Introduction

The Covid-19 pandemic impacted global health on an unprecedented scale, and had a devastating effect on the global economy and living conditions. In the absence of vaccines, countries implemented various non-pharmaceutical interventions to curb the spread, amongst which were economic lockdowns, social distancing, and various hygiene specifications (United Nations, 2020). Conversely, Covid-19 was responsible for the most significant acceleration of the digital revolution, and fundamentally changed the way in which we live, work, socialise, and participate in leisure (International Telecommunication Union 2020). United Nations Secretary-General Antonio Guterres said that the internet saved lives because people could work, study, and socialise safely online (Szymański & Rathman, 2021).

Covid-19’s impact on South Africa was no different. South Africa declared a state of disaster under the Disaster Management Act (57/2002) (South African Government, 2003) on 15 March 2020 and entered one of the most stringent global lockdowns. Citizens were forced to stay home, and were allowed out only for medical care and essential goods and services (Harding, 2020). Shortly thereafter South Africa’s Risk Adjusted Strategy for dealing with Covid-19 was introduced, allowing decision-makers to determine the required lockdown levels versus the disease risk. This resulted in a five-level alert system ranging from Level 5 (a high-volume of Covid-19 spread, causing very serious damage) to Level 1 (a low-volume of Covid-19 spread causing minimal damage but involving high-level readiness from the health sector). Other non-pharmaceutical interventions included the wearing of facemasks, handwashing with soap and water or using alcohol-based hand-sanitizer, and social distancing. The government thus attempted to strike a balance between the preservation of social stability and economic relief (Makokoane, 2021). Unfortunately, Covid-19 exposed structural weaknesses in the South African economy and disproportionately impacted poor, vulnerable households, and the youth of society, as they absorbed most of the initial shock and were pushed into food insecurity. South Africa’s economy contracted by roughly 7% in 2020, wages decreasing by 10–15%, and the number of unemployed increased by approximately 1.5 million, thus exacerbating severe inequality in society (Arndt & Robinson, 2020; World Bank Group, 2021).

Conversely, Covid-19 sparked one of the biggest transformations of the South African economy in recent years as geographical flexibility in workspaces towards decentralised locations became the norm. Businesses were forced to transform their operating models and their expectations of employees, as people could now work how, where, and when they wanted within a work-from-home and hybrid-work model (BizzCommunity, 2021). Significant improvements in smart office technology through improved bandwidth, faster internet speeds, Wi-Fi hotspots, cloud-based storage solutions, smart phone usage as computers, and online video conferencing platforms such as Zoom, Microsoft Teams, and Google Meet provided infinite possibilities for remote working (De et al., 2020; Szymański & Rathman, 2021). The so called ‘Zoom-boom’ enabled a ‘Franchise of flexible office space’ or a ‘hub-and-spoke’ office model to emerge in the suburbs (BizzCommunity, 2021). The ‘hub-and-spoke’ office model involves a larger centralised ‘hub’ where employees congregate, while smaller decentralised satellite offices create ‘spokes of productivity’, taking cognisance of employee wellness (Robinson, 2021).

Geographical flexibility in work is reshaping South Africa’s urban centres as more people can migrate from cities to smaller towns or suburbs; a trend described as ‘semigration’, i.e., the act of moving from one location to another within one’s home country (Creamer Media, 2020). The term was developed in the 1990s to describe migration from Johannesburg to Cape Town (Ballard, 2004), and migration to coastal and estate living (BizzCommunity, 2021). More recently semigration refers to an increased pattern of decentralised urban living, moving away from the larger metropoles towards smaller towns, subrubs and peripheral areas within metropolitan areas, or secondary cities and towns in relative proximity to metropolitan areas. Semigration is now driven by the desire to improve living conditions, while participating in work-from-home or hybrid work. It is no longer just the wealthy or retired community who semigrate, but people of all ages, married, single, with or without children, and a wider variety of wealth segments (BizzCommunity, 2021).

It should however be noted that, to some extent, the more recent semigration phenomenon still resembles the somewhat similar patterns Ballard described in 2004: i.e., that semigration is affected by white people in particular, as a way to withdraw from democratic society which is attempting to achieve assimilation and integration between various racial and income groups (Ballard, 2004). In that sense, semigration is similar to emigration, but without leaving the actual borders of the country. Instead, in this regard it is a spatial practice to achieve ‘sanitised spaces’ (or ‘privatised enclaves’) reserved for a select homogeneous few. Stated differently, Ballard (2004: 60) summarises the work of Hook and Vrdoljak (2002: 202) by stating that in its extreme, ‘semigration is the creation of a “self-contained town” from which residents seldom need to venture.’ Henama (2021) more recently described semigration as a unique form of post-apartheid labour migration, confirming the findings of Ballard (2004). He describes the typical semigrant as being wealthy and highly skilled–and usually white. Semigration locations become the primary residence for the entire family, with adults being much younger than retirement age. Significant investments are made in fibre internet and Wi-Fi services to increase the speed and reliability of communication and limit the need to travel to urban centres for work. Despite this, semigrants are ‘super-commuters’–they are highly mobile and have the ability to commute via private transport or by air travel if the need arises. The work by Ballard (2004) and Henama (2021) implies that semigration is a relatively upmarket trend, which remains unattainable for low-income marginalised communities in South Africa. Media reports during the Covid-19 pandemic confirm these semigration characteristics (BizzCommunity, 2023; Cohen, 2022; Erasmus 2022; Hoek, 2022; Leadhome Properties, 2021; Matloha, 2022; Meintjies, 2023; Smit, 2023).

According to Lightstone, the Western Cape and KwaZulu-Natal have seen the most significant increases in semigration patterns due to the Covid-19 pandemic. These provinces offer greater environmental attractions and coastal locations, better economic potential, efficient infrastructure, services and facilities, flexible work and lifestyle opportunities, an improved sense of safety, holiday ambience, and its retirement potential offers great returns on investments (Lightstone, 2021; The Newspaper, 2022). Semigration is mostly intra-provincial; approximately 44% of Gauteng residents, 26–30% of Western Cape residents, and 11–12% of KwaZulu-Natal residents semigrated to smaller towns during the Covid-19 period. The ratio between semigration to small towns versus major cities and metropolitan areas is 60:40 in the Western Cape and KwaZulu-Natal, while the ratio is even higher in Gauteng (80:20). Semigrants to the Western Cape are mostly 36–49 years old and purchase mid-value to high-value freehold properties. Generally, semigrants purchased properties of greater value and smaller size, but those who purchased larger homes probably work from home. Popular semigration towns in the Western Cape include Hermanus, Langebaan, St Helena Bay, Groot Brakriver, Plettenberg Bay, and Knysna (Lightstone, 2021; PropertyWheel, 2021).

A distinct gap has been identified in the academic literature regarding the application of the concept of semigration to case-study areas during the Covid-19 pandemic. Existing sources have been written predominantly by journalists–and numerous opinion pieces by real estate agents, though quite popular, do not contextualise the semigration phenomenon within applicable theories (BizzCommunity, 2023; Cohen, 2022; Erasmus 2022; Hoek, 2022; Leadhome Properties, 2021; Matloha, 2022; Meintjies, 2023; Smit, 2023). This article therefore attempts to fill this gap by exploring changes that have occurred in the property market of a case-study area (Hermanus) that has been identified as a key semigration town by the media. The aim of this study is to investigate the extent and characteristics of semigration to Hermanus.

The study thus seeks to answer certain research questions, such as:

  1. 1.

    How the number of property purchases and their spatial distribution has changed over time in Hermanus?

  2. 2.

    How the property prices vary across both space (i.e., suburbs) and time in Hermanus?

  3. 3.

    What the typical characteristics of semigrants to Hermanus are?

  4. 4.

    What the potential policy implications of semigration to Hermanus might be?

Literature review

In the absence of a distinct academic literature base that applies the concept of semigration to specific case studies, it was decided to approach this literature review from a more theoretical perspective, contextualising the topic within relevant and applicable theories that could explain this phenomenon within the study area of Hermanus. The theories selected include: 1) the synthetic model of migration–providing a summary of the various factors influencing why people migrate; 2) productionism and environmentalism–describing divergent spatial distribution patterns of development associated with changing socio-economic circumstances and societal mobility patterns; and 3) life-course theory and residential location theories–explaining how people choose to live their lives within changing socio-historical contexts, which ultimately influence their residential choices within varying geographical locations.

The synthetic model of migration

Various theories explain why people migrate, but the synthetic migration model combines key factors from different models, including spatial reward structure, individual characteristics and structural variables, and information sources, flows and filters (obstacles and facilitators). The spatial reward structure refers to push-and-pull factors, which create a spatial disequilibrium between the origin and destination location, (both have positive and negative externalities, respectively attracting and repelling migration) influenced by objective circumstances and subjective evaluations (Gelderblom, 2006).

Individual characteristics comprise age, gender, education, employment, and occupation, while structural variables refer to household composition, e.g., each household member weighs the pros and cons of migrating to ensure optimum living conditions for everyone (Jacobs & Du Plessis, 2016). The decision to migrate is about minimising the individual’s or household’s risks and maximising overall economic welfare (Gelderblom, 2006). It is also influenced by pre-existing values associated with income and wealth (higher income, higher standards of living, economic stability, and crime-free environments), comfort (pleasant, environmentally healthy, and socially amenable communities), stimulation (access to people, entertainment, and educational opportunities), and affiliation (access to family and friends when in need) (De Jong, 2000). Migration disrupts pre-existing routines causing discomfort and insecurity. The destination location should thus offer sufficient resources to build a new routine and sense of belonging within an unfamiliar environment (Gelderblom, 2006).

Contemporary international migration patterns are embedded in the dynamics of interconnected information and communication flows, which form a societal informational network, improving the interaction between people and places within the global economy (Ros et al., 2007). Improvements in information and communication technologies (ICTs) allow people and places to interact within the global economy (Gelderblom, 2006), and friends and family can facilitate migration through generalised information about the destination location, accommodation options, and economic opportunities (Jacobs & Du Plessis, 2016).

Productionism and environmentalism

The theories of productionism and environmentalism also describe the migration process. The decision to migrate is underscored by how well people can improve their living conditions and satisfy their needs; how to achieve this is guided by two opposing belief systems and principles: production-driven and environment-driven migration (Gutiérrez & Munoz, 2021). Hart (1983) coined these terms by describing divergent spatial distribution patterns of development associated with the shift from a Fordist economy (mass production and consumption) to a post-Fordist economy (industrial economy emphasising labour specialisation and suburban business decentralisation) (Geyer & Geyer, 2017).

Productionism is the lifecycle stage where people prioritise improvements in employment, income, and upward social mobility (Jacobs & Du Plessis, 2016). Increased economic production, growth, and consumption is thus better for society, as needs are satisfied and social order is maintained (Gutiérrez & Munoz, 2021). Work (paid employment) is a distinct domain of people’s lives providing a sense of worth and social value (Giddens, 1994), resulting in a trade-off between the positive externalities associated with increased socio-economic productivity, versus the negative externalities created by deteriorating ecological systems as natural systems are discounted (Gutiérrez & Munoz, 2021).

Environmentalism proposes a fundamental rethinking of productive activity based on the principles of biophilia, sustainability, ecology, restoration, regeneration, and the protection of the natural environment. In this system, a mutualistic relationship between economic activity and nature is required to satisfy human needs (Gutiérrez & Munoz, 2021). Productionism manifests through urbanisation to the largest cities as people migrate in search of economic opportunities. Conversely, at a certain tipping point, there comes a deconcentration and decentralisation towards the periphery of cities and to smaller towns in the rural hinterland, as people migrate in search chiefly of improved living conditions–which is effectively environmentalism (Jacobs & Du Plessis, 2016).

Life course theory and residential location theories

The decision to migrate is also influenced by life course theory, or the life course perspective, and residential location theories. Life course theory was pioneered in the late 1920s and popularised by Glen Elder in the 1960s. According to life course theory, people’s lives are shaped by structural age stratification and trajectories such as work, career, and family pathways, socio-developmental lifespan psychology, and cultural and intergenerational contexts (Elder Jr 1994).

Life course theory is characterised by several principles, including socio-historical and geographical location, the different time-spans of people’s lives, linked or independent lives, and human agency in decision-making. Socio-historical and geographic location focusses on how people live their lives through changing historical times and across varying geographical locations and contexts (Elder Jr et al. 2003; JRank, 2022). According to Elder, people’s lives comprise three main time-spans, including individual (their chronological age), generational (age groups comprising heterogeneous people from different genders, social classes, family structures, ethnicities, and religions), and historical (societal events influencing people). The transition and trajectories of time are also important. Transitions involve changes in status, social identity, and role involvement–while trajectories are longer-term age-based pathways of development within major social institutions, such as education, family, or career (JRank, 2022). ‘Linked or independent lives’ categorises the extent to which people’s lives become embedded in social relationships with friends, family and colleagues across a person’s lifespan–and to what degree this determines socialisation patterns, behavioural exchange, and generational succession (Elder Jr 1994). ‘Human agency in decision-making’ refers to people’s ability to construct their own life course through choices and actions (Elder Jr et al. 2003), associated with their institutional involvement and social relationships, bounded by the opportunities and constraints presented by various life events and experiences (JRank, 2022).

The decision to migrate is a major turning point in people’s life courses. People migrate when structural forces (the economy, politics, organisations, and ICT) and cultural forces (work, education, family life, and leisure) offer more convenience or flexible opportunities to improve living conditions (Jasso, 2003). The flexibility of opportunities in turn influences people’s choices of where to live, as found in residential location theory (Wang, 2018). This choice is complex and involves trade-offs between several factors, including the household’s socio-demographic and socio-economic characteristics, the residential unit itself, the quality of (and attachment to) the neighbourhood, the availability of various facilities and amenities, the accessibility to frequented destinations, and transportation options (Deeyah et al., 2021).

These factors work together to create the household’s lifestyle, which influences daily behaviour patterns and preferences towards building densities, accessibility to retail and services, lot sizes, and modes of transport (Schirmer et al., 2014). Generally, younger people reside in the inner cities because of easy access to job and housing opportunities. Once the life course has taken its path, usually when economic independence and social stability are achieved, preference is usually given to family-orientated suburban lifestyles, implying that several factors influence residential location choices (Wang, 2018).

The household’s socio-demographic and socio-economic characteristics (age, gender, ethnicity, income, size and composition) determine its lifecycle (Schirmer et al., 2014). Choosing a residential unit involves a trade-off between providing proper shelter (Deeyah et al., 2021), property prices, unit sizes, housing types, and architectural features. The quality of residential neighbourhood locations, and people’s attachment to them, is influenced by a myriad of complex factors: the built environment, population densities, land-use mix, open space availability, accessibility to facilities and amenities such as employment, education, retail, and leisure, efficient infrastructure and service networks, transportation options, and the time and costs associated with commuting (Schirmer et al., 2014).

Methodology

South African data on semigration is difficult and costly to obtain. The methodology of the study is therefore discussed from the perspective of the data sources used, as these informed the analyses that could be performed.

Lightstone data analysis

During the Covid-19 pandemic, the South African media reported findings from Lightstone (a company providing information, valuations, and market intelligence on properties in South Africa) showing significant increases in semigration to the Western Cape in 2020 and 2021. Popular towns included Hermanus, Langebaan, St Helena Bay, Groot Brakriver, Plettenberg Bay and Knysna (Creamer Media, 2020; BizzCommunity, 2021; BusinessTech, 2021, MyHome sine anno; MyProperty, 2021; Philpot, 2020; PropertyWheel, 2021). Lightstone was selected to enquire about semigration data to destination locations in the Western Cape for two reasons: firstly, it was identified in media reports as the largest company collecting and analysing semigration data in the country; secondly, Lightstone had developed the only Automated Property Valuation Model in South Africa that is approved by Fitch and Moody’s for mortgage securitisation. Lightstone obtains their data directly from the South African Deeds Office, with access to information on the legal ownership of properties, not merely the households living in properties, and therefore tracks semigration on an individual, personal level. Lightstone defines semigration as the movement from any town in South Africa to a destination location, but it does not include movement from one property to another within the same town (Lightstone, 2021).

Extensive discussions were held with the company throughout 2021 to ensure the collection of a variety of variables available at the city and town geographical level, while balancing the costs of the data. After purchasing the data in December 2021 (Table 1), an initial data screening and cleaning process was initiated: all destination locations and their associated variables were moved into one Excel spreadsheet; the heading structures were changed to align with IBM SPSS 28 specifications, and initial frequency counts were run for each destination location. Through this screening and cleaning process, the main limitation of the Lightstone data was revealed–it had been sampled from a larger dataset. The sample was based only on households with one and two members, thus limiting the number of observations in the dataset, such as the number of properties purchased per destination location. A detailed breakdown of the suburbs where semigrants settled in the various destination locations was not included, and in some instances, data was not provided for all three years requested (2019–2021). Despite this, the Lightstone data provided an indication of the most prominent origin and destination locations where semigration occurred.

Table 1 Lightstone variables used in the study
Full size table

Hermanus was selected for this research because the sampled data provided by Lightstone identified it as one of the top semigration destinations in the Western Cape. The sample for Hermanus contained 305 observations of propertiy purchases, and data was provided for 2020 and 2021. Hermanus is predominantly perceived as a tourism and retirement town (Hermanus Property Sales, 2022; Hermanus Tourism, 2019, MyProperty, 2021), so observing changing semigration trends would make a valuable case study. Only one case study was considered sufficient because the main aim of the study was to explore the extent and characteristics of semigration on a lower geographical level of analysis (suburbs) within a case study area, which was eventually achieved through Property24 data, described in the following sub-section. Because the Lightstone sample for Hermanus was relatively small, it was not possible to perform more advanced statistical calculations; the results should be interpreted in the form of a trend analysis, and no statistical generalisations can be made regarding the broader Hermanus property market. Frequency counts and crosstabulations were calculated on the data in IBM SPSS 28 and figures were created in Excel.

Property24 data and analyses

To overcome the limitations of Lightstone’s sample data, additional data was collected in two phases from Property24 (Property24 sine anno). Property24 is one of South Africa’s leading websites for searching property sales and rentals online. The author did not purchase data from Property24, because they offer data only for the last 50 properties sold. Instead, the open-source data available on the website was used in the first phase of the data collection process. This involved copying open-source data from 2013 to 2022 for Hermanus and its suburbs and putting it manually into Excel (Table 2), and creating figures showing frequency counts.

Table 2 Property24 variables used in the study
Full size table

The second phase involved gathering data on property purchases in Hermanus manually, on a street-by-street level. The Property24 website contains a list of streets allocated to geographic areas smaller than the suburban level. An Excel dataset was created indicating the physical address, the year sold, and the purchase price. The year 2019 was selected to provide a pre-Covid-19 perspective; 2020 provides a Covid-19 perspective, and 2021 and 2022 represent glimpses of a post-Covid-19 property market. The physical address was then used to copy the GPS coordinates from Google Maps, after which the tabular GPS coordinates were converted into a shapefile containing points, of the properties purchased, in ArcMap 10.8.1.

A few limitations were encountered with the Property24 street-level data: in some instances, the purchase dates or prices were unavailable, some geographic areas or streets had ‘no results’, and the data on flats or apartments was not extracted, because it was difficult to obtain accurate GPS coordinate, which are often shared by buildings of multiple occupancy. The geographic areas used to categorise the streets do not align fully with the suburban level, and it is readily acknowledged that there is difficulty in obtaining accurate and complete information for Zwelihle, as informal property market arrangements dominate in such informal settlements. Further, the 2022 data is incomplete because only purchases loaded onto the website between June and September are reflected, and purchases could also have been added to streets that had already been covered during the manual data collection process.

The intersect tool in ArcMap 10.8.1 was used to overlay the shapefile containing points, with the Hermanus sub-place shapefile. It computes a geometric intersection of the input features and shows only the features that overlap in all the layers and/or shapefiles (ArcMap Helpfilea sine anno). Thereafter, the shapefile containing the property purchases was analysed using spatial statistical techniques, including directional distributions, spatial autocorrelation (Moran’s I), optimized hot spot analysis, and hot spot analysis.

Four directional distributions were calculated, one for each year from 2019 to 2022, to measure the standard deviation from the mean centre for point features. Results are presented in the form of ellipses indicating the compactness and orientation of the distribution. Compactness refers to the distribution of the points: points that are closer together have a ‘compact distribution’ (a more circular ellipse), while points that are further apart have a ‘dispersed distribution’ (a more elongated ellipse). Orientation refers to the direction in which the ellipse is facing (ArcMap Helpfileb sine anno).

Six spatial autocorrelations (Moran’s I) were calculated, one for the year variable (from 2019 to 2022 combined), one for the purchase-price variable (from 2019 to 2022 combined), and one for each individual year. This technique assesses whether the distribution of the values is dependent on the spatial distribution of the features. The values can be either distributed randomly, clustered, or dispersed, and are interpreted through the Moran’s I (ranging between 1, 0, and −1), and z-score (ranging from z > 0, z < 0; z = 0). A positive Moran’s I (closer to 1) and z > 0 show clustered values; high values cluster near each other, and low values cluster near each other. A Moran’s I of 0 and z = 0 resemble a random distribution pattern. A negative Moran’s I (closer to −1) and z < 0 indicate dispersed values, high values clustering near low values, and vice-versa. The p-value indicates the level of statistical significance (if any) of the results (ArcMap Helpfilec sine anno).

Optimized hot spot analyses were calculated for each individual year and all years combined (2019–2022). The tool works best with incident data, (points representing events or objects with a focus on the presence or absence of points), rather than the measured attributes of points (ArcGIS Pro Helpfilea sine anno). The Property24 street-level data represents the number of properties purchased per year, (an incident count variable) and the purchase prices (an attribute of the properties). The tool creates ‘hot spots’, areas where more properties were purchased, ‘cold spots’ where fewer properties were purchased, and a random distribution, where the number of properties purchased was not too high nor too low. The analysis field was left blank, making it possible to see where the point clustering is unusually (statistically significantly) intense/sparse (i.e., areas with more and fewer points). The most common aggregation shape (a fishnet polygon mesh) was used as the ‘count incident data aggregation method’, to create a boundary for the calculation to be done. Aggregation polygons could not be used, because the study area does not have a minimum of at least thirty prescribed polygons (ArcGIS Pro Helpfileb sine anno).

In contrast to the optimized hot spot analyses, the hot spot analysis tool examines the distribution of an attribute of a spatial feature, (ArcGIS Pro Helpfilea sine anno) and was used to identify statistically significant hot and cold spots for property prices. Property prices were combined for all years (2019–2022) in the first analysis, followed by separate analyses for each individual year. The tool creates hot spots (areas where property prices are highest), cold spots (areas where property prices are lowest), and a random distribution of property prices. (ArcGIS Pro Helpfilea sine anno).

Results from the optimized hot spot and hot spot analyses are interpreted in the same way: hot spots on the map are represented in red, cold spots in blue, and a random distribution in light yellow. These colours are further distinguished through different shades, indicating varying intensities of clustering. Hot spots range from a dark red, showing areas where the highest values are clustering most intensely, to a lighter red and orange where high values are clustering less intensely. Conversely, cold spots range from a dark blue, in areas where the lowest values are clustering most intensely, to a lighter blue and greyish blue, where low values are clustering less intensely. For the optimized hot spot analyses, the darkest red indicates areas with the highest incident count (the highest number of properties purchased), while the darkest blue indicates areas with the lowest incident count (the lowest number of properties purchased). For the hot spot analyses, the darkest red indicates where the highest property prices are clustered, while the darkest blue indicates where the lowest property prices are clustered (ArcGIS Pro Helpfileb sine anno).

Results and discussions

This section starts by providing a situational overview of the historical development of Hermanus and the possible reasons for its growth as a semigration town, as revealed in the literature. This is followed by a discussion of the number and spatial distribution of sales, property prices and their spatial distribution, and the characteristics of the semigrants to Hermanus.

The appeal of hermanus

Hermanus is situated on the south coast of the Western Cape province in South Africa, and is located between two lagoons, the mountains and the ocean (Fig. 1). It is approximately 120 km southeast of the City of Cape Town metropolitan area. Its history dates to the early 1800s, when it was settled by a travelling teacher and shepherd, Hermanus Pieters. With its natural spring, the area was named ‘Hermanus Pieters se Fonteyn’, or Hermanuspietersfontein, meaning ‘Hermanus Pieters’ Fountain’. This fertile location attracted many farmers during the summer months, while fishermen settled there permanently owing to its abundant fish stocks. As the town grew, a church and school were built in 1886, and it gained a reputation for its sanatorium (a medical establishment for treating chronic illnesses) amongst the British medical profession in the late 1800 s. The town’s name was shortened to ‘Hermanus’ in 1902 by the post office, and it achieved municipal status in 1904. Word of this charming town spread to affluent Britons, and many of the sanatorium buildings were subsequently converted into hotels as the well-established international tourist trade flourished just before World War II. These hotels operated buses from Cape Town to collect Britons who travelled by mail ships to escape the English winter searching for opulent, summery seaside lifestyles. Many of these ‘swallows’ (seasonal migrants) made Hermanus their permanent home after World War II. One of these swallows, Sir William Hoy, settled permanently in Hermanus, and played a critical role in maintaining the quaint and sleepy seaside village atmosphere, when he refused to extend the railway line into the town, fearing it would spoil its natural charm. Hermanus thus remains the only place in South Africa with a railway station but without any trains or railway line (Hermanus Tourism, 2019).

Fig. 1
figure 1

(Source: Author’s creation)

Study area map: Hermanus in the Western Cape, South Africa

Full size image

Hermanus is one of the most popular whale-watching holiday destinations in South Africa and has several pristine beaches and world-class restaurants (Traveller Butlers sine anno). Hermanus has consequently gained popularity as a semigration destination. Younger families settle in Hermanus to improve their living conditions through spacious living options in freehold developments and security estates; the area offers various leisure and recreational amenities, including sea and mountain activities, medical facilities, and educational opportunities. At the same time, residents can work from home or participate in hybrid forms of work within relative proximity to the City of Cape Town metropolitan area (BusinessTech, 2021; MyProperty, 2021).

BetterBond, a mortgage broker in South Africa, experienced a 27% increase in the number of mortgage applications for Hermanus from November 2020 to November 2021. During this period, Hermanus attracted semigrants from Gauteng who were predominantly 35–55, while the retirement trend to the town continued for 55–70 year-olds. The desire for a safe, relaxed, and idyllic lifestyle along the whale coast, the ability to work from home, and the town offering similar amenities and attractions as a city within a coastal setting were the main reasons for semigrating (MyHome sine anno; MyProperty, 2021; Philpot, 2020).

People also took advantage of the record low interest rates during Covid-19 and invested in property in Hermanus at a 60–90% loan rate, making future semigration possible. Many of these homes were furnished and are now rented as Airbnbs. The ratio of investment properties in Hermanus remained steady at 45%, making up the bulk of homeownership in the town. Freehold property prices steadily increased from R1.5 million (2019) to approximately R3 million (2021). The desire for secure lifestyle estates, such as lock-up-and-go living, has also intensified, with the average value of sectional title properties increasing from roughly R1.4 million in the first quarter of 2020, to almost R2.3 million in the last quarter of 2021 (MyProperty, 2021).

The number of sales and their spatial distribution

Property24’s data shows that the number of property purchases remained consistent from 2013 to 2016, with a slight increase in 2017 followed by a sharp decline in 2019, and a further decrease in 2020 (Fig. 2). The 2019 decline was caused by sluggish economic conditions, both locally and globally, thus depressing the labour market and increasing interest rates and inflation, which negatively impacted the housing market (BusinessTech, 2022; News24, 2020). In 2021, the number of property purchases more than doubled–however, compared to the pre-2019 peak, the growth is not exceptionally high, minimising Covid-19’s impact as many people most likely postponed their semigration until 2021. Lightstone’s sample indicates that Covid-19 impacted property purchases as expected, but most purchases occurred in months with lower infection risk levels, when greater societal freedoms were allowed: in 2020, most of the 76 transactions occurred between September and November during lockdown Level 1. There was more than a three-fold increase in transactions in 2021 (229), with greater numbers in April to June and September to October, during the moderate lockdown Levels 1 and 2. However, far lower numbers are found in January to March and July to August, in the more severe lockdown Levels 3 and 4 (South African Government, 2022).

Fig. 2
figure 2

(Source: Property24 sine anno)

Number of properties purchased in Hermanus

Full size image

Property24’s data indicates that the most popular suburbs in Hermanus from 2013 to 2022 include Onrus River, Sandbaai, Vermont, around the town centre of Westcliff, Voëlklip (to the east), and Hawston and Fisherhaven (to the west). Figure 3a maps the spatial locations of properties purchased from 2019 to 2022. It is difficult to discern differences (if any) from this map, but the spatial autocorrelation (Moran’s I) indicates a statistically significant (p < 0.01) clustered pattern for these years (Moran’s I = 0.35, z-score = 17.79). The properties purchased in 2019 are clustered mostly near each other, just as those properties purchased in 2020, 2021, and 2022 are clustered mostly near each other as well. Directional distributions help explore these differences (Fig. 3b): the ellipses for 2019–2021 all have roughly the same sizes–the compactness levels remained relatively unchanged between the years; the 2019 ellipse lies along a west-north-west to east-south-east axis, showing that in 2019 more properties were purchased in Fisherhaven and Hawston; the 2020 and 2021 ellipses have roughly similar orientations facing an easterly direction, indicating that in these years more properties were purchased in the town centre (Hermanus Heights, Eastcliff, and Westcliff), and towards Voëlklip and Fernkloof. The 2022 ellipse differs significantly from the others in compactness and orientation; it is more elongated and lies parallel to the Indian Ocean which indicates that, by contrast to previous years, properties were purchased throughout dispersed locations but mostly along the Indian Ocean coastline.

Fig. 3
figure 3

(Source: Property24 sine anno), b Directional distributions of the properties purchased (2019–2022) (Source: The author’s creation)

a Spatial locations of properties purchased (2019–2022)

Full size image

The optimized hot spot analyses show the presence or absence of points; hot spots indicate where more properties were purchased, while cold spots show the opposite. Vermont and Sandbaai had the most property purchases (darkest red) from 2019 to 2022, while Onrus River, Hemel en Aarde Estate, and Voëlklip performed relatively well (shades of red and orange). Conversely, the town centre, Fernkloof Estate, Hawston, and parts of Fisherhaven had the lowest number of property purchases (darkest blue, and shades of lighter blue/grey) (Fig. 4a). Figure 4b shows the changing pattern over the years. More properties were purchased in Fisherhaven and Hawston in 2019–2020 (more hot spots in red), with a decline in 2021–2022 (more cold spots in blue). Conversely, Fernkloof Estate and Voëlklip experienced the opposite effect over the same period. Sandbaai and Vermont, and to some extent Onrus River, saw major increases in the number of property purchases in 2021 and 2022 respectively.

Fig. 4
figure 4

(Source: Property24 sine anno), b Optimized hot spots of properties purchased (2019–2022, individual) (Source: Property24 sine anno)

a Optimized hot spots of properties purchased (2019–2022 combined)

Full size image

Property prices and their spatial distribution

Property24 data indicates consistently higher average asking prices than actual sale prices from 2013 to 2020 (Fig. 5a). Limited variations are visible from 2013 to 2018, but average asking prices increased drastically in 2019–2020 due to the volatile property market at the time (News24, 2020), and Covid-19 (BusinessTech, 2022). Covid-19’s impact is however more evident when comparing the small gap between the average asking and sale prices for 2021, while 2022 shows signs of recovery as average sale prices are the highest they had ever been. The mean was calculated for average asking prices and sale prices from 2013 to 2022 (Fig. 5b). Suburbs that have higher average asking prices versus sale prices include Kwaaiwater, Voëlklip, Hemel en Aarde Estate, Fernkloof Estate, and Eastcliff. The opposite pattern is observed in Fernkloof, Hermanus Central, Hermanus Beach Club, and Chanteclair. Higher prices in the town centre are influenced partially by the greater number of flats, apartments and guesthouses, while other factors like demographic change, progression or regression (neighbourhood growth or decay), and supply and demand also influence the average asking and sale price (Harcourts, 2012).

Fig. 5
figure 5

(Source: Property24 sine anno), b Average asking and sale prices per suburb (across 2013–2022) (Source: Property24 sine anno)

a Average asking and sale prices (2013–2022)

Full size image

The spatial autocorrelation (Moran’s I) indicates a statistically significant (p < 0.01) clustered pattern for the property prices of 2019–2022 (Moran’s I = 0.28; z-score = 14.72), implying that most higher prices clustered together, as did most lower prices. This is probably because property prices generally do not vary greatly within suburbs. Comparable results were obtained for 2019, 2020 and 2022 (the Moran’s I scores range from 0.21 to 0.22, z-scores from 4.07 to 4.50). Results for 2021 are slightly more clustered (Moran’s I = 0.27, z-score = 7.43), suggesting less variation in prices across suburbs as the market attempted to stabilise after the more stringent lockdown periods of 2020 (South African Government, 2022).

Figure 6a shows a hot spot analysis for property prices from 2019 to 2022. Hot spots are displayed in shades of red, indicating higher property prices, while cold spots represent the opposite. Voëlklip, Fernkloof, Fernkloof Estate, parts of the town centre, Onrus River, and the Hemel en Aarde Estate had the highest property prices (in darkest red). Voëlklip is a traditional holiday location with easy access to pristine beaches and mountain slopes and a bustling town centre (Xplorio sine anno). Onrus River is also an ideal holiday location with easy access to various outdoor activities, such as hiking trails, horse riding, cycling paths, and easy access to the lagoon (SA-venues, 2022). Fernkloof Estate offers luxury golf estate living, while the Hemel en Aarde Estate sits between stellar wine farms and offers various clubhouse activities and spacious properties (Hermanus Property Sales, 2022). All of these factors increase the property prices. The lowest property prices (in darkest blue) were in Zwelihle, parts of Mount Pleasant, Sandbaai, Hawston, and Fisherhaven. Mount Pleasant is one of the oldest suburbs in Hermanus, and has lower property prices (Harcourts, 2012). Zwelihle is the local informal settlement (Cele et al., 2014), and Hawston is a traditional fishing town (Xplorio sine anno); both of which attract mostly poorer households who are temporarily employed in the informal sector, and subsequently earn low incomes. Sandbaai has the most densely populated housing and caters for the older generation, who are often downsizing and need to be close to a wide variety of facilities, resulting in more affordable housing (Xplorio sine anno).

Fig. 6
figure 6

(Source: Property24 sine anno), b Hot spot analysis of property prices (2019–2022, individual) (Source: Property24 sine anno)

a Hot spot analysis of property prices (2019–2022, combined)

Full size image

Figure 6b indicates changing patterns in property prices over the years. The biggest changes were observed in 2021–2022 with Fisherhaven, Hawston, Sandbaai, Mount Pleasant, and parts of Vermont experiencing a sharp decrease in property prices, and consequently greater numbers of dark blue cold spots. Conversely, parts of the Hemel en Aarde Estate, especially along the coast of the town centre, the Fernkloof Estate, Fernkloof, and Voëlklip, saw significantly greater numbers of dark red hot spots, representing increasing property prices. This means that already expensive suburbs became even more expensive due to the market recovery after Covid-19, while less expensive suburbs became even cheaper, thus potentially making it easier for a changing demographic to enter the Hermanus property market.

Characteristics of semigrants

The characteristics of semigrants include their origin locations and provinces, wealth segments, the prices of properties sold in the origin locations, tenure status and length of tenure in the origin location, and age group. Lightstone’s sample shows that most semigrants originated from the Western Cape (169), Gauteng (82), KwaZulu-Natal (18), and Free State (13), with the remaining provinces comprising the remaining number.Footnote 1 Popular origin towns in the Western Cape include Cape Town, Durbanville, Milnerton, and Paarl, while Johannesburg, Tshwane and Sandton were most prominent in Gauteng. eThekwini and Pietermaritzburg (KwaZulu-Natal) and Bloemfontein (Free State) were also common origin locations. Semigrants move from capital cities (Cape Town is the legislative capital, Tshwane the administrative capital, and Bloemfontein the juridical capital) (South Africa Visa sine anno), as well as larger metropolitan areas, cities, and economic-development hubs (South African Cities Network [SACN] 2021). Hermanus’ relative proximity to the City of Cape Town suggests that semigrants can access the functional and economic possibilities of this metropolitan municipality, and therefore could potentially be working remotely or in hybrid format (BizzCommunity, 2021).

Lightstone’s sample for 2020–2021 supports their findings, reported in the media (BizzCommunity, 2021), that people of all wealth segments were semigrating (Fig. 7a). Most semigrants (127) were categorised in the ‘high-value’ wealth segment (R250 000-R700 000 per year), followed by the ‘luxury/wealthy’ and ‘mid-value’ segments. The prices of properties sold in the origin locations (Fig. 7b) also suggest variation in the wealth segments. Although most properties were sold for R2 million-R4 999 999 in 2020 and 2021, less variation is evident in the property prices of 2020, most probably due to the impact of Covid-19. The year 2021 saw a wider spread in property prices, as some properties were much cheaper, e.g., below R1 million, while others were more expensive (up to R5 million), with twelve properties being sold for over R5 million (suggesting the ‘super-wealthy’ wealth segment). Of Lightstone’s sample, 163 semigrants stayed in freehold properties in the origin location in 2020–2021, followed by 73 in estates, and 69 in sectional title deeds. Of these, 140 resided there for less than 5 years; 45 stayed for 5–7 years, and 70 for 7–11 years. The Property24 data indicates that semigrants purchased freehold properties in Hermanus mostly from 2013 to 2022. Freehold properties were especially popular in Onrus River, Sandbaai, Vermont, Voëlklip, Westcliff, Hawston, Fisherhaven, and Fernkloof Estate, whereas sectional title deeds were more prevalent in Onrus River, Sandbaai, Westcliff, and Eastcliff.

Fig. 7
figure 7

(Source: Lightstone’s sample), b Prices of the properties sold in the origin locations (Source: Lightstone’s sample)

a Wealth segments of semigrants

Full size image

Lightstone’s sample data suggests that younger age groups (i.e., other than retirees) are entering the Hermanus property market: 123 of the semigrants were 36–49 years old, followed closely by 122 in the 50 + category, while the number of 18–35 year-olds increased from 9 in 2020 to 51 in 2021. Property24 splits its age profiles according to sellers, buyers, and owners. Sellers and buyers are those who participated in the property market within the last six months, while owners purchased properties more than six months ago. Although the age data extracted in September 2022 does not align fully with the dates for the other variables used in the study, it was included because it still yields valuable results (Fig. 8a). The sellers and owners in Hermanus and in most suburbs were predominantly over 65, followed by 50–64 year-olds. Just over 36% of buyers were 50–64, indicating potential preparation for retirement. When looking at the 18–35 and 36–49 age groups, buyers comprise the highest percentages (just over 12% and 28%, respectively), suggesting that younger age groups are entering the property market.

Fig. 8
figure 8

(Source: Property24 sine anno)

a-e Age categorisation of sellers, buyers, and owners

Full size image

Although each suburb has a unique categorisation of sellers, buyers and owners, some general trends were observed (Figs. 8b–e). Suburbs where people 50–64 and over 65 act either as sellers, buyers, or owners in the property market include Eastcliff, Fernkloof, Fernkloof Estate, Hemel en Aarde Estate, Hermanus Beach Club, and Hermanus Central. Conversely, suburbs that experienced an increase in the younger age groups 18–35 and 36–49 year olds include Fisherhaven, Hawston, Sandbaai, and Westcliff. Fisherhaven and Hawston had some of the cheapest property prices in Hermanus (Fig. 6a), and it takes only some fifteen minutes to reach the town centre (Xplorio sine anno). The more affordable property prices in Sandbaai (Fig. 6a), and the more density-built environment, could be attracting the younger population with its access to a variety of facilities (Harcourts, 2012). Conveniently situated close to the town centre, Westcliff is one of the older suburbs in Hermanus, has private and public hospitals, and homes within walking distance of the beach (Xplorio sine anno).

Conclusions and policy implications

The aim of the study was to investigate the extent and characteristics of semigration to Hermanus, South Africa. This study was initiated by media reports of a boom in semigration patterns to the Western Cape, and more specifically Hermanus, due to the Covid-19 pandemic. Covid-19 brought substantial changes to societal life; significant improvements to smart office technology enabled a geographical shift in the South African economic landscape, as flexible work-from-home and hybrid forms of work became the norm, which in turn affected semigration patterns as people moved to more decentralised locations to improve their living conditions.

Results from this study indicate a depressed housing market in Hermanus since 2019, with Covid-19 slowing the market even further in 2020. Not only did the number of property purchases decrease, but prices decreased as well–not an unexpected result as the pandemic brought the global economy to a near standstill, and governments worldwide sought to manage the disease. The year 2020 was marked by stringent lockdowns, which made it difficult to semigrate, therefore an increase in property purchases and property prices could be anticipated for 2021. Additionally, the number of properties purchased in 2021 is similar to pre-2019 levels, showing that Covid-19 slowed the pace of purchases, but did not necessarily increase it quite as drastically as media reports had suggested. This offers some anecdotal evidence of potential semigration, but does not necessarily confirm it.

A geographic shift in the locations where properties were purchased is however noticeable from 2019 to 2022, suggesting that Covid-19 had some influence on the patterns of semigration. More properties were purchased in the west of Hermanus in 2019 (Fisherhaven and Hawston), in the east in 2020–2021 (Hermanus Heights, Eastcliff, Westcliff, Voëlklip, and Fernkloof), and along the Indian Ocean in a dispersed pattern in 2022. These results concur with changing property prices, as some suburbs experienced price increases such as the Hemel en Aarde and Fernkloof Estates, Fernkloof, Voëlklip, and parts of the town centre, while others showed decreases, e.g., Fisherhaven, Hawston, Sandbaai, Mount Pleasant, and parts of Vermont. These results could be evidence of a changing demographic property market, which in turn could hint at a shifting semigration pattern.

Covid-19’s potential impact on the semigration patterns to Hermanus is most evident in the changing characteristics of semigrants, such as their origin locations, wealth segments, and age groups. Most of the semigrants originated from the Western Cape, especially the City of Cape Town, indicating that the functional proximity to the metropolitan area is important possibly to maintain work-from-home or a hybrid work status. Results indicate that people of various wealth segments are entering the property market in Hermanus, not just the super-wealthy; younger age groups (18–35 and 36–49) are also buying into the market, both indicating changing demographic patterns associated with the semigration phenomenon.

When considering all the results, it is difficult to discern whether changes in the Hermanus property market are attributable to various forms of migration, or more specifically to semigration due solely to the Covid-19 pandemic. Results do however indicate some anecdotal evidence of change to the semigration phenomenon in the town. Semigration can be used as a measure of the overall quality and efficiency of the municipality that governs and manages these hot spots. The municipalities tend to be well-managed, offer a slower-paced lifestyle, reduced crime levels, and adequate service delivery and infrastructure availability (BizzCommunity, 2023).

The greater Hermanus area was identified as a regional node in the municipal Spatial Development Framework (Overstrand Municipality, 2020). This implies that the findings from this study, such as anecdotal evidence of increased semigration to the town, could have significant impact on municipal policies and strategies. The most important of these could be the expansion of economic development within Hermanus, beyond the tourism and hospitality industries which currently dominate it, as highly skilled semigrants have the entrepreneurial and technical expertise to create new and different employment opportunities within the town. The municipality could boost the tax base in the town, which could be used to address some of the current service-delivery and infrastructural challenges, exacerbated by increasing semigration. There are a number of specific services that require attention, including additional bulk water services, especially for Hermanus West; repairing the water network services in Hermanus West, (particulatly in the suburbs of Onrus River and Sandbaai), and Hermanus Central, just like those works currently underway in Hermanus East; upgrading the stormwater and wastewater infrastructure in Hermanus Central, and to a lesser extent in Hermanus West; monitoring the demand and usage of electricity, especially in Hermanus Central; ensuring adequate upgrades to the road network within the town to accommodate heavier traffic flows and congestion throughout the year, rather than only during tourist and holiday seasons.

Other important policy aspects to consider include supply-side constraints to development that can arise within the town due to geographical and topographical limitations (e.g., the coastline, mountain ranges, and the surrounding nature reserves). It could be suggested that a higher-density development approach would be needed, yet tempered by a smaller town context, as such measures could change the character and ambience of the town for which it is so well known. Further policy considerations include additional requirements for amenities and facilities like hospitals and clinics, primary and secondary schools, libraries, community halls, post offices, open spaces, recreational and sports facilities, and ICT services, beyond what is currently anticipated within the Overstrand Municipality (2020: 110). Lastly, the possibility of privatised enclaves for homogenous groups, such as gated security estates, could negatively impact the town’s progress towards achieving integration and desegregation.

Recommendations for further research include doing a qualitative ethnographic case study of locals’ perceptions and experiences regarding semigration in the town, and a quantitative analysis of changing patterns of infrastructure and service-delivery usage.