Rational, adaptive and learning behavior of voters: Evidence from disaggregated popularity functions for Sweden

The disaggregated study confirms earlier aggregated studies showing that unemployment strongly affects the popularity of the Swedish government. The results presented here, however, suggest that there are differences among popularity functions estimated across groups differing in age, income and gender. Consequently, popularity functions estimated at an aggregated level ignore reaction patterns within various groups in society. Age: A smaller part of the variation in popularity was explained for the youngest and the oldest age groups than for the middle age groups. Older voters appear to be less influenced by economic factors than other groups. They are economically less influenced by changes in unemployment and probably have party loyalties and political beliefs that remain less affected by economic events. The young voters are influenced as much as those 25–39 and 40–49 years of age by unemployment. Since the young show a much greater variability in their political allegiances, the unemployment variable explains much less of the variation. Young voters may be more strongly affected by issues other than economic ones, e.g., the nuclear power issue, Vietnam, etc. Income: A comparison of different income groups shows that unemployment influences the popularity of the government much more among those of middle income than those of high or low income. A greater part of the variation in popularity is also explained for this group than for the other groups. Gender: We find that men and women respond in roughly a similar manner to unemployment and inflation. The use of group specific measures of unemployment and of the food price inflation did not bring out any significant differences between these measures and (economy-wide) general measures of unemployment and inflation. Thus voters across age, income and gender appear to exploit the same informational sources when judging the economic performance of the government. We raised the question initially of whether people learn with age and thus becoming more rational and knowledgeable. Our results are consistent with this view. However, the patterns obtained are also consistent with the view that voters' preferences are influenced by their economic status, since income moves upwards with growing age. Copyright Martinus Nijhoff Publishers 1987