Determinants of future anxiety across individual, household, and regional levels in South Korea using a social ecological model

Abstract

This study is the first to examine the determinants of future anxiety in South Korea using the Social Ecological Model (SEM). It aimed to show that, beyond individual factors, mezzo- and macro-level aspects, particularly those related to housing, may influence anxiety. Utilizing 2018 data from the Korean Health Panel Survey, we employed a three-level multilevel analysis to investigate how these factors contribute to the perception of future anxiety among Koreans. Our findings reveal that future anxiety is influenced by a complex interplay of demographic and socio-economic factors, including gender, age, health status, economic stability, living conditions, and regional socio-economic indicators. Notably, the study highlights the significant role of household-level factors, such as income and housing status, in shaping individuals’ anxieties about the future. Additionally, regional characteristics like housing vacancy rates and suicide rates are found to have an impact, suggesting the importance of broader socio-economic and cultural contexts in understanding future anxiety. Given South Korea’s socio-economic challenges, including the real estate crisis and demographic shifts, our study underscores the need for targeted policies to alleviate future anxiety, enhance mental well-being, and promote a more equitable society.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-87387-9.

Introduction

The future world is filled with uncertainty, as no one can be sure what will happen next¹. Especially in societies characterized by significant socio-economic volatility, such as contemporary contexts, individuals often experience future uncertainty. Moreover, when outcomes are contingent upon individual choices rather than dictated by deterministic patterns, individuals may become susceptible to anxiety due to the unpredictable future. The anxiety rooted in future negative thinking is highly likely to undermine mental well-being².

Uncertainty about the future is closely related to anxiety by its nature³. Zalesk⁴ defined the anxiety caused by future uncertainty as future anxiety. He described it as a condition marked by unease, doubt, fear, and worry regarding potential negative developments in one’s distant future. Future anxiety is essential because it can lead to fear, terror, and even depression or suicidal ideation⁵. Also, it is associated with health conditions such as hypertension, disability, and oral health^6,7.

Several studies have explored the socio-environmental determinants of mental health in South Korea^8–10. However, these studies have typically focused on broader mental health outcomes such as depression or generalized anxiety disorder, rather than specifically examining future anxiety. While these studies provide valuable insights into the social and environmental factors affecting mental health, they did not address future anxiety, which can have profound and unique implications for mental well-being. Given the socio-economic volatility in South Korea, this study can fill a significant gap by focusing specifically on future anxiety within the framework of the Social Ecological Model (SEM)¹¹, offering new insights into the micro (individual), mezzo (household or interpersonal), and macro (community or societal) level of factors. This exploration will help deepen our understanding of future anxiety and its determinants.

Furthermore, home ownership plays a crucial role in mental health, with the loss of one’s home often being a deeply traumatic experience in modern society¹². In particular, South Korea has experienced a dramatic real estate bubble since 2017, with apartment prices in Seoul Capital City (SCC) rising by 69% and in Gyeonggi-do Province (GGP) increasing by 34% during this period¹³. The repercussiaons of the real estate crisis in Korea were manifold, including aggressive purchasing of properties, speculation in bitcoin and stocks, and a surge in bankruptcy filings. In addition, it is believed that the economic and housing insecurity prevalent in South Korea is exacerbating the nation’s record-low fertility rate of 0.65 births per year^14,15. Consequently, urgent attention is required to address these pressing issues.

This is the first study to analyze the factors related to future anxiety within the Social Ecological Model (SEM) framework in South Korea. Given that future anxiety is influenced by various levels of factors, including the micro (individual level), mezzo (households), and macro (regional levels), this study employed multilevel analysis. The multilevel analysis is considered an optimal approach to consider the above level differences¹⁶. Many precedent researches related to mental health employed the multilevel analysis¹⁷. Given the lack of sufficient evidence on the social-environmental determinants of future anxiety, we conducted an exploratory study that does not establish specific hypotheses but assumes, based on prior findings, that individual, household, and regional characteristics will have an impact.

Literature review

Anxiety is generally understood to consist of two distinct dimensions: ‘state anxiety’ and ‘trait anxiety’¹⁸. State anxiety refers to individuals’ psychological reactions to adverse events, whereas trait anxiety reflects an inherent personal tendency to experience anxiety¹⁸. Because trait anxiety is related to personality—a fixed characteristic of an individual—clinical treatment or medication is often prioritized over social and environmental interventions¹⁸. As a result, research on anxiety has predominantly been conducted within the field of psychology or treated as an affective disorder, focusing on individual mental health issues. Alternatively, some studies have examined social-environmental factors that trigger anxiety, such as COVID-19 pandemic, or have targeted vulnerable groups, including adolescents, pregnant women, and students^19–24.

Also, several studies have indicated that demographic factors can influence anxiety^25–32. However, most of them have primarily focused on individual factors such as gender, age, and other mental health, while macro-level perspectives—such as social relationships, social capital, community, and social environment—have often been overlooked. Although the SEM has been extensively applied in studies on depression and suicide³³, its use in anxiety research remains extremely limited. Moreover, given the limited research on generalized anxiety, it is reasonable to infer that studies analyzing future anxiety within the SEM framework are even more scarce.

In light of this research gap, and given that future anxiety inherently involves individuals’ expectations of the future shaped by their present circumstances, Bronfenbrenner’s SEM framework¹¹ provides a suitable theoretical lens for examining the multi-level determinants. The SEM conceptualizes health as being influenced by distinct levels: micro (individual), mezzo (interpersonal or household), macro (societal)^34,35, making it particularly appropriate for analyzing complex socio-environmental factors contributing to future anxiety. From this perspective, factors across these hierarchical levels could theoretically affect future anxiety.

For instance, at the micro level, educational attainment can significantly impact an individual’s ability to cope with future uncertainty. Lower levels of education may limit an individual’s capacity to effectively analyze and respond to unfavorable future events, secure stable employment, or navigate complex systems to access necessary resources. At the mezzo level, household income and housing insecurity emerge as key determinants of future anxiety. Households with lower income levels are more vulnerable to economic uncertainties due to their limited ability to absorb financial shocks or unexpected expenses. This financial instability can exacerbate concerns about future unpredictability, increasing anxiety^32,36. Similarly, housing insecurity, such as high rent burdens or the risk of eviction, further undermines a household’s sense of security. These factors amplify anxiety related to future uncertainty, particularly in contexts like South Korea, where housing affordability is a significant societal issue^14,15. Moreover, financial challenges linked to housing, such as unaffordable housing costs, high mortgage payments, or excessive rent, have been identified as major contributors to poor mental health, especially in high-income countries^37–41.

Moreover, the regional and societal factors at the macro level can play a significant role in shaping future anxiety. For instance, economic volatility, such as fluctuations in regional housing prices, can create a collective sense of instability. Zhang and Wang⁴² found that regional economic development correlates with emotional anxiety, while Hankinson⁴³ noted that rising housing prices reduce home ownership rates, which in turn contribute to increased anxiety. Additionally, limited welfare support at the community level can exacerbate these effects by leaving individuals with fewer resources to buffer against socio-economic challenges¹⁷. Collectively, these findings highlight the critical role of macro-level factors in shaping the broader determinants of future anxiety.

Study data and methods

Data sources

The study combined two data sets from different sources to conduct a multilevel analysis based on an ecological model: the Korea Health Panel (KHP) survey data at the individual and household levels, and the regional-level data from the Korean Statistical Information Service (KOSIS). First, the KHP is conducted by the National Health Insurance Services and Korea Institute for Health and Social Affairs. The KHP survey is notable for several reasons. Initially, it ensures representativeness through a two-stage stratified random cluster sampling method based on the Population and Housing Census, encompassing the entire Korean population. Additionally, the KHP employs face-to-face interviews conducted by trained investigators to survey participants. A key strength of the KHP lies in its ability to mitigate information loss and recall bias errors by cross-referencing participants’ self-reported data with medical receipt checks from the administrative data in NHIS. The KHP spans the period from 2008 to 2018. For our research, we used data from 2018, focusing on adult household members aged 18 and over. The sample size consisted of 12,694 individuals nested within 5,762 households.

In addition, KOSIS is the most comprehensive data source released by the Korean National Statistical Office. KOSIS currently provides 343 types of statistics, ranging from domestic to international statistics, produced by 87 organizations in Korea. Additionally, KOSIS publishes 100 major indicators of Korea, which provide information on the country’s economic and social conditions, as well as health indicators such as birth and death rates. Although KOSIS lacks individual-level data and focuses solely on macro-level indicators, it provided sufficient regional-level data for this study. We extracted regional-level data from KOSIS and matched them with the regional variables in the KHP dataset.

The KOSIS dataset consists of 17 administrative districts, including Seoul Capital City (SCC), Busan Metropolitan City (BMC), Daegu Metropolitan City (DG), Incheon Metropolitan City (IMC), Gwangju Metropolitan City (GMC), Daejeon Metropolitan City (DJ), Ulsan Metropolitan City (UMC), Sejong City (SJC), Gyeonggi-do Province (GGP), Gangwon-do Province (GWP), Chungcheongbuk-do Province (CBP), Chungcheongnam-do Province (CNP), Jeollabuk-do Province (JBP), Jeollanam-do Province (JNP), Gyeongsangbuk-do Province (GBP), Gyeongsangnam-do Province (GNP), and Jeju-do Province (JJP). Thus, the 12,694 individuals were nested within these 17 regions. This study received approval from the institutional review board of Yonsei University (approval number: 1041849-202403-SB-055-01).

Outcome variables

The dependent variable in this study was future anxiety. The KHP addressed future anxiety with the following question: “In the past month, have you felt uncertain or anxious about the future?” The initial response categories consist of ‘always’, ‘almost always’, ‘often’, ‘sometimes’, and ‘never’. We re-coded these categories for several reasons. Firstly, only about 5% of respondents answered ‘always’ or ‘almost always’, while 40% reported ‘often’ or ‘sometimes’ (refer to the table S1 in the supplementary material). Secondly, the question about future anxiety pertained to the past month, in contrast to other variables, which referred to the past year, leading to discrepancies in the time frames. Thirdly, the multilevel analysis used in this study requires sufficient sample sizes in each group of the dependent variable. Therefore, we re-categorized these categories as ‘no’ (0) for ‘never’ responses and ‘yes’ (1) for all other responses.

Independent variables

This study used ‘individual level’, ‘household level’, and ‘regional level’ for the independent variables. As for individual level, it contained variables such as gender, age, marital status, subjective health status, Charlson Comorbidity Index (CCI), disabled, oral health condition, depression, psychiatric drug use, and physical and mental stress. While for the household level, the variables were about income quintile, number of household members, house type (Single family house, multi-family house, row-house, rental and studio), and home ownership (Ownership, deposit-based, monthly rent, provided housing). Regarding the regional level, we considered regional economic growth rate, regional dilapidated housing ratio, regional housing vacancy rate, region suicide rate, health and welfare budget share of local government expenditure, and financial autonomy of local governments. The detailed calculations and coding for each variable are presented in Table S2 of the supplementary material.

Analyses

This study used descriptive analysis including frequency, mean, percentage and chi-square for confirming general characteristics of the samples. For the major analysis, this study employed three-level multilevel logistic regression analyses to examine the determinant factors of future anxiety. The study sample consists of 12,694 individuals as level 1, nested within 5762 households as level 2, which are further nested within 17 regions as level 3.

However, the 17 regions at level 3 do not provide a sufficient sample size for multilevel analysis. This is because the Maximum Likelihood (ML) estimation method in multilevel analysis is susceptible to small sample sizes, which can lead to biased estimation. While the recommended adequate sample sizes vary across studies, most suggest a minimum of 30 to 50^44–46.

Therefore, given the limited number of regions (n = 17), we opted for Bayesian Markov Chain Monte Carlo (MCMC) estimation methods instead of maximum likelihood (ML) methods. The Bayesian MCMC estimation is the latest statistical method as known for useful models with small sample sizes in multilevel analysis^47–49. Recent studies revealed that the Bayesian MCMC obtained unbiased estimates than likelihood methods especially with fewer than 10 clusters^50,51.

This study proceeded with a series of five models. We introduced the initial null model (Model 1), consisting solely of an intercept, to establish a baseline for assessing the variability in future anxiety across the three levels. Models 2 to 4 sequentially incorporated independent variables at different levels into the null model. Model 2 introduced individual-level independent variables, Model 3 added household-level factors, and Model 4 included regional-level factors alongside the baseline Model 1. The final model (Model 5) integrated all levels of factors simultaneously. All statistical analyses were performed using Stata/MP version 18.0 (Stata Corp LP, College Station, TX, USA). For multilevel modeling with maximum likelihood estimation for model fit, the MCMCglmm command was employed.

Study results

Descriptive analyses

Individual level of characteristics

Table 1 shows the individual characteristics of future anxiety. Out of a total of 12,694 individuals, 6,146 (48.6%) reported that they have experienced future anxiety. Across different age groups, younger individuals exhibited higher levels of future anxiety compared to older age groups. Specifically, those aged 29 and under had the highest prevalence at 57.17%, while those aged 76 or older had the lowest at 39.74%. Regarding marital status, single individuals (57.8%) reported the highest levels of anxiety about their future compared to married individuals (45.77%). Similarly, individuals with poor subjective health reported higher future anxiety rates (57.92%) compared to those with good subjective health (43.74%). Concerning the CCI, individuals scoring 2 or more had the highest future anxiety rates (50.45%), whereas those who scored 0 had the lowest (45.03%). Moreover, individuals with poor oral health and have experienced physical & mental stress were more likely to report future anxiety compared to those with better conditions. Likewise, individuals with depression or prescribed psychiatric medication were more likely to report future anxiety.

Table 1.

Individual-level characteristics of the samples (level 1).

	Future anxiety		Χ2 (P-value)
	No	Yes
	n (%)	n (%)
Gender
Male	2,912 (50.73)	2,828 (49.27)	1.885 (0.17)
Female	3,613 (51.96)	3,341 (48.04)
Age group (years)
≤29	711 (42.83)	949 (57.17)	205.68 (0.000)
30–39	717 (48.51)	761 (51.49)
40–49	1,066 (44.64)	1,322 (55.36)
50–65	1,913 (52.51)	1,730 (47.49)
66–75	1,228 (59.96)	820 (40.04)
≥ 76	890 (60.26)	587 (39.74)
Marital status
Married	4,607 (54.23)	3,889 (45.77)	107.54 (0.000)
Single	985 (42.20)	1,349 (57.8)
Separated	933 (50.05)	931 (49.95)
Subjective health status
Good	2,796 (56.26)	2,174 (43.74)	118.46 (0.000)
Mild	2,890 (50.44)	2,840 (49.56)
Bad	839 (42.08)	1,155 (57.92)
Charlson Comorbidity Index
0	4,315 (51.05)	4,138 (48.95)	13.68 (0.001)
1	1,101 (54.97)	902 (45.03)
≥ 2	1,109 (49.55)	1,129 (50.45)
Disabled
No	6,056 (51.34)	5,740 (48.66)	0.263 (0.608)
Yes	469 (52.23)	429 (47.77)
Oral health condition
Good	3,765 (53.56)	3,264 (46.44)	55.99 (0.000)
Mild	1,585 (51.88)	1,470 (48.12)
Bad	1,175 (45.02)	1,435 (54.98)
Depression
No	6,392 (53.22)	5,619 (46.78)	294.59 (0.000)
Yes	133 (19.47)	550 (80.53)
Psychiatric drug
No	6,211 (51.95)	5,744 (48.05)	24.95 (0.000)
Yes	314 (42.49)	425 (57.51)
Physical & mental stress
Good	4,727 (71.01)	1,930 (28.99)	2.2e + 03 (0.000)
Mild	1,682 (31.17)	3,715 (68.83)
Bad	116 (18.13)	524 (81.88)
Total	6,525 (51.4)	6,169 (48.6)

Household level of characteristics

Table 2 displays household characteristics associated with future anxiety. Of the 5,762 households, 2,783 (48.3%) reported at least one experienced future anxiety. Regarding the household income quintile, the wealthiest households exhibited the lowest prevalence of future anxiety (43.52%), with negligible differences observed among other income groups. In terms of home ownership, individuals who resided in provided housing reported the least future anxiety (43.76%), whereas those in monthly rental houses reported the highest (62.54%). Concerning the household size, those with two members had the lowest prevalence of future anxiety (39.89%). Lastly, concerning housing types, individuals who lived in studios reported the highest future anxiety (56.44%), while those in single-family houses reported the lowest (44.99%).

Table 2.

Household-level characteristics of the samples (level 2).

	Future anxiety		Χ2 (P-value)
	No	Yes
	n (%)	n (%)
Household income quintile
1 (lowest)	617 (51.33)	585 (48.67)	15.44 (0.004)
2	594 (50.38)	585 (49.62)
3	542 (48.7)	571 (51.3)
4	577 (51.56)	542 (48.44)
5 (richest)	649 (56.48)	500 (43.52)
Home ownership
Ownership	2,198 (54.77)	1,815 (45.23)	94.26 (0.000)
Deposit-based	222 (42.21)	304 (57.79)
Monthly rent	257 (37.46)	429 (62.54)
Provided housing	302 (56.24)	235 (43.76)
Number of household’s members
1	568 (49.43)	581 (50.57)	83.26 (0.000)
2	1,118 (60.11)	742 (39.89)
3	483 (49.39)	495 (50.61)
>4	810 (45.63)	965 (54.37)
Types of house
Single family house	1,141 (55.01)	933 (44.99)	18.5 (0.001)
Multi-family house	346 (48.19)	372 (51.81)
Row-house	91 (52.3)	83 (47.7)
Rental	1,330 (50.51)	1,303 (49.49)
Studio	71 (43.56)	92 (56.44)
Total	2,979 (51.7)	2,783 (48.3)

Regional level of characteristics

Table 3 shows the descriptive statistics of the regional variables. Both SCC and CNP exhibited the highest anxiety rates in Korea, at 0.56% each. For regional economic growth rate, CBP had shown the highest rate with 6.3%, while UMC showed the lowest with − 2.2%. CNP also showed the highest regional suicide rate with 35.5%, while SCC showed the lowest with 22.5%. For the regional dilapidated housing ratio, JNP had the highest with 33.7, while SJC had the lowest with 6.7. As for the regional housing vacancy rate, JJP appeared to be the highest with 14%, with the lowest being 3.2% for SCC. The regional suicide rate was the highest in JJP (30.6%), while SCC (22.5%) was discovered to be the lowest. When it comes to the health and welfare budget share of local government expenditure, DMC had the highest with 44.7%, with the lowest being JJP with 21.7%. Lastly, for the local government fiscal independence ratio, the highest was 79.2% for SCC, while the lowest was 19.8% for JNP.

Table 3.

Regional-level characteristics of the samples (level 3).

	17 administrative districts
	SCC	BMC	DG	IMC	GMC	DJ	UMC	SC	GGP	GWP	CBP	CNP	JBP	JNP	GBP	GNP	JP
Sample size (n)	1,565	1,024	821	735	382	519	322	22	2,517	439	481	576	646	660	698	927	360
Anxiety rate (%)	0.56	0.29	0.53	0.51	0.35	0.49	0.46	0.41	0.53	0.49	0.50	0.56	0.46	0.35	0.43	0.48	0.53
Regional economic growth rate (%)*	3.6	1.7	2.4	0.7	5	0.9	-2.2	2.8	6	1.9	6.3	0.6	1.7	2.2	-1.2	0.6	-0.9
Regional dilapidated housing ratio	17.6	22.8	17	13.5	15.8	15.2	12.2	6.7	8.5	23.4	20.5	19.9	25.8	33.7	26.9	20.7	20.5
Regional housing vacancy rate (%)	3.2	8.1	5.2	6.5	7.2	6.1	7.7	12	6	12	12.3	12.7	12.2	15.3	12.9	10.7	14
Region Suicide Rate (%)	22.5	27.9	26.8	27.9	25.7	28.3	27.1	26	24.2	33.1	31.1	35.5	29.7	28	29.6	28.9	30.6
Health and welfare budget share of local government expenditure (%)	39.6	44.5	44.6	42.4	46.6	44.7	33.9	24.6	34.9	26.8	29.9	27.6	30.6	25.4	26.3	30.3	21.7
Local government fiscal independence ratio (%)	79.2	52.3	47.6	60.3	43.8	47.1	59.9	57.1	61.9	23.5	29.6	34.5	21.5	19.8	26.2	37.7	34.5

Note: SCC: Seoul Capital City; BMC: Busan Metropolitan City; DG: Daegu Metropolitan City; IMC: Incheon Metropolitan City; GMC: Gwangju Metropolitan City; DJ: Daejeon Metropolitan City; UMC: Ulsan Metropolitan City; SC: Sejong City; GGP: Gyeonggi-do Province; GWP: Gangwon-do Province; CBP: Chungcheongbuk-do Province; CNP: Chungcheongnam-do Province; JBP: Jeollabuk-do Province; JNP: Jeollanam-do Province; GBP: Gyeongsangbuk-do Province; GNP: Gyeongsangnam-do Province; JP: Jeju-do Province; *We added the 3.2 for log-transformation for treating minus values when conducting multi-level logistic regression.

Multilevel logistic regression analysis

This study consists of 5 models (model 1, 2, 3, 4, 5), and Table 4 shows the results of multilevel logistic regression analysis for future anxiety. Model 1 contained the model without any explanatory variables. Model 2 only contained the variables at the individual level, while Model 3 contained the variables in household level, and Model 4 contained the variables at the regional level.

Table 4.

Multilevel logistic regression results for future anxiety.

Variables	Categories	Future anxiety
		Model 1	Model 2	Model 3	Model 4	Model 5
Gender (Male)	Female		0.791***			0.781***
Generation (≤29)	30–39		0.863			0.919
	40–49		0.964			1.024
	50–65		0.535***			0.664**
	66–75		0.25***			0.311***
	≥ 76		0.161***			0.196***
Marital status (Married)	Single		1.78***			1.966***
	Separated		1.599***			1.349**
Subjective health status (Good)	Mild		1.589***			1.572***
	Bad		2.043***			2.011***
Charlson Comorbidity Index (0)	1		0.917			0.93
	≥ 2		0.977			0.963
Disabled (No)	Yes		0.869			0.835
Oral problem (No)	Mild		1.359***			1.333**
	Severe		1.743***			1.69***
Depression (No)	Yes		3.037***			3.065***
Psychiatric drug (No)	Yes		1.114			1.104
Physical mental stress (No)	Sometimes		6.819***			6.95***
	Often		11.744***			11.924***
Frustrating event (No)	Yes		12.262***			12.013***
Household income quintile (Richest)	1 (Poor)			2.693***		3.066***
	2			2.461***		2.52***
	3			1.983***		2.068***
	4			1.537***		1.666***
Home-ownership (Home owners)	Rental for a year unit			1.736***		1.298*
	Monthly rental			2.751***		1.328*
	Provided housing			0.811		0.722**
No. of household member (< 4)	1			0.573***		0.673**
	2			0.322***		0.515***
	3			0.856		0.85
Types of house (Single family house)	Multi-family house			1.48**		1.26
	Row-house			1.105		1.214
	Rental			1.167		1.15
	Studio			2.12**		1.728**
Regional economic growth rate (log)				0.996	1.009
Regional dilapidated housing ratio (log)				0.685*	0.84
Local housing vacancy rate (log)				0.231***	0.358***
Region Suicide Rate (log)				1.909**	1.711**
Health and welfare budget share of local government expenditure (log)				0.51***	0.559***
Local Government Fiscal Autonomy (log)				0.707	0.923
Constants		0.795***	0.18***	0.533**	0.871	0.106***
Random effects
Intra-class correlation
Regional level		0.014	0.009	0.011	0.002	0.004
Region > Household level		0.14	0.139	0.139	0.151	0.142

Note: MCMC: 55,000 iterations; burn-in = 5,000; thinning 20 = eff. Sample size ~ 2,500 per chain; OR = adjusted odds ratio of posterior mean, ICC = intraclass-correlation coeffient; The acceptance ratios for liability set in all models were between 0.2 to 0.5; pMCMC = Bayesian p-value; *: <0.1 **: <0.05 ***: <0.01.

The ICC (intra-class correlation coefficient) in the null model suggests the model’s power on the baseline. for the regional level presenting 0.014 means that the regional cluster has an impact of 1.4% on individual mental health and future anxiety. As for the household level, it revealed that household cluster power has an effect of 14% on the individuals’ future anxiety.

In model 2, female had lower future anxiety than men (odds ratio = 0.791, p < 0.001). As for the generation, the older the person, the lower the likelihood of experiencing future anxiety. For marital status, people who were either single (odds ratio = 1.78, p < 0.001) or separated (odds ratio = 1.599, p < 0.001) had a significantly higher future anxiety than those who were married. Also, for subjective health status, the future anxiety levels rose, with bad (odds ratio = 2.043, p < 0.001) being the worst. Similarly, as the subjects’ oral problems got worse, they showed higher future anxiety at a statistically significant level. People who had depression (odds ratio = 3.037, p < 0.001) showed future anxiety compared to people with no depression at a statistically significant level. Future anxiety increased as people’s physical and mental stress deteriorated at an extraordinarily significant level, from sometimes (odds ratio = 6.819, p < 0.001) to often (odds ratio = 11.744, p < 0.001) compared to no. Lastly, people who suffered from a frustrating event had the highest future anxiety than any other category with 12.262 odds ratio (p < 0.001).

In model 3, all the variables were statistically significant for household income quintile, with the odds ratio getting higher as the subjects got poorer. People who lived in rental homes for a year unit (odds ratio = 1.736, p < 0.001) showed future anxiety at a significant level, while people who lived in monthly rental houses (odds ratio = 2.751, p < 0.001) showed the highest. As for the number of household members, people who lived alone or lived in a household of 2 showed more future anxiety at a statistically significant level than those with 4 household members. Lastly when it comes to the types of houses, people who lived in a multi-family house (odds ratio = 1.48, p < 0.05) and commercial building (odds ratio = 2.12, p < 0.05) showed statistically significant higher future anxiety level. In model 4, the regional dilapidated housing ratio showed a statistically significant level of 0.685 odds (p < 0.1). Local housing vacancy rate (odds ratio = 0.231, p < 0.001) showed a significant future anxiety level, as well as the regional suicide rate (odds ratio = 1.909, p < 0.05). As for the health and welfare budget share of local government expenditure, it showed a statistically significant future anxiety level of 0.51 odds ratio (p < 0.001).

In the final model (Model 5), all the variables including individual, household and regional levels were considered. Almost all the variables showed the same statistical significance level as models 2–4, except for a few. The significance level that changed from under 0.001, to under 0.05 in model 5 are the following: generation for 50–65, marital status for separated, mild oral problems, and 1 member in a household. The home-ownership variables all had a change in the significance level, with rental for a year unit and monthly rentals becoming under 0.1 from 0.001. In model 3 provided housing turned out to be not statistically significant, but in model 5, it turned out statistically significant, with the odds ratio being 0.722 (p < 0.0.5). Lastly, the multi-family house in model 3 turned out to be not statistically significant in model 5, as well as the regional dilapidated housing ratio.

Discussions

This study analyzed the determinant factors of future anxiety in the framework of ecological models by using a three-level multilevel analysis. The study considered predisposing, social, physical & mental health as individual factors (micro level), income quintile, home ownership, number of household members, and type of house as household level (mezzo level), regional economic growth rate, dilapidated housing ratio, housing vacancy rate, suicide rate, health and welfare budget share, fiscal independence ratio as regional level (macro level).

As a result, first of all, the ICC presented that the household level was as substantial as 14% in the null model, which can be interpreted as the household factors determining the 14% of individuals’ future anxiety. Meanwhile, the regional level was quite low as 1.4% in the null model, which had limited effect on the individuals’ experiences. This low value indicates that it may not be the best geographical unit for estimating individual future anxiety and measuring regional variables¹⁷. Han and Lee¹⁷ explained that the low value of ICC at a higher level might be attributed to the large administrative areas, each encompassing only a few districts. The small number of regions, combined with an excessive number of individual samples nested within the same region, introduces significant heterogeneity in the samples. Nevertheless, it is perhaps natural for the impact of factors to diminish as they extend to higher levels, radiating outward from the individual⁵².

For instance, factors examined at the micro level, like residential characteristics, are thought to have a more immediate impact on individual outcomes compared to macro-level factors, such as regional economic growth⁵³. In the results of individual level, factors showed that the females were less likely to experience future anxiety, which was completely contrary to the traditional results of depression. Most studies reported that females are more vulnerable to mental health, especially for depression than male. While there is limited evidence on the determining factors of future anxiety, we contend that Korean males are more prone to worry about their future lives due to the enduring patriarchal nature of Korean society⁵⁴. This heightened anxiety often stems from a greater sense of responsibility for the economy. Even single young males are mostly concerned about the future such as, marriage expenses, and savings. In particular, in the Korean marriage market, there still exists a common belief that a man must own a house. A man without property is often perceived as deficient, which can lead to future anxiety.

In terms of age, the older, the lower the likelihood of future anxiety. This result was correspondent to the existing study⁵⁵. Susulowska⁵⁵ found that the fear of the future first appears at age 11 ~ 14, increases thereafter, and highest frequency at age 20 ~ 29, and decreases as older. Furthermore, the results in marital status were also consistent with those of previous studies⁵⁶, in which unmarried separated, divorced, or widowed are more likely to recognize future anxiety than those married. The most significant factors contributing to future anxiety include health-related issues, particularly depression, physical and mental stress, as well as experiences of frustration. Early examinations have already revealed a high correlation between depression and anxiety^29–31.

Moreover, existing studies further elucidate the distinctions between anxiety and depression, indicating that anxiety is associated with dangerous events and future threat, while depression is linked to past experiences of loss⁵⁷. With this knowledge, the relatively low correlation could be explained by that this study focuses on future anxiety. Instead, it was the combination of physical-mental stress and experiences of frustrating events that exhibited an extraordinarily significant effect. While Borkovec, Metzger, and Pruzinsky⁵⁸ argued that individuals tend to worry most about the possibility of non-rewarding outcomes from their efforts, our findings revealed unexpectedly high odds ratios.

The most important findings in this study are at the household level which have not explored in the previous studies. All the household-level variables had significant effects on future anxiety. The lower the income, the more likely to experience future anxiety. This indicates that not only the householder, but the most of household members are highly likely to worry about the various economic circumstances, such as shortage of living equipment, food costs, education expenses, health expenses, and so on.

Regarding home ownership, renters experience significantly more future anxiety than homeowners. But without owning a house, individuals do not worry significantly about the future if they live in provided housing that is free of charge. These effects are noteworthy in that they remained significant even after controlling for individual-level mental anxiety. This suggests that rental residents experience anxiety about the future due to the prospect of having to relocate when their housing contracts expire.

In Korea, anxiety about obtaining housing has surged since 2017 due to soaring real estate prices and overcrowding in Seoul⁵⁹. This has escalated to a point in which purchasing a home, regardless of one’s income or savings accumulated over the years, has become exceedingly difficult. The prospect of being unable to own a home despite diligent saving has led to feelings of frustration for many⁶⁰. While some individuals have resorted to taking out loans to purchase homes, they later experienced significant financial strain due to the high-interest rates associated with these loans⁶¹.

Meanwhile, individuals living in one to two-person households have shown lower levels of future anxiety compared to those in four-person households. In Korea, the majority of four-person households consist of parent-child or parent-grandparent arrangements, often influenced by the increased economic and caregiving responsibilities placed on the household heads. Consequently, members of these four-person households may harbor concerns about their livelihoods once the household heads retire from the workforce. This pervasive sense of future anxiety extends beyond these households, impacting surrounding single or two-person households and contributing to a broader atmosphere of social unease. Recently, this trend has been identified as one of the contributing factors to the unprecedented phenomenon of a total fertility rate of 0.78⁶².

Moreover, house type also had an effect on future anxiety. Especially for those living in studios were significantly more likely to express future anxiety. At the time of 2017, most studios were temporal houses where students for the certification exam candidates such as the Civil service, Bar, or CPA exam stayed, and it is interpreted that there was great anxiety about the future because these were people preparing for urgent future exams. Also, most residential studios in Korea are structurally and environmentally poor for living. Even without exam preparation, many individuals may find themselves desiring to escape from such challenging environments. However, if circumstances, whether economic or social, prevent them from doing so, there may be feelings of anxiety about continuing to live in the studio in the future.

The results in the factors of the regional level are also meaningful, though, its’ ICC value was too low that it could only be used as a reference. Despite its limited contribution to interpreting the results, it seems that a higher local housing vacancy rate is associated with a lower level of future anxiety among individuals. In South Korea, the housing vacancy rate is usually low in metropolitan cities, especially for the SCC, but high in the provinces. In particular, most SCC citizens experienced skyrocketing real estate, feeling literally “fear and anxiety”¹³. Hence, when the vacancy rate in an area is low, it is likely that people will experience anxiety about the future, as they perceive having little chance to purchase houses.

Meanwhile, the regional suicide rate was presented as significantly affecting the future anxiety of individuals. When hearing news of a neighbor or fellow resident in the area committing suicide, it often triggers worry. As the higher regional suicide rate indicates an increased occurrence of such news, fears and future anxieties would proliferate. Therefore, the mood of the regional society should be carefully administered. In this regard, the other results, “health and welfare budget share” could suggest a solution. The result showed that the higher the budget share on health and welfare, the lower the likelihood of future anxiety. Indeed, the impact will vary depending on the allocation and utilization of the budget. However, the emphasis placed by local governments on allocating a higher proportion of the welfare budget underscores their commitment to the well-being of residents. It seems that it is essential to allocate budget resources at a reasonable level.

A notable limitation of this study is that it does not specify the exact nature of future anxiety being analyzed. Future anxiety can stem from various sources, such as economic uncertainty, health concerns, or security issues. While the study’s findings regarding economic and housing-related factors are significant, the lack of specificity about the source of future anxiety limits the interpretability of the results. Future research would benefit from distinguishing between different types of future anxiety to provide more targeted insights. In addition, multilevel analysis is a method for separating each level; therefore, this study could not provide results or discuss the interaction effects between levels (e.g., micro-mezzo, mezzo-macro).

Conclusion

This study provides a comprehensive analysis of factors contributing to future anxiety among individuals in South Korea, using a three-level multilevel analysis framework that includes individual (micro), household (mezzo), and regional (macro) factors. Our findings indicate that future anxiety is significantly influenced by a myriad of factors across these levels, highlighting the complex interplay between individual, household, and regional elements. At the individual level, factors such as gender, age, marital status, health status, and experiences of depression and stress play a pivotal role in shaping perceptions of future anxiety. Household factors, including income quintile, home ownership, and household composition, also emerge as critical determinants, suggesting that economic stability and living conditions within one’s immediate environment are key to understanding variations in future anxiety. Moreover, regional factors such as the housing vacancy rate and suicide rates underscore the impact of broader socio-economic and cultural contexts. Our analysis underscores the need for a holistic approach to addressing future anxiety, one that considers interventions at multiple levels of influence. As South Korea suffers from socio-economic challenges, including a real estate crisis and demographic shifts, our study calls for targeted policies that address the root causes of future anxiety, promote mental well-being, and foster a more equitable and supportive society.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

HWJ: design of the work, formal analysis, drafting the article; MC: review and revise. KSL: supervision, review, and revise.

Funding

This study was funded through a grant from the Korea Health Technology R&D Project, supported by the Korea Health Industry Development Institute (KHIDI) and the Ministry of Health & Welfare, Republic of Korea (Grant No. RS-2024-00438829).

Data availability

We obtained raw data from KHP (https://www.khp.re.kr:444/eng/main.do) which is available to the public with authorization. One can get permission to use data from the Korea Institute for Health and Social Affairs (KIHASA) and the National Health Insurance Services (NHIS).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

We conducted this study following the Declaration of Helsinki, and the study protocol was approved by the Yonsei University Mirae Institutional Review Board (approval IRB number: 1041849-202403-SB-055-01). The need for written informed consent was waived by the Institutional Review Board ethics committee due to the retrospective nature of the study.