analysis correlates five year forecasts of three political risk assessment models, namely EIU, PRS and BERI, with actual losses incurred by insured companies over the 1982-1994 period. (4)
In testing the reliability of political risk forecasting, the task is to theoretically link the political acts resulting in loss to the causes of the loss, by establishing an index of losses due to political risk. In order to do this, a list of loss actions must be constructed, which includes the acts or events that are political in nature that result in the respective subsequent losses. Accordingly, Howell and Chaddick explain that "the modeler would try to envision the circumstances under which these events will occur". (5) In projecting the circumstances under which these events or actions occur, the modeler constructs a list of variables. In other words, the modeler identifies political risks that would be the cause of loss. (6)
In terms of the aforesaid, "a measure of loss due to political causes was needed to test the model and theory, in this case employing multiple correlation and regression". (7) Hence Howell and Chaddick started the assessment of three approaches by creating a 'measure of loss' index. (8) The loss index is based on actual losses incurred in the 1982-1994 period. …