Unemployment In Indonesia : Analysis Of Census Data

It is generally recognised that the working poor in less-developed countries are more numerous and represent a more serious problem than the totally unemployed. Apart irom these difficulties with the concept, there is great conlusion about the ievei of unemployment in indonesia, because the Indonesian census has published two sets of figures, a set of preliminaiy figures in its Series C report and a set it tiiial figures Dy provinces (so tar completely available only for Java) in its Series E reports. The Series C showed the piopoition of the economically active population who were unemployed in the whole country as 2.2 per cent (a proportion lower than the 5.4 per cent reported in the 1961 Census) and as 2.3 per cent for Java, while the series E showed the unemployment rate for Java to be 8.3 per cent. Both series defined the unemployed as those who were not working or were working less than two days during the reference week preceding the date of enumeration, but excluded farmers not working tor seasonal reasons. Series C estimates were based on a 10 per cent sample tabulation of the Sample Census covering about 3 per cent of all households, while Series B estimates were based on a complete tabulation of the Sample Census The differences between the two estimates is, however, mainly due to differences in imputation procedures by which returns with incomplete or inconsistent answers were assigned to various categories. The definition of unemployment based on a week as the reference period generally gives lower estimates than one based on a longer reference period or one based on some other approach, such as that of usual activity. But even on the basis of a one week reference period, the Series C estimates are too low. At the same time, it appears that the changes in imputation procedures which produced the Series E estimates have result in an excessive adjustment of the overall Unemployment rate. There is some danger that these official estimates may be neglected in future work because of the wide disparity in them. This note, therefore, attemps to analyse the differences between the two estimates in order to suggest a more realistic estimate that might be useful for purposes of further analysis and policy-making