![Sample our Area Studies journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5922/TOC-Subject-Sample-2021-Area-Studies.jpg"})
Abstract
Although many are under the impression that income inequality in Hong Kong has been continually on the rise, the picture presented by official statistics is mixed. This article suggests that official statistics, due to limitations in their design, have underestimated the impact of the top earners. Utilising taxation data, a continuous and strong growth in inequality can be identified when the distribution among taxpayers is considered, providing an alternative picture of inequality in Hong Kong. It is argued that the surge in income inequality is driven in part by the rich earning more rather than by more people becoming rich. The Hong Kong government, instead of adopting redistributive measures to alleviate the situation, has exacerbated inequality through policy choices that strongly favour the richest group. Although these actions might be explained in relation to the political influence wielded by the business sector and the upper class, the social implications of inequality cannot be neglected. Apart from a stronger demand for democracy, perceptions of income inequality and the pro-rich bias in the political system significantly affect people’s political views, as seen in the support for the recent Occupy Movement.
Acknowledgments
I would like to thank Lui Hon-Kwong and two anonymous reviewers for their valuable comments and suggestions on earlier drafts of this paper. Special thanks must also go to Sing Ming for generously sharing two sets of survey data.
Notes
1. I would like to thank an anonymous reviewer for suggesting this point.
2. It is reported that, following the redistributive process, the Gini in 2011 would be reduced from 0.537 to 0.521 (post-tax) and 0.475 (post tax and post-social benefits including education, housing and medical services) (Census and Statistics Department, Citation2012a, p. 107; see also Table below). Similar efforts have also been made by Lui (Citation1997; Citation2007) to estimate the redistributive effect of public housing, but he identifies only marginal effects on inequality.
3. Even though the survey collects the actual income levels, the figures are top-coded in the final report.
4. See Census and Statistics Department (Citation2012a, p. 171) for the formula and details of this estimation method.
5. From 2001 to 2011, the average number of working members per household remained constant (Census and Statistics Department, Citation2012a, p. 104). The share of “nuclear family households” changed slightly from 66.2 per cent to 66.3 per cent. The average number of children/elderly per household was also virtually unchanged (Census and Statistics Department, Citation2012c, p. 17).
6. Allowances include personal allowances, allowances for married persons, dependent siblings or parents, home loans etc.
7. A yearly income of HK$108,000 in 2001 would become HK$118,648 in 2011 in real terms. The calculation is then done by adjusting the income for inclusion to those earning HK$120,001 or more. The Gini for this population of taxpayers changes from 0.457 to 0.450.
8. Economically-active households are households with at least one person currently working or unemployed.
9. As the tax system changed slightly over the period under investigation, doubts can also be raised about the effect of such changes on the level of reported income. In particular, a supply-side interpretation suggests that a change in tax rate can influence the reported incomes of those affected. With a long time series in the US, Piketty and Saez (Citation2003) found little evidence of such lasting effects. Changes in the tax bands and standard tax rates are also discussed below.
10. Dividing taxpayers into 24 groups (according to their income level), data from the Inland Revenue Department (multiple years) report the number of taxpayers and the amount contributed for each group. The Gini based on the whole population can then be calculated based on the Lorenz curve. Supporting materials for this calculation are available upon request.
11. The EC has 1,200 members elected by a group of around 250,000 professionals, political and business elites. However, sectoral bias is also apparent in the generation of this group of electors. For example, the Employers’ Federation has 16 seats among just over 100 corporate members, while the Education sub-sector elects 30 members from close to 100,000 individuals (teachers). In reality, as many as one-third of the seats have been returned uncontested in previous elections (e.g. Young & Cullen, Citation2010, p.71).
12. For example, private members’ bills require majorities from both the functional and the geographical members separately, granting functional members the crucial veto power over popular bills that might damage business interests.
13. While it is not the case that the dutiable cases for each year necessarily contributed to the revenue for that particular year (there might be delays in assessment or payment), the number of dutiable cases and the amount of revenue received each year were very stable from one year to the other, making the analysis here valid.
14. Some other potentially redistributive measures were implemented in this Budget, such as a subsidy on electricity charges and extra old-age allowances. However, these measures were not explicitly directed towards a certain class, which would make a direct comparison more difficult.
15. The interviews were conducted in Cantonese, the main local dialect, in late November 2014. Following the usual procedure for this type of study in Hong Kong (e.g. Chan & Chan, Citation2006; Sing, Citation2010), a two-stage sampling method was adopted. First, based on telephone directories, a sample was generated by a computer system, replacing the last two digits of numbers to cover all unlisted phone numbers. Interviews were then conducted from numbers randomly chosen from the sample. Second, if the interviewed household had more than one qualified member, another random selection (the next birthday method) was used to select a respondent. According to official figures, 96.7 per cent of households in Hong Kong have access to a fixed line telephone, ensuring that our sample was reasonably representative of the entire Hong Kong population. 806 interviews were completed in total. The overall response rate was 67.7 per cent with a ± 3.5 per cent sampling error, which is fairly comparable to similar studies in Hong Kong (Chan & Chan, Citation2006; Sing, Citation2010). It is further noted that, as in all individual opinion polls, we work on the assumption that the population is homogeneous. I thank an anonymous reviewer for suggesting this point.
16. The same response variable has been used in other studies on the movement (e.g. Kurata, Citation2015, p. 26). Results with the inclusion of age, gender and education level are largely the same. They are not reported but are available from the author upon request.