1 Introduction

Single parenthood is not a new social phenomenon. In the past, it may have arisen because low life expectancy led to the death of a spouse. However, single parenthood nowadays is increasingly the result of divorce and premarital birth. This change can be partly attributed to cultural factors such as the growing importance of individualism and the changing views on families shared by many modern industrialized societies (Ambert 2006). Although Hong Kong is a Chinese society where traditional family values remain important, it has also been influenced by Western liberal values and the social changes associated with urbanization (Lau 1981; Lee 1991; Baker 1993). Over the past two decades, the number of single-parent families in Hong Kong has increased by 137 %, from 34,538 in 1991 to 81,705 in 2011 (Census and Statistics Department 2001, 2011). This is mainly the result of an increase in the number of divorced parents, which surged by 244 %, from 18,700 in 1991 to 64,324 in 2011 (Census and Statistics Department 2001, 2011). In 2011, 9.69 % of children aged under 18 were living in single-parent households (Census and Statistics Department 2011). Single-parent families exert strong demands for government welfare support; according to the Social Welfare Department, they accounted for the second largest group (after the elderly) making Comprehensive Social Security Assistance (CSSA) applications from 2007 to 2011 (Table 1).

Table 1 CSSA distribution by case nature (SWD Review, various years)
Full size table

The risks to children living in these families are a specific concern within the government’s anti-poverty initiative. Following the reestablishment of the Commission of Poverty in 2012, the government is now committed to supporting the underprivileged with special attention being given to disadvantaged groups including single-parent families (Commission on Poverty 2013). In Hong Kong, an affluent society (GDP per capita was USD 36,796 in 2012; Source: World Bank) with adequate social security protection, the chance of children living below subsistence level is very small. However, socioeconomic disparity in Hong Kong is serious, with the Gini coefficient reaching 0.537 in 2011; and underprivileged groups often suffer more because of their relative position in society rather than because of material deprivation in absolute terms. As such, it is more appropriate to assess the relative poverty of children in single-parent families. Despite the importance of the problem, however, we still have little knowledge of the state of child poverty among single-parent families. The current study, therefore, set out to investigate this in terms of both trends and causes.

2 Consequences for Children of Living in Single-Parent Families

It is well established that children in single-parent families have a higher chance of growing up poor (Yoo and Lim 2009; Christopher et al. 2002; Chzhen and Bradshaw 2012; Rainwater and Smeeding 2003; Robson and Berthoud 2003; Bradshaw and Chzhen 2009; Atkinson and Marlier 2010). The obvious explanation for this is that in single-parent families there is only one breadwinner whereas married-couple families often have two. Another important reason is the reduction in household earnings as a result of the barriers to labor market participation encountered by single parents. Unlike married-couple families, in which both parents can share the responsibilities of breadwinning and child rearing, single parents, especially those with young children, have to struggle to balance childcare and employment (Citizen Advice Bureau 2008; Millar and Ridge 2009; Marsh 2001). Although there are variations across societies and organizations in terms of policies that accommodate the needs of single parents, including childcare service provision and flexible working hours and locations, single parents in general suffer from lower employment participation and reduced earnings (Bradshaw et al. 1996). Furthermore, they are more sensitive to unemployment. With only one breadwinner in a single-parent household, loss of employment means the total loss of the family’s income. Lower incomes and higher unemployment rates also mean single-parent families are less capable of accumulating the financial resources needed to sustain expenditure during periods of unemployment (Ambert 2006). The immediate result is that children living in single-parent families are more likely to suffer from poverty and material deprivation, which not only have detrimental effects on physical health, psychological development, and school performance (Brooks-Gunn and Duncan 1997; Ridge 2011; Van Hulst et al. 2011), but also have negative consequences for the long-term economic outlook in adulthood (Duncan et al. 2012; Holzer et al. 2007).

However, the problems faced by single-parent families are not limited to a higher risk of child poverty. Children in these families are also more likely to become single parents themselves (Woodward et al. 2001). Poverty and lack of social support is strongly linked to juvenile pregnancy, as teenagers growing up in impoverished families and neighborhoods often lack good parenting and social support (Ambert 2006; Fischer 1993). In addition, teenagers growing up in poverty with low educational and career expectations often consider that there is little to lose by having unprotected sex and undergoing early pregnancy (Furstenberg 1992). Motherhood may be a source of psychological satisfaction that is difficult for these teenagers to achieve through other means (Cohler and Musick 1996). In this sense, single parenthood is likely to pass down to the next generation and result in intergenerational poverty.

3 Trends in Child Poverty in Single-Parent Families

Since child poverty in single-parent families is likely to reproduce itself, it is imperative to monitor the trends in the poverty risk of children in this category. Nevertheless, few studies have focused solely on poverty among single-parent families and examined the trends in child poverty in these families (Yoo and Lim 2009; Ozawa and Kim 1999). According to Yoo and Lim (2009), who assessed the poverty level of single parents in the United States, the incomes of both single-parent and married-couple families improved between 1991 and 2005. The poverty trends for both types are similar whether or not welfare income transfer is included. However, the improvement in the economic status of children in single-parent families lags far behind that of children in married-couple families at all income stages. These results suggest a widening gap between the poverty rates of children in single-parent and married-couple families in the United States. In Hong Kong, there has so far been no systematic investigation of the trend in child poverty in single-parent families. We therefore have little knowledge of the changes in their economic outlook and how they fare compared with married-couple families. To close this research gap, the primary objective of this study is to examine trends in child poverty in single-parent families and compare them with those of married-couple families. We are also interested in identifying the factors that influence and explain change over time. These factors fall into two broad categories. The first emphasizes the social and demographic characteristics of individual households. The second focuses on the broader socioeconomic context of the society under study.

4 Household Characteristics

Although the government plays an important role in redistributing national resources and providing a safety net for vulnerable children (Hirsch 2013; Musgrave 2009), the family remains the primary social institution responsible for nurturing the growth and development of children (Parke and Buriel 2006). As discussed above, the structure of the single-parent household itself is an important factor that contributes to child poverty. There are also other family characteristics that mediate the link between single parenthood and poverty. Families headed by a single mother have higher poverty rates than those where the father is alone (Christopher et al. 2002; Chzhen and Bradshaw 2012; Sorensen 1994). This could be explained by gender discrimination in employment that puts single mothers in a still more disadvantaged position with regard to employment opportunities and income (Darity and Mason 1998; Cheung 2002; Davies and Joshi 1998). In addition, parental age and the route to single parenthood are also indicators of the material wellbeing of the single-parent family. Younger single parents tend to have accumulated fewer assets in terms of both human and social capital, such as education, employment experience, and social networks, and also tend to lack tangible assets, such as savings and property. On the other hand, never-married mothers, who are often younger, also suffer greater financial risk than married women who become single parents following death or divorce, as the latter group may benefit from inheritance or maintenance provided by the former spouse (Ambert 2006). Family size is also indicative of the level of financial support and burden of single-parent families. Single-parent families with more dependent children are at a greater risk of child poverty as the intra-household distribution is more spread out.

As well as mediating factors, there are also selection factors that contribute to both single parenthood and child poverty. As discussed earlier, low educational attainment and career expectations may increase the chance of single parenthood among teenagers. At the same time, low educational attainment also contributes to the low employment status and income level of the parent (Julian and Kominski 2011). Similarly, parental immigration status is also related to both single parenthood and poverty, since divorce rates for cross-border marriages are usually high (Yang and Lu 2010) and immigrants are often disadvantaged in terms of employment. This factor contributes to the child poverty risk in single-parent families in Hong Kong (Chou et al. 2013) and elsewhere (Lichter et al. 2005; Oropesa and Landale 2000).

5 Socioeconomic Context

As well as changes in the demographic and social characteristics of families, broader socioeconomic development also changes the poverty risk associated with these family characteristics. Economic globalization and the international division of labor have polarized the economic conditions of low-skilled workers and educated professionals in most developed economies, and Hong Kong as an open economy and global city is no exception. Over the past few decades, driven by the comparative advantage of low labor costs, most of the manufacturing jobs in Hong Kong have been relocated to Guangdong province in Mainland China. The process of deindustrialization and the growing importance of the service and knowledge economy in Hong Kong demand a higher level of education and professional training from its workforce. At the same time, wages in low-skilled and elementary jobs in general have suffered from downward pressure because of surplus labor from the former manufacturing sectors and the continual supply of immigrants who tend to compete with local residents for low-end jobs (Lee and Wong 2004; Chiu and Lui 2004; Lee et al. 2007; Lam and Liu 2011). The situation of working single parents might have improved with the introduction of the minimum wage in May 2011, but they remain vulnerable to economic restructuring.

Considering the prevalence of single parents in Hong Kong and the implications of this for intergenerational poverty, we have very little knowledge about the characteristics of such families and their role in child poverty trends. Studies of single parents in Hong Kong have focused on the effect of social welfare (Census and Statistics Department 2011), life satisfaction (Lau 2003; Leung 1998), the psychological wellbeing of single parents (Lee et al. 1999), family involvement in education (Choy and Moneta 2002; Rudowicz 2001), and the educational aspirations of their children (Ho 2006). As discussed above, studies have been conducted in North American and European contexts to investigate the effects of family characteristics on the economic conditions of single-parent families. Although it is possible to generalize the effects of these characteristics on child poverty across different societies, there are important variations as a result of specific local contextual factors (Wang and Ngai 2011). To fill this research gap, the second objective of this paper is to analyze the factors that influence trends in child poverty in single-parent families in Hong Kong. We use data from the Population Census 1981–2011 to examine the trends and contributing factors using the decomposition method.

6 Methodology

6.1 Data

In this study, we used the 5 % samples of Population Census data for 1991, 2001, and 2011 as well as the 1 % samples for the 1981 Census because a 5 % sample for that year is not available for dissemination (Census and Statistics Department 1982, 1992, 2002, 2012). According to the Census and Statistics Department, the sampling unit for selecting the 5 % sample from the Population Census data set is a unit of occupied quarters for the land population and occupied vessels for the marine population. Though the original sample of the Population Census is not strictly Equal Probability of Selection of Every Sampling Units (EPSEM), all quarters and vessels carry equal weight in the sample data set. Within the 5 % sample, we focused on children aged 17 and below living with a parent, so those living with nonparents such as grandparents were excluded. In addition, because we sought to compare Chinese children living in local and immigrant families, we also excluded children who were not Chinese or whose parents had not been born in Mainland China or Hong Kong. As a result, we excluded 879, 8,372, 6,028, and 5,385 children respectively from the 1981, 1991, 2001, and 2011 Censuses, so the respective resulting sample sizes were 12,643, 57,016, 58,350, and 47,190.

The dependent variable in this study was whether the child was living in poverty. Poverty status was defined as living in a household whose income was beneath the poverty line. In this study, the poverty line was considered to be half of the median household income in the corresponding Census period. Household income refers to gross (pre-tax, pre-transfer) incomeFootnote 1; the median household income data were available in the respective reports for each Census (Census and Statistics Department 1982, 1992, 2002, 2012). The median household incomes for five household sizes (2, 3, 4, 5, and 6+) were used to calculate the poverty lines of the corresponding household size. We adopted this relative definition because it is the official definition of poverty used by the Hong Kong Government (Wenweipo 2013) and it is also widely used in the child poverty literature (Chou et al. 2013). As well as the poverty level at half of the median household income, we have also examined the poverty level at 40 % and 60 % of the median household income, as well as the subsistence rate of CSSA, to assess the situation of single-parent families at different degrees of poverty. Since we are focusing our analysis on the official poverty line in Hong Kong (i.e., 50 % of the median household income), the results for the other degrees of poverty are listed in the Appendix.

With regard to independent variables, we have measured the selection factors including parental education level (less than high school, high school, subdegree, university or above) and parental immigration status (both parents born in Hong Kong, at least one parent emigrated from Mainland China, at least one parent immigrated to Hong Kong less than 7 years earlier). The mediating factors used in this study included parental employment status (both parents not working vs. at least one parent working), gender of single parent (male vs. female), age of parent (younger than 30, 30–39, 40 and older), and the route to single parenthood (never married, widowed, divorced/separated). Other mediating factors, such as the number of children (1, 2, 3, 4, or more) living in the household, which indicates household size, and the age composition of children measured by the proportion of children living in the household matched with the following age groups—4 or younger, 5–11, and 12–17—were also measured. This information was used to illustrate the life cycle stage of families and their financial burden, which may affect the economic wellbeing of children.

6.2 Data Analysis

Our data analysis consisted of three stages. Firstly, we computed the frequency distribution of the descriptive statistics of all the dependent and independent variables used in the logistic regression. Secondly, we constructed four logistic regression models to evaluate the associations of the dependent variable, namely poverty status, with all the independent variables (parent, children, and household characteristics) for each Census year. Lastly, the Blinder-Oaxaca decomposition method (Blinder 1973; Oaxaca 1973) was used for our compositional analysis. One drawback of this decomposition method is that it is used to quantify the different contributions of group deviations to a continuous outcome. However, our outcome measure, namely poverty status, is a binary outcome. To overcome this problem, the modified Blinder-Oaxaca decomposition technique (Fairlie 2005) was used, so that the coefficient estimates from the logistic regressions and the sample mean of the independent variables entered in the regression could be utilized in the decomposition by using STATA Version 12.

Different estimates of the probability of being poor by year were generated by these logistic regression analyses and these estimations were utilized in the decomposition analysis. Because the means of predicted likelihood were actually equal to the means of the dependent variable, we were able to decompose the change in the mean of the response into a composition component by keeping the coefficients unchanged. At the same time, a rate component could be estimated by holding the independent variables constant while varying the coefficients. Consequently, the composition component is the change in the likelihood of being poor, which was accounted for by the independent variables (family characteristics) we used, and the rate component is the socioeconomic contexts associated with the particular family characteristics we used in the decomposition analysis. As such, the composition and rate components reflect the relative contribution of the family characteristics we measured and the socioeconomic context to the poverty risk of children.

7 Results

7.1 Trends in Child Poverty

Table 2 shows the overall trend in child poverty in Hong Kong between 1981 and 2011 and the corresponding figures for single-parent and married-couple families. Over the past 30 years, the overall child poverty rate has risen steadily, from 17.6 in 1981 to 26.2 in 2001, before leveling off in the past decade. Child poverty rates among married-couple families have followed a similar trend, rising from 17.2 to 24.7 between 1981 and 2001, and then declining slightly to 23.6 in 2011. While the child poverty rate in Hong Kong has remained constant in the past 10 years, the reduction among married-couple families has been compensated for by a continued increase in single-parent families, where the rate has climbed steadily from 24.8 in 1981 to 49.6 in 2011.Footnote 2 During the period under study, the child poverty rate for single-mother families was significantly higher than for both single-father and married-couple families.

Table 2 Poverty rates among children of single parents and children of married couples, 1981–2011 (Poverty = 1 if income <50 %)
Full size table

7.2 Changes in Social and Demographic Characteristics

Table 3 shows the changes in social and demographic characteristics among single-parent and married-couple families between 1981 and 2011. When assessing the mediating factors, it can be seen that the employment contribution of married couples was significantly higher than that of single parents throughout the period under study. While labor participation has remained relatively constant over the past three decades for married couples, it decreased continuously for single parents between 1981 and 2001, and has only slightly recovered in the past 10 years. During the period under study, the percentage of single fathers has continued to decrease while the percentage of single mothers has risen substantially, from 47.7 in 1981 to 78.8 in 2011. In terms of family size, the number of children in single-parent households was evenly spread from one to four or more in 1981, but this then changed dramatically with 93.1 % of single-parent households in 2011 having only one or two children. A similar trend was observed in married-couple families, but the fertility rate of married couples dropped more sharply, with couples bearing only one child increasing from 14.2 in 1981 to 43.8 in 2011, and couples bearing 4 or more children dropping from 30.5 in 1981 to 1.0 in 2011. Single-parent families also tended to have older children. There was also a general increase in the average age of parents in both family types. In particular, the percentage of single parents below the age of 30 declined significantly, from 35.5 % in 1981 to 4.0 % in 2011.

Table 3 Social and demographic characteristics of children of single parents and children of married couples, 1981–2011
Full size table

With regard to selection factors, there are an increasing number of single parents who are university educated. However, more than three times as many graduates lived in married-couple families (25.2) than single-parent families (6.7) in 2011, whereas twice as many single parents (51.2) as married couples (25.1) had less than a high-school education. In terms of time in Hong Kong, the percentage of local parents in married-couple families increased significantly, from 21.0 % in 1981 to 47.2 % in 2011, while the percentage of immigrant parents decreased from 74.4 to 38.1 % over the same period. Although the percentage of local and immigrant parents in single-mother households followed a similar trend, the rate of change was more moderate and resulted in more immigrant parents in single-parent families (40.8) than in married-couple families (38.1) in 2011. In 1981, a majority of single parents were widows/widowers or never married (90.1 %), while only a few were divorced or separated (9.9 %). This situation reversed in 2011 when most single parents were divorced or separated (79.5 %).

7.3 Components of Change

While the social and demographic characteristics of single-parent and married-couple families may affect the poverty risk of children living in these families, the association between these characteristics and poverty may change over time due to changes in the broader socioeconomic context. For instance, a person with a university degree will have a different level of poverty risk in 1969 than in 1999. To isolate the effect of the change in characteristics (e.g., whether or not the person has a university degree) and the change in the risk level associated with that characteristic (i.e. the poverty risk associated with a university degree), we decomposed the changes in poverty for each decade between 1981 and 2011 into a composition component (family characteristics we measured) and a rate component (the contextual change that altered the poverty risk level of that characteristic). The result of our estimation is shown in Table 4. In the upper portion of the table, we hold the composition constant at the end of each decade to illustrate the rate component, and hold the coefficients constant at the beginning of each decade to illustrate the composition component. In the lower portion of the table, we hold the composition constant at the beginning of each decade and the coefficients constant at the end.

Table 4 Changes in poverty that are due to changes in composition versus changes in subcategory poverty risks (Poverty = 1 if income <50 %)
Full size table

The differences between these two sets of estimates represent the relative effect of changes in family characteristics and in social conditions on the risk of child poverty. For instance, the poverty rate of children in married-couple families would have declined by 26.4 % between 1981 and 2011 based on the change in composition when we used the 1981 coefficients for calculation. However, the decline in poverty would have reduced to 10.4 % if we used the 2011 coefficients. This means that the poverty risk for children in single-parent families tended to be higher in 2011 than in 1981 for the characteristics these families shared. Similarly, if we fixed the composition of married couples at 2011, the effect of the change in the rate component (32.8) would be much higher than that when we fixed the composition at 1981 (16.8). A possible explanation for this difference is that although more married couples had attained a university education in 2011, the depreciation of value of a university degree was also much larger toward the end of the period of assessment.

A comparison of the poverty risks of single-parent and married-couple families shows that the family characteristics of married-couple families have stronger effects on poverty reduction than those of single-parent families. As can be seen in Table 4 (left panel, upper portion), the poverty risk of children in single-parent families increased by 24.9 % during the period 1981–2011. This was mainly due to changes in the socioeconomic context because the rate component contributed 34.7 % of the increase in poverty over the past three decades, while the characteristics of single-parent families (i.e. the composition component) offset the poverty risk to children in these families by 9.8 %. Similar trends are observed among married-couple families, but the changes in composition have a stronger effect on reducing their poverty (26.4 %). The result is a widening poverty gap between children in single-parent and married-couple families over the period 1981–2011.

7.4 Compositional Subcomponents

In this section, we examine the individual compositional factors that contribute to the widening of the gap between children in single-parent and married-couple families. The decomposition result of the family characteristics of both types is shown in Table 5. As in Table 4, we show here two sets of results with variations in terms of the year at which the coefficients are fixed. The decomposition result shows that the effect of family characteristics for married-couple families in reducing the child poverty risk is much stronger than that for single-parent families if we fix the coefficient at the beginning of each period. Several factors contribute to reducing the child poverty risk for married-couple families. Parental education level reduces the risk by 1.71 %, 2.68 %, and 4.83 % in each respective decade between 1981 and 2011, and this effect was consistently higher than for single-parent families over the same period. If we look at the overall poverty reduction rate during the whole period from 1981 to 2011, the contribution of parental education for married-couple families (11.6), defined as the highest educational attainment of the couple, is more than double that for single-parent families (5.02). It is notable that the effect of parental education in alleviating poverty is further reduced when evaluated using coefficients at 2011 than at 1981 for both single-parent and married-couple families.

Table 5 Contributions of changes in composition to changes in child poverty (Poverty = 1 if income <50 %)
Full size table

Parental employment also contributed to widening the poverty gap between single-parent and married-couple families, but in the opposite direction. As shown in Table 5, parental employment increased the child poverty risk for both family types, but the impact was stronger for single-parent families (5.57) than for married-couple families (0.29). Other factors such as family size and parental experience all contributed, albeit to a lesser extent, to widening the poverty gap between single-parent and married-couple families. With regard to factors unique to single-parent families, parental gender had the effect of increasing the poverty risk in the 1980s and 1990s, but its effect became more or less neutral in the 2000s. The route to single parenthood increased the child poverty risk in the 1980s but reduced it thereafter.

8 Discussion

In this study, we compare the trends of child poverty risk for single-parent and married-couple families in Hong Kong and assess the contributory factors. Over the past three decades, the composition of single-parent families in Hong Kong has changed significantly, with most becoming single parents through divorce and separation rather than through death. This trend is similar to that in most Western societies, except that single parenthood resulting from premarital birth only accounts for a small percentage of cases in Hong Kong in recent decades (Ambert 2006; Chzhen and Bradshaw 2012; Bianchi 1994). Similarly, as is the case in most European cities, single-parent families in Hong Kong tend to have fewer and older children than their married counterparts (Chzhen and Bradshaw 2012). Given the changes in composition and size, and the higher average age of both parents and children, we would expect most single-parent families to be living in better economic conditions nowadays than in the past. However, our results show that their risk of child poverty continued to increase throughout the period of study. In addition, there is a significant poverty gap between single-parent and married-couple families, which has widened in the past three decades with more children in the former being exposed to poverty. The main reason for this is that although the living conditions of both single-parent and married-couple families have improved, the improvement for married-couple families occurred at a much faster rate than that for single-parent families. This trend is consistent with earlier findings in European and North American contexts that showed a similar gap between family types (Chzhen and Bradshaw 2012; Yoo and Lim 2009).

Among the family characteristics we examined in this study, parental education level and labor market participation account for a substantial component of the poverty gap between single-parent and married-couple families. The situation of single parents in Hong Kong is similar to that in most European countries where they are less likely to have received higher education than married parents (Chzhen and Bradshaw 2012). Although the number of Hong Kong single parents who have had higher education has increased, in 2011 over 85 % still had only a high school education or below compared to about 62 % of married parents. A possible explanation to the continue low educational level of single parents is that many single-parent families were the result of divorces in cross-border marriages, in which the immigrant mothers usually have low educational level. With regard to labor market participation, fewer single than married parents were in employment, consistent with other studies predicting low labor participation for single parents because of childcare responsibilities (Marsh 2001; Millar and Ridge 2009). However, not only is the labor market participation of single parents in Hong Kong lower than that of married couples, it dropped conspicuously between 1981 and 2001, and has only slightly recovered during the last decade.

A number of different factors have contributed to this trend. Firstly, since the percentage of single mothers has grown substantially over the past three decades and the majority of current single-parent households are headed by a female, the decrease in labor market participation and corresponding increase in poverty risk is consistent with studies showing a negative relationship between single motherhood and employability, and the adverse consequences of this on children’s wellbeing (Millar and Ridge 2009; Snyder et al. 2006). Secondly, the changing socioeconomic context has also contributed to the widening gap in the labor market participation rates of single-parent and married-couple families. As discussed earlier in this paper, the economic restructuring in Hong Kong since the 1980s has had a negative impact on both employment opportunities and wages for workers with low skills and little education. This social trend is more unfavorable to single parents, who tend to have less education than their married counterparts, thus worsening their socioeconomic conditions over the past three decades. The effect of the socioeconomic context in Hong Kong on single parents is similar to that on other disadvantaged groups such as immigrants (Chou et al. 2013).

As well as employment participation and the effect of the changing socioeconomic context, family size and composition have also affected the economic condition of these families. As the fertility rate of two-parent families has decreased more rapidly than that of single-parent families over the past three decades, so has their financial burden with regard to childcare. In other words, the living conditions for children in two-parent families have improved faster, thus further setting them apart from children in single-parent families.

The results of this study have important policy implications for the government’s strategy for tackling child poverty in single-parent families in Hong Kong. Firstly, our findings show that the poverty gap between single-parent and married-couple families has continued to rise in the past three decades. This indicates that the fruits of Hong Kong’s economic growth have not been shared evenly among these two groups, and the economic status of single-parent families has not improved even in the face of the sustained economic success enjoyed during this period. The government may need to rethink its policy of relying on economic growth and the trickle-down effect on income distribution to achieve a reduction in poverty among disadvantaged groups such as single-parent families.

Secondly, although economic growth creates the necessary resources to provide financial support to single parents, cash allowances alone are insufficient as this approach does not address the structural causes of poverty in these families. According to our study, the majority of single-parent families in Hong Kong are headed by women whose labor market participation has declined. While these single mothers share the same context of economic restructuring that puts all low-skilled workers with limited education at a disadvantage, they also have unique attributes that keep them out of employment. In order to help single parents to return to the labor market, the government needs to introduce policy measures that reduce their childcare burden. These could include more publicly funded childcare support centers and policy initiatives that will encourage the implementation of family-friendly policies among employers, which will provide more flexibility for working lone parents in both the public and private sectors. Broader legislative measures that address possible discrimination against single mothers in the workplace may also be required.

This study has a number of limitations. Firstly, it is based on cross-sectional data, but the causal and temporal relations between risk factors and poverty status in single-parent and married-couple families in Hong Kong must be ascertained using longitudinal data. Secondly, the threshold of poverty—that is, half the median household income—was a relative one. This is also a solely income-based measure that does not take into account family assets. Further studies need to be conducted to generalize our findings by using an absolute threshold of poverty and a measure based on the material deprivation and wellbeing of children. Thirdly, while the overall sample size was relatively large, the sizes of some specific cells, such as those with at least one parent being a new arrival, or households with four or more children, were quite small. This may have reduced the statistical power of our data analysis. Lastly, although we included a wide range of characteristics of children, parents, and households among the factors contributing to the child poverty risk, some other important issues, such as the history of marriage and employment, as well as the relationship between single parents and their former spouses, were not examined because they are not covered in Census data. This issue is one of the common limitations faced by studies employing secondary data analysis and future studies need to be undertaken to address these further, potentially important, factors.

In this study, we analyzed four cross-sectional datasets from the Hong Kong Population Census collected over the past three decades to identify trends in child poverty rates and the factors underlying these trends for both single-parent and married-couple families. To summarize, the differences in the position of single-parent and married-couple families in Hong Kong generally conform to the trends found in other developed societies. Our findings show that the socioeconomic and demographic characteristics of single parents in Hong King have been changing, with more divorced single mothers heading families with dependent children. Although single parents’ education levels have improved, in general these families still possess less human capital than their married counterparts. These differences made single-parent families more vulnerable to the changing socioeconomic context in Hong Kong in the 1980s and 1990s, which led to a widening gap in labor participation between family types. The resulting trend is that children in single-parent families have suffered and continue to suffer from an increased risk of poverty and their economic outlook remains much worse than that of children in married-couple families.