Trends and dynamics of homelessness in Hong Kong: evidence from population survey in 2015 and 2021

Abstract

This study examines trends in homelessness in Hong Kong, comparing data from surveys conducted in 2015 and 2021. The research aims to identify trends and contributing factors, explore context-specific factors, analyze the relationship between housing, employment, and homelessness, and determine factors influencing its duration. Data from the H.O.P.E. Hong Kong 2015 and Hong Kong Homeless Census 2021 surveys were analyzed, encompassing demographic information, conditions of homelessness, and social relationships, health, and personal behavior. Descriptive statistics, chi-square tests, and ANOVA analyses were employed. The study identifies significant differences in the duration of homelessness based on demographics and employment status. Individuals with disabilities, those unemployed for longer periods, and CSSA recipients experience longer durations. Regular contact with family and friends was associated with shorter durations of homelessness. The study emphasizes the importance of improving access to public housing, addressing employment challenges, and providing support services to the homeless population.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-24749-y.

Introduction

Homelessness is a critical issue worldwide, particularly in global cities like New York City and Tokyo, where high living costs and socio-economic factors drive its prevalence [1]. In New York, a staggering 62,391 people use homeless shelters nightly, with a 59% increase in service use over the past decade and an annual 8.6% rise in female shelter residents since 2015 [2, 3]. Conversely, in Tokyo, homelessness primarily affects older male blue-collar workers, with street homelessness dropping from 5,400 in 2004 to just over 700 in 2017 [4]. The UK sees diverse patterns, such as a doubling of rough sleepers in England from 2012 to 2015 and a 20% increase in homelessness from rental losses in Wales from 2009 to 2014 [5]. Despite various initiatives, challenges like inadequate affordable housing, rising income inequality, and legal mandates to provide temporary shelter complicate solutions, emphasizing the need for comprehensive strategies that address both the symptoms and causes of homelessness [6].

Understanding of homelessness

The concept of homelessness varies widely, influenced by differing legislative definitions in countries like the United Kingdom, the United States, and Australia. The UK has enacted multiple Homelessness Acts (1977, 1996, 2017) which mandate local authorities to assist eligible, unintentionally people experiencing homelessness (PEH) in urgent need [7]. In the US, the definition has evolved with the McKinney-Vento Act [8] and the HEARTH Act [9], which categorize PEH into four groups, including individuals and families lacking a stable nighttime residence, being at imminent risk of losing it, being classified as homeless under other federal laws, or fleeing dangerous situations like domestic violence [10]. Australia’s Supported Accommodation Assistance Act and the Australian Bureau of Statistics describe homelessness as lacking secure and adequate housing due to various reasons, including inadequate dwellings and unstable tenure [11, 12].

Homelessness is triggered by factors like eviction, relationship breakdowns, and job loss [13], viewed as immediate causes within broader theoretical frameworks. Positivist approaches focus on empirical relationships between such factors and homelessness, while social constructionism considers homelessness a result of societal norms and values shaping perceptions [14]. Feminist theories discuss how gender inequalities contribute to homelessness, emphasizing the role of systemic gender discrimination. Critical realism blends these views, acknowledging both individual and structural causes and exploring the interplay between societal structures and individual circumstances in perpetuating homelessness [15].

Homelessness and socio-economic development

Homelessness is deeply intertwined with socio-economic policies, exhibiting significant variations across nations like Japan, the United States, and the United Kingdom. In Japan, comprehensive policies including an increase in shelters and stringent public space management have reduced visible homelessness [4], though they may not fully address its root causes, combining punitive and welfare strategies [16]. In the U.S., public assistance programs like Supplemental Security Income crucially support vulnerable groups, though their effectiveness is curtailed by strict eligibility criteria and insufficient housing integration [17]. The UK approaches homelessness with a blend of housing policies, welfare benefits, and social services, yet continues to face challenges, necessitating a more integrated strategy [18].

These efforts underscore the complex relationship between homelessness and socio-economic development, highlighting the need for multifaceted solutions that tackle root causes and provide immediate relief. Discrepancies in resource allocation and persistent health disparities suggest that addressing homelessness also requires addressing underlying social, economic, and political inequities [19–21]. Further research is essential to refine these policies and explore innovative solutions, ensuring they effectively address the diverse needs of PEH.

The context of Hong Kong

Homelessness in Hong Kong can be traced back to the post-World War II era, evolving significantly due to changing social welfare policies [22]. Initially boosted by overseas capital, the welfare system saw support wane by the 1960s. However, in the 1970 s, the government began supporting NgOs to maintain stable governance, acknowledging nearly 10,000 people lacked stable housing [23]. The 1980 s marked a shift towards more proactive welfare measures aimed at marginalized groups, including the PEH, promoting registration and collaborative efforts for welfare solutions [24–26].

The period from 1993 to 2001 saw attempts to provide temporary housing, though the impact on human rights and housing prices remained contentious [24]. Post-1997, the adoption of neoliberal policies led to reduced support for the marginalized, which, following the financial crisis, was recognized as a welfare issue addressed through NGO-led initiatives [27]. A significant governmental effort was the “Three-year Action Plan to Help the Street Sleepers” initiated in 2001, leading to a temporary reduction in homelessness [28, 29]. However, by 2015, the numbers had risen, highlighting the transient effectiveness of these measures and the challenge of hidden homelessness, such as those in 24-hour fast food outlets [24, 30].

Living conditions for the PEH remain dire, with many residing in unsafe or illegal accommodations as cage homes, partitioned units, or improvised shelters in public spaces [31]. These environments contribute to significant health and psychological risks, including higher rates of lung disease, psychiatric disorders, and self-harming behavior [30, 32, 33]. In response, the Hong Kong government and NGOs have developed several initiatives. The “Three-Year Action Plan to Help the Street Sleepers” launched in 2001, and more recently, Integrated Services Teams for Street Sleepers (ISTs) provide outreach, case management, and temporary shelters [28, 29]. Transitional housing projects and long-stay hostels also offer alternative pathways to housing [34]. However, these services remain insufficient. Primary healthcare support is fragmented and largely NGO-led rather than systematically provided by public services, leaving many without adequate medical care [35]. Waiting times for institutional accommodation are long, and eligibility criteria are restrictive, compounding the structural and personal challenges faced by the homeless and heightening risks of addiction and criminal involvement [32, 33, 35].

Aims of the study

The present study aimed, (1) to analyze the demographic and socioeconomic profiles of people experiencing homelessness in 2015 and 2021; (2) to examine the relationships between housing, employment, and homelessness in a comparative context; (3) to investigate factors influencing the duration of homelessness.

Methods

Data and sample

This study utilizes survey data from two projects, H.O.P.E. Hong Kong 2015 (HOPE HK 2015) [36] and Hong Kong Homeless Census 2021 (Homeless Census 2021) [37], which aimed to examine the homeless situation and the characteristics of PEH in Hong Kong in the respective years. Academic institutions and NgOs conducted both surveys in 2015 and 2021, respectively, and they were the most extensive territory-wide surveys of the PEH in Hong Kong at the time.

The data for HOPE HK 2015 was collected through an overnight street count on 29 October 2015. More than 300 student volunteers from six universities participated, with 40 of them serving as team leaders. Each team was assisted by at least one social worker from each community organization. Before the survey, volunteers attended two training sessions co-organized by community organizations and universities on 4 and 16 October 2015. During these sessions, volunteers were trained on the causes of homelessness, the background of HOPE HK 2015, conversational techniques to engage with homeless individuals, and locations for street counts. Social workers also accompanied volunteers on a night visit approximately a week before the survey, during which they had the opportunity to become familiar with the survey’s rundown and to visit and talk with homeless individuals to build relationships and prepare them for the survey. During the street count, social workers helped handle any problems that arose and ensured the safety of volunteers while guiding the flow of interviews. The interviewers successfully collected 641 questionnaires. Of the collected questionnaires, 372 were valid for analysis. The HOPE HK 2015 methodology and results are detailed in a report [36].

The data for the Homeless Census 2021 was collected from the evening of 9 July 2021 to the early morning of 10 July 2021. Over 300 volunteers from NgOs, universities, and tertiary institutions were trained as interviewers through sessions co-organized by universities and NgOs. Social workers planned the survey routes to cover over 360 hot spots of homeless individuals in various Hong Kong districts. A team consisting of two to three interviewers and one to two team leaders was assigned to each route. The survey attempted to reach all homeless individuals in Hong Kong during the data collection period and revisited the same location thrice if they found empty beds without occupants. Interviewers obtained consent from each participant before inviting them to complete the questionnaire. The survey successfully collected 719 questionnaires. Of the collected questionnaires, 711 were valid for analysis. The Homeless Census 2021 is outlined in another report [37] and other publications [35, 38].

Pilot testing for both surveys was conducted in collaboration with academics, social workers, and individuals with lived experience of homelessness. Feedback from NgO workers, scholars, and outreach-recruited homeless individuals informed adjustments to enhance clarity, usability, and accessibility. Revisions included simplifying open-ended questions, refining language, and providing interviewer assistance for participants with literacy or cognitive needs. Core questions remained consistent between the 2015 and 2021 surveys to ensure comparability. These included key variables such as demographics, homelessness duration, causes, housing applications, social relationships, health, and behaviors. However, the 2021 survey included additional questions addressing emerging issues, such as the impact of COVID-19 on unemployment and mental health.

Measurements

Demographic and socio-economic variables

The demographic and socio-economic characteristics of the respondents were used in the analysis, including sex, age, education level, marital status, ethnicity, and current employment status. The questionnaire was designed to be brief and concise, ensuring that it would be easily understood by the participants and would not impose a significant burden on them.

Conditions and characteristics of PEH

The duration of those respondents in both survey had been experiencing homelessness was assessed by requesting them to select a category indicating the duration from the following options: less than 6 months, 6 months to 1 years, 1 to 2 years, 2 to 3 years, 3 to 5 years, 5 to 10 years, 10 to 20 years, or 20 years or more; the mid-point of these pre-established intervals was subsequently utilized as a continuous variable for further analysis. Besides, two open-ended questions were included to determine if respondents were experiencing homelessness for the first time and to assess the frequency of their mobility during the two years prior to becoming homeless.

Respondents were instructed to select one or more options that accurately represented the reason(s) of their homelessness. The options are then categorized into four distinct groups: housing-related factors (e.g., rent too expensive, evicted or rejected by the landlord, previous accommodation demolished/redeveloped, convenient for daily life/work, previous accommodation too crowded/conditions too poor, previous accommodation infested by fleas, and evicted from temporary shelters); employment-related factors (e.g., to save money, and became unemployed); personal reasons (e.g., homeless after being discharged from prison/hospital/drug treatment centre; drug/alcohol problems, health reasons, personal choices, and gambling addiction); family reason (e.g., had relationship problems with family/tenants; family in mainland China/overseas). These response alternatives were derived from a comprehensive review of existing literature on homelessness and were designed to encompass the most frequently reported causes in previous research [39–42].

A single question was posed to ascertain whether respondents had pursued public housing opportunities. Furthermore, respondents were asked to select one or more options that accurately described their reasons for not applying for public housing. The reasons include complicated application procedure, long waiting period, no preferred district, in process of a divorce, family members in mainland, personal reason, not eligible. Those two questions aimed to identify the barriers faced by homeless individuals when considering public housing as a potential solution.

Social relationship, health, and personal behaviour

The social relationship, physical health and personal behavior of the PEH were examined in the survey. In terms of social relationships, two inquiries were posed, ‘whether they have regular contact with family and friends and ‘whether they contact social workers or social service agencies’. The respondents were given the option to answer with either a ‘yes’ or ‘no’ to each question. As for the physical health questions, participants were asked whether they suffered from chronic illnesses that required frequent treatment, whether they were prescribed psychotropic medications within the previous six months, and whether they had any physical disabilities. They were required to answer either ‘yes’ or ‘no’ to each query. Concerning personal behavior, the questions were about gambling, alcohol consumption, and drug usage, to which participants could only respond with either ‘yes’ or ‘no’.

Statistical analysis

Descriptive statistics were generated for demographic information of the cohort, conditions of characteristics of the homelessness, and social relationship, health, and personal behavior. Chi-square tests were used to compare the proportion of the characteristics across years. ANOVA was conducted to compare the duration of homeless experience in sub-populations with different demographic backgrounds. Significance level was set at 0.05, and p-values were reported. Missing data were handled via listwise deletion. Analyses were conducted using SPSS version 28.

Results

Comparison of demographic information of people experiencing homelessness

The demographic and socioeconomic profiles of homeless individuals surveyed in 2015 and 2021, were shown in Table 1. The majority of respondents were male, with the proportion of females increasing from 7.6% in 2015 to 16.1% in 2021. The median age rose from 55 in 2015 to 58 in 2021, with an increased proportion of respondents aged 60 or above in 2021 (46.2%) compared to 2015 (37.9%). Education levels in 2021 exhibited a higher percentage of respondents with tertiary or above education (9.3%) compared to 2015 (3.4%). Marital status distribution remained relatively stable, with single and separated or divorced individuals forming the majority.

Table 1.

Demographic and socioeconomic backgrounds

	2015		2021
	N	%	N	%
Sex
- Male	354	92.4	594	83.8
- Female	29	7.6	114	16.1
Age
- 18–29	8	2.2	16	2.3
- 30–39	29	8.1	41	5.9
- 40–49	71	19.8	108	15.6
- 50–59	115	32.0	208	30.0
- >=60	136	37.9	320	46.2
- Median	55		58
Education
- Primary or below	152	46.5	236	35.1
- Secondary	164	50.2	380	56.5
- Tertiary or above	11	3.4	56	9.3
Marital Status
- Single	146	42.7	267	39.7
- Married	63	18.4	118	17.6
- Separated/Divorced	127	37.1	268	39.9
- Widowed	6	1.8	18	2.8
Ethnicity
- Chinese	332	90.0	641	91.3
- Non-Chinese	37	10.0	61	8.7
Current Employment Status
- Working	135	36.1	219	31.1
- Not working	239	63.9	486	68.4
Current/Last Income from Work
- Median	6000 HKD		8000HKD
Duration of Unemployment
- Under 3 months	18	8.9	49	10.5
- 3 to 6 months	18	8.9	23	4.9
- 6 months to 1 year	11	5.4	49	10.5
- 1 to 2 years	22	10.8	73	15.7
- Longer than 2 years	134	66.0	271	58.3
Unemployed because of COVID-19 pandemic
- Yes	/	/	104	25.4
- No	/	/	274	66.8
- Not sure	/	/	32	7.8
CSSA recipients
- Yes	179	48.8	317	45.3
- No	188	51.2	383	54.7

Most respondents were of Chinese ethnicity (90.0% in 2015 and 91.3% in 2021), with a slight decrease in non-Chinese respondents from 10.0% in 2015 to 8.7% in 2021. In terms of employment, a higher percentage were not working in 2021 (68.4%) compared to 2015 (62.4%). The median current or last income increased from 6000 HKD in 2015 to 8000 HKD in 2021. The majority of unemployed respondents had been so for over two years in both years, with a slight decrease from 66.0% in 2015 to 58.3% in 2021 In 2021, 25.4% reported unemployment due to COVID-19. The percentage of CSSA recipients decreased from 48.8% in 2015 to 45.3% in 2021.

Of the interviewee, 28.7% of respondents had been homeless for 5 or more years in 2015, comparing to 22.8% in year 2021. Notably, the percentage of respondents experiencing homelessness for 2 to less than 5 years dropped to 22.7%, and for 5 years or more decreased to 22.8% in 2021.

Comparsion of homeless experience

The median duration of homelessness decreased from 30 months in 2015 to 18 months in 2021. A larger proportion experienced homelessness for the first time in 2015 (62.8%) than in 2021 (58.6%). The median number of recurrences remained stable at three times for both years whilst the mean value are 4.14 times in 2015 and 4.72 times in 2021. Housing insecurity was the most common cause of homelessness (68.0% in 2015; 64.4% in 2021), followed by economic insecurity (34.3% in 2015; 36.0% in 2021), personal reasons (25.4% in 2015 and 24.4% in 2021), and family issues (21.2% in 2015 and 21.3% in 2021). The proportion applying for public housing increased slightly from 31.6% in 2015 to 34.4% in 2021. The median waiting time for public houses increased from 30 months in 2015 to 36 months in 2021 (See Table 2). In 2015, the most common reasons for not applying for public housing were related to complicated application procedures or uncertainty about how to apply (43.9%), followed by long waiting periods (22.9%). In 2021, the top reasons were similar, with complications in application procedures or uncertainty decreasing to 27.9%, and long waiting periods increasing to 26.1%.

Table 2.

Conditions and characteristics of homelessness

	2015		2021
	N	%	N	%
Duration of homeless
- Less than 3 months	41	11.2	157	23.6
- 3 to less than 6 months	32	8.7	48	7.2
- 6 months to less than 1 year	34	9.3	58	8.7
- 1 to less than 2 years	61	16.7	100	15.0
- 2 to less than 5 years	93	25.4	151	22.7
- 5 years or above	105	28.7	152	22.8
- Median	30 months		18 months
- Mean	56 months		46 months
Homeless for the first time
- Yes	223	62.8	350	58.6
- No	132	37.2	247	41.4
Homelessness Recurrence (how many times)
- Median	3 times		3 times
- Mean	4.14 times		4.72 times
Causes of homelessness
- Housing related	214	68.0	397	64.4
- Economic related	108	34.3	222	36.0
- Personal reasons	80	25.4	150	24.4
- Family reasons	67	21.2	131	21.3
Applied for public housing
- Yes	118	31.6	242	34.4
- No	255	68.4	461	65.6
Reasons for Not Applying Public Housing
- Application procedure too complicated/not sure how to apply	98	43.9	124	27.9
- Waiting period is too long, unhelpful	51	22.9	116	26.1
- No units in preferred district/urban area found	15	6.7	14	3.2
- Already have a public housing unit/in process of a divorce	27	12.1	90	20.3
- Families in mainland	0	0.0	10	2.3
- Personal reasons	49	22.0	88	19.8
- Others	33	14.8	165	37.2
- Not eligible (income, asset, or identity)	/	/	51	11.5
Waiting time of public housing
- Median	30 months		36 months
- Mean	43 months		41 months

Comparison of homeless social relationship, health and personal behavior

Table 3 shows the social relationships, health, and personal behavior of homeless individuals in 2015 and 2021. Regular contact with family and friends increased from 44.5% in 2015 to 52.2% in 2021, and contact with social workers or agencies increased from 59.1% in 2015 to 73.2% in 2021. The proportion with chronic diseases requiring regular treatment rose from 31.3% in 2015 to 39.0% in 2021, while the use of psychotropic medications remained stable. The proportion of physically disabled respondents was similar in both years. Personal behavior Changes included a decrease in gambling habits from 26.6% in 2015 to 17.7% in 2021, a decrease in alcohol consumption from 28.8% in 2015 to 19.4% in 2021, and a decrease in drug abuse from 12.8% in 2015 to 8.2% in 2021.

Table 3.

Social relationship, health and personal behavior

	2015		2021
	N	%	N	%
Regular Contact with Family and Friends
- Yes	163	44.5	364	52.2
- No	203	55.5	333	47.8
Contact with Social Workers/Social Service Agencies
- Yes	220	59.1	507	73.2
- No	152	40.9	186	26.8
Chronic Diseases Which Require Regular Treatment
- Yes	113	31.3	272	39.0
- No	248	68.7	425	61.0
Have You Taken Any Psychotropic Medications in the Past 6 Months
- Yes	24	24.0	68	25.4
- No	76	75.0	200	74.6
Physical Disabled
- Yes	52	14.8	108	15.5
- No	299	85.2	589	84.5
Gambling Habit
- Yes	91	26.6	123	17.7
- No	251	73.4	573	82.3
Habit of Drinking Alcohol
- Yes	101	28.8	135	19.4
- No	250	71.2	560	80.6
Drug abuse
- Yes	43	12.8	56	8.2
- No	294	87.2	629	91.8

Comparison of homeless duraction across different demographic populations

An ANOVA analysis of the duration of homelessness were conducted, based on various factors in 2015 and 2021, as shown in Table 4. Notably, there was a significant difference in the duration of homelessness based on sex in both 2015 (F = 5.14, p <.05) and 2021 (F = 4.64, p <.05). Female individuals reported an average duration of 25.1 months, while their male counterparts experienced an average duration of 58.1 months, which was more than a double. In 2021, the pattern persisted, with females experiencing an average duration of 32.8 months, while males averaged 48.2 months.

Table 4.

ANOVA of duration of homeless

	2015				2021
	M	SD	F	p	M	SD	F	p
Sex
- Male	58.1	75.2	5.14	0.024	48.2	64.7	4.64	0.032
- Female	25.1	30.7			32.8	49.8
Age
- 18–29	12.3	19.5	4.28	0.002	17.8	28.6	3.81	0.005
- 30–39	27.0	38.4			31.3	52.0
- 40–49	37.1	45.8			38.0	53.5
- 50–59	62.4	73.2			39.7	57.5
- >=60	66.5	81.5			55.8	70.3
Education
- Primary or below	65.8	76.7	4.30	0.014	55.4	69.5	5.87	0.003
- Secondary	42.6	63.0			37.5	54.6
- Tertiary or above	47.5	46.1			46.9	69.6
Marital Status
- Single	53.9	64.1	1.33	0.264	46.0	60.8	1.44	0.218
- Married	40.8	64.6			35.2	56.1
- Separated/Divorced	63.1	86.0			50.1	68.1
- Widowed	69.8	85.7			30.9	43.9
Ethnicity
- Chinese	59.1	76.1	4.04	0.045	47.9	64.4	3.88	0.049
- Non-Chinese	33.1	46.7			31.1	49.5
Current Employment Status
- Working	57.6	73.7	0.10	0.758	37.9	57.0	2.88	0.090
- Not working	55.1	74.2			50.0	65.4
Duration of Unemployment
- Under 3 months	23.3	26.3	2.72	0.020	29.0	58.2	8.02	< 0.001
- 3 to 6 months	30.6	69.9			68.2	89.3
- 6 months to 1 year	36.5	54.0			19.8	33.6
- 1 to 2 years	33.8	53.8			33.1	47.6
- Longer than 2 years	69.7	77.8			62.3	69.3
CSSA recipients
- Yes	65.5	79.2	4.91	0.027	53.4	68.4	7.63	0.006
- No	48.1	68.0			39.8	57.4
Homeless for the first time
- Yes	50.0	63.3	5.70	0.018	38.5	56.5	25.0	< 0.001
- No	68.4	87.1			65.8	72.2
Causes of homelessness
- Housing insecurity	52.3	68.6	2.90	0.089	46.1	61.9	0.01	0.931
- Not Housing insecurity	68.2	88.4			46.5	65.2
- Economic insecurity	51.4	78.8	0.94	0.334	34.7	50.4	11.3	0.001
- Not Economic insecurity	60.3	73.9			52.0	68.0
- Personal reasons	73.5	94.8	4.73	0.030	45.8	63.8	0.008	0.929
- Not Personal reasons	51.8	67.3			46.4	63.1
- Family issue	77.8	94.1	6.10	0.014	54.6	66.8	2.88	0.090
- Not Family issue	51.7	69.0			44.2	62.2
Applied for public housing
- Yes	60.2	74.9	1.00	0.367	38.4	52.6	5.13	0.024
- No	48.6	71.6			50.0	67.2
Regular Contact with Family and Friends
- Yes	54.1	73.7	0.14	0.707	40.2	59.8	5.06	0.025
- No	57.1	72.4			51.2	65.0
Contact with Social Workers/Social Service Agencies
- Yes	62.5	79.3	4.52	0.034	44.9	62.0	0.22	0.638
- No	45.8	60.4			47.5	64.0
Chronic Diseases Which Require Regular Treatment
- Yes	53.8	73.3	0.44	0.508	46.2	63.6	0.01	0.932
- No	59.6	75.4			45.7	62.1
Physical Disabled
- Yes	61.9	75.0	0.41	0.522	63.4	79.4	8.85	0.003
- No	54.9	72.7			42.8	58.8
Gambling Habit
- Yes	67.0	79.9	3.58	0.059	51.9	63.6	1.48	0.224
- No	50.6	65.6			44.2	62.1
Habit of Drinking Alcohol
- Yes	69.3	85.2	5.32	0.022	50.5	60.4	0.93	0.335
- No	50.0	62.5			44.6	63.5
Drug abuse
- Yes	60.1	72.8	0.23	0.632	65.9	73.1	6.09	0.014
- No	54.4	72.1			44.2	61.7

Age groups revealed diverse experiences in 2015 (F = 4.28, p <.05) and 2021 (F = 3.81, p <.01). Individuals aged 50–59 experienced an average duration of homelessness of 62.4 months in 2015, while those aged 60 and above reported an average duration of 66.5 months. In 2021, individuals in the 50–59 age range had an average duration of 39.7 months, whereas those aged 60 and above reported an average duration of 55.8 months. Education levels also displayed a noteworthy impact in 2015 (F = 4.30, p <.05) and 2021 (F = 5.87, p <.01). In 2015, respondents with primary education or below had an average duration of homelessness of 65.8 months, compared to those with tertiary education or above who reported an average duration of 55.4 months. By 2021, this pattern remained consistent, with individuals with primary education or below experiencing an average duration of 55.4 months, and those with tertiary education or above having an average duration of 46.9 months.

There is also a significant difference in ethnicity in 2015 (F = 4.04, p <.05) and 2021 (F = 3.88, p <.05). In 2015, non-Chinese individuals reported an average duration of homelessness of 33.1 months. In 2021, this figure slightly decreased to 31.1 months. Regarding the duration of unemployment, there are significant differences in 2015 (F = 2.72, p <.05) and 2021 (F = 8.02, p <.001). For those unemployed for longer than 2 years, the average duration of homelessness was 69.7 months in 2015. In 2021, this figure decreased slightly to 62.3 months. For CSSA recipients in 2015 (F = 4.91, p <.05) and 2021 (F = 7.63, p <.01), those people experienced an average duration of homelessness of 65.5 months in 2015 and 53.4 months in 2021. Those who had not experienced homelessness before reported an average duration of 68.4 months in 2015 (F = 5.70, p =.05), while in 2021 (F = 25.0, p <.001), this figure decreased to 65.8 months. Several patterns emerged in relation to the causes of homelessness. In 2015, those facing employment insecurity (F = 6.23, p <.05) had an average duration of homelessness of 39.0 months, whereas those not facing this issue averaged 63.6 months. By 2021, the difference persisted, with those facing employment insecurity (F = 6.45, p <.05) reporting an average duration of 36.0 months, and those not facing this issue experiencing an average duration of 50.0 months. Similarly, the average duration for individuals facing economic insecurity was 51.4 months in 2015, and this decreased to 34.7 months in 2021. However, the difference in marital status (F = 1.33, p =.264), current employment status (F = 0.10, p =.758), housing insecurity (F = 2.90, p =.089), and economic insecurity (F = 0.9, p =.334) were not significant in 2015, and the difference in marital status (F = 1.44, p =.218), current employment status (F = 2.88, p =.090), and housing insecurity (F = 0.01, p =.931) were still not significant in 2021.

In 2015, individuals without regular contact with family and friends experienced an average duration of homelessness of 57.1 months (M = 57.1, SD = 72.4), which decreased to 51.2 months (M = 51.2, SD = 65.0) in 2021. In 2015, individuals with such contact experienced an average duration of homelessness of 62.5 months, which decreased to 44.9 months in 2021. Regarding the factor of physical disabled, in 2015, individuals with physical disabilities had an average duration of homelessness of 61.9 months (M = 61.9, SD = 75.0)), which increased slightly to 63.4 months (M = 63.4, SD = 79.4) in 2021. Individuals with a habit of drinking alcohol reported an average duration of homelessness of 69.3 months in 2015, while this decreased to 50.5 months in 2021. For drug abuse, in 2015, those with a history of drug abuse had an average duration of homelessness of 60.1 months, which increased slightly to 65.9 months in 2021.

Discussion

Novel findings

This study examined trends and factors influencing homelessness in Hong Kong by comparing survey data from 2015 to 2021. Key findings include an increase in elderly and female individuals among people experiencing homelessness, a rise in chronic health conditions, and a decrease in harmful behaviors such as gambling and alcohol use. The median duration of homelessness also dropped from 30 months in 2015 to 18 months in 2021, indicating potential improvements in support services. However, persistent challenges such as unemployment, long waiting time for public housing, and the impacts of COVID-19 highlight the need for targeted interventions, and significant ANOVA differences in duration by demographics [43–45].

The increasing proportion of elderly PEH points to the effects of population aging, inadequate pensions, and rising healthcare costs. This trend parallels findings in Japan, where the homeless population has also aged considerably over the past two decades [4], while differing from patterns in New York and England, where youth homelessness and family homelessness remain more prominent concerns [5].

Unemployment remained a central driver of PEH, with the COVID-19 pandemic intensifying both economic instability and health risks. Consistent with international findings, pandemic-related job losses deepened housing precarity among already marginalized groups [46–50]. For example, research in the United States and Canada documented that COVID-19 not only disrupted employment but also heightened food and housing insecurity among low-income households [47, 48]. In Hong Kong, Chan et al. [35] demonstrated how pandemic-related shocks magnified multidimensional precarity, particularly the interaction between insecure labor, poor health, and unstable housing, which aligns closely with our results.

Meanwhile, pandemic-related policy deepened housing precarity among already marginalized groups. In the United States, eviction moratoria temporarily shielded many at-risk households but did not eliminate housing insecurity [47, 49], while in England, government programs such as “Everyone In” offered emergency accommodation to rough sleepers [50]. By contrast, Hong Kong lacked comparable large-scale schemes, and NGO reports indicate that many homeless individuals remained in precarious housing or public spaces throughout the pandemic [35]. These findings highlight the importance of integrated prevention strategies that address employment, housing, and health together, echoing global calls for cross-sectoral responses to PEH.

Encouragingly, the median duration of homelessness declined from 30 months in 2015 to 18 months in 2021. This may reflect greater service engagement, as seen in Hong Kong research on long-stay hostels as transitional housing alternatives [34], or the gradual expansion of NGO-led welfare initiatives [31]. Internationally, similar reductions in duration have been linked to Housing First and rapid rehousing programs, which prioritize immediate housing alongside wraparound supports [51, 52]. However, the reduced proportion of first-time homeless suggests persistent barriers for those already unhoused, reinforcing the need for reintegration pathways as well as prevention [53, 54]. This dual challenge mirrors findings in Europe and North America, where prevention measures reduced new entries into homelessness but long-term homelessness remained entrenched [50, 55].

Patterns in health and behavior also shifted. While chronic disease prevalence rose, harmful behaviors such as gambling, alcohol use, and drug abuse declined, possibly reflecting targeted interventions. These results align with Hong Kong findings on multi-dimensional precarity and rough sleeping [35], which emphasize the interlinkages between health, labor instability, and housing insecurity.

Theoretical implications

The current study contributes to the evolving theoretical understanding of homelessness, emphasizing its interconnectedness with multiple societal and health issues, thereby challenging the notion of homelessness as an isolated problem. Through our descriptive analysis, the findings suggest that homelessness, unemployment, and health may be linked, highlighting the need for a more nuanced understanding of these potential interrelations in future policy-making and research. This study revealed that homelessness is substantially influenced by factors such as job loss and increased chronic disease incidence, thus illustrating the complex interplay of social determinants [56]. Consequently, our study provides preliminary evidence supporting the need for an integrated perspective that situates homelessness within a broader societal context. However, further research is required to substantiate this, given the methodological limitations of our study.

Homelessness, as underscored by the presented results, is a complex phenomenon shaped by a web of structural and individual factors. While societal changes and economic conditions, play a pivotal role, the root causes can be intricately connected [57]. However, these foundational drivers are interconnected, rendering their individual influences on homelessness challenging to isolate distinctly. The current finding illuminated the profound impact of substantial societal episodes, notably the COVID-19 pandemic, on the complexion of the PEH [53].

Consistent with the previous study, the binary debate between structural versus individual causes is made even more evident by current results [24]. Housing and economic insecurities emerged as dominant structural factors precipitating homelessness. This resonates with evidence from New York, where rising rents and limited affordable housing are key drivers of homelessness [58], and with Tokyo, where precarious employment and day-laborer markets shape entry into homelessness [4]. Yet Hong Kong’s case illustrates a hybrid scenario: both high housing costs and family-related factors, such as relationship breakdown, remain salient [59].

Another crucial factor in the equation is the intersection of health and homelessness. The increasing prevalence of chronic diseases among homeless individuals elucidates an urgent need to integrate healthcare services with housing and employment solutions [60, 61]. Individuals’ health status can profoundly affect their ability to find stable housing and secure employment, necessitating an integrative approach in the development of homelessness interventions. Meanwhile, the results also revealed the potential importance of social support networks in alleviating homelessness. The observed increase in contact between homeless individuals and their family, friends, and social services hints at the potential protective effects of strong social connections. This aligns with international evidence showing that interventions that explicitly build social capital can support transitions out of homelessness. Housing First programs in the United States and Europe, for example, provide immediate, stable housing alongside individualized support and community integration, thereby reducing chronic homelessness and improving wellbeing [51, 52]. Social enterprise and peer-support models further engage homeless individuals in work and community participation, enhancing resilience [60, 62]. Integrated care approaches that combine housing, healthcare, and social services have also proven effective in reducing recurrence of homelessness and improving health outcomes [50, 63]. Strengthening the social capital through such evidence-based interventions may therefore facilitate the transition out of homelessness, while simultaneously addressing broader health and social vulnerabilities [64, 65].

Policy implications

In recent years, the discussion of homelessness changed from “individual fault” to systematic problems [63, 66]. Historically, blame was largely placed on individual shortcomings or personal failures, casting homelessness as a result of personal inadequacies [67]. However, the current results corroborated a more recent and evolving perspective, which viewed homelessness through the lens of systemic failures and societal inadequacies [68]. This change in perception has given rise to criticism of measures that are punitive in nature. Such measures often manifest as evictions of homeless individuals from public spaces, further entrenching the vulnerabilities they face [68]. This tension between punitive and welfare approaches is evident not only in Japan, where policies oscillate between exclusion and support [4], but also in England, where welfare-oriented programs such as transitional housing exist alongside restrictive local ordinances targeting rough sleeping [69]. Hong Kong mirrors this contradiction: while transitional housing projects have expanded, punitive practices such as evictions from public spaces persist [35].

Drawing from our results, it is evident that systemic issues, like housing and economic insecurity, underpin a significant portion of homelessness cases. Therefore, measures that solely emphasize individual blame are not only misdirected but exacerbate the problem. Instead, comprehensive strategies that address the systemic roots of homelessness and provide supportive interventions are imperative. Our results highlight the need for systemic reforms that address the structural roots of homelessness, combined with supportive interventions at the individual level [55, 70].

There is also an increasing focus on studies related to social work and welfare measures that aim to improve the effectiveness and efficiency of supporting homeless individuals in reintegrating into society [71, 72]. While refining each welfare countermeasure may seem positive, it is noticed that there is a risk that it could strengthen the justification for excluding those who do not meet the specific criteria or requirements of these measures. In other words, if research is not conducted objectively and fails to acknowledge the complementary nature of welfare measures and punitive countermeasures, it may unconsciously serve the role of justifying the removal or elimination of homeless individuals from the streets. Furthermore, the perspectives of homeless individuals are crucial and research can be inclusive and acknowledge the resilience they possess [73].

For in-work homeless, there were above 30% of the homeless have jobs in Hong Kong. Jones, Ahmed [74] held the view that the simplistic assumption that employment alone can resolve homelessness. The emphasis on individual responsibility for work overlooks broader structural issues such as low pay, job insecurity, and the lack of affordable housing. In Hong Kong, it is imperative to prioritize the importance of promoting decent work opportunities and ensuring the availability of affordable housing in both the labor and housing markets. This study suggested that policies designed to prevent and alleviate homelessness may not succeed in a labor market that fails to provide sufficient income and stability; thus, the potential improvements in prevention policy, such as providing financial assistance to manage property transitions and removing discriminatory barriers imposed by landlords would be needed [62, 75].

Between 2015 and 2021, Hong Kong witnessed a surge in the number of female homeless individuals, reflecting an alarming trend, even as significant gender disparities persisted. This diverges from patterns in Tokyo, where male PEH remains the predominant issue [4]. However, it is more comparable to recent trends in England and the United States, where gendered vulnerabilities, such as domestic violence, wage inequality, and caregiving responsibilities, have increasingly been acknowledged as pathways to homelessness [76, 77].

To strengthen support for PEH, this study emphasizes the need for expanding transitional and light public housing programs to provide temporary stability and pathways to long-term solutions. Enhancing targeted healthcare services, particularly for mental health and chronic illnesses, and fostering collaboration between government agencies and NGOs can improve outreach and service delivery [63, 65]. Additionally, employment support programs, such as skill-building workshops and wage subsidies, are crucial for improving economic stability and facilitating reintegration into society [78, 79]. These recommendations offer actionable insights for policymakers addressing homelessness in Hong Kong.

Limitations

This study relies on data collected from surveys conducted in 2015 and 2021, which May not capture real-time Changes and nuances in the PEH. More frequent data collection would provide a more comprehensive understanding of the dynamics of homelessness. Also, focusing on the demographic shifts and trends in homelessness, but not delving deeply into the underlying causes and experiences of homelessness among different groups, is another limitation. Further qualitative research could provide richer insights into the experiences and perspectives of PEH. Moreover, the HOPE survey had a relatively high incompletion rate, which may have introduced selection bias. While the core survey questions remained consistent across 2015 and 2021, there were minor differences in supplementary questions, which could affect direct comparisons. The lack of multivariate analysis limits our ability to control for potential confounding factors, such as the interaction between health and employment status.

The research does not explore the perspectives and voices of homeless individuals themselves. Including their experiences and insights would provide a more holistic understanding of homelessness and inform more relevant and effective interventions.

Conclusions

The study highlights significant shifts in the demographic profile of homeless individuals in Hong Kong from 2015 to 2021, with increases in female, elderly, and educated homeless populations, challenging existing stereotypes and underscoring the need for targeted interventions. The COVID-19 pandemic has exacerbated issues of unemployment and health vulnerabilities, emphasizing the necessity for comprehensive solutions that address housing, healthcare, and economic factors. These findings expand the understanding of homelessness as a multifaceted issue intertwined with broader societal factors, advocating for integrated policy approaches that include housing, healthcare, employment, and social support, while cautioning against punitive measures and promoting inclusivity in policy development.

Abbreviations

PEH

People Experiencing Homelessness

Authors’ contributions

S.C. was responsible for the literature search, data analysis, data interpretation, and writing of this paper. H.W. was responsible for the overall research design, data collection of the data set. Y.C. and S.L. were responsible for the writing of this paper. G.C. and C.C. were responsible for data interpretation and writing of this paper. All authors read and approved the final manuscript.

Funding

N/A.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

In accordance with the Declaration of Helsinki, all participants were fully informed about the purpose and procedures of the study and provided their informed consent prior to participation. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.