The dynamic relationship between house prices and farmland prices: evidence from the Australian region

In a broader historical context, Australian farmland values have soared. Broadacre farmland prices in Australia have generally experienced strong growth since 1992 (Australian Bureau of Agricultural and Resource Economics and Sciences ABARES, 2025). The price per hectare increased from around $1,165 in 1992 to roughly $9,617 in 2025. Over the last decade (2015–2025), the average price per hectare of broadacre farmland has grown at an average annual rate of 10.2%. The recent surge over the past decade has been attributed to high demand from diverse buyers, optimism in agriculture due to good commodity prices and favourable weather, low interest rates and limited farmland availability in Australia (Rural Bank, 2022; Gholipour et al., 2025).

The consequences of the rise in farmland prices have attracted considerable attention among policymakers, farming industry analysts, media and academics (e.g. Greenville, 2025; Mackintosh, 2021; Gholipour et al., 2025). For example, Gholipour et al. (2025) show that rising farmland prices indirectly diminish the farming industry’s profitability by increasing production costs. Greenville (2025) argues that higher land prices boost farmer wealth and equity levels, which in turn increase lending capacity and support productivity growth by enabling farmers to access capital for further investments in land and technology. On the other hand, he contends that relatively high land prices may create barriers to entry and expansion.

We build on recent literature examining the impact of rising farmland prices on the Australian economy by investigating whether these prices are linked to house prices in regional areas. Our study focuses on housing markets, given arguments that the recent surge in farmland values has contributed to higher house prices in these areas. Conversely, some argue that escalating house prices in urban areas near farming regions have, in turn, driven up farmland values (e.g. Jasper and Honan, 2018; Claughton, 2018; McDonald, 2018).

The purpose of this study is to clarify these conflicting views and examine the dynamic relationship between house prices and farmland prices in the Central Coast region in the state of New South Wales (NSW), Australia, over the period 1992–2024. The Central Coast region is one of the closest communities to Greater Sydney, offering a residential option for priced-out households and renters from the metropolitan city. With a coastline of almost 80 km, the region’s farming activities are dominated by poultry meat production, nurseries, cut flowers, turf and vegetables (New South Wales Department of Primary Industries [NSWDPI], 2020). Therefore, proximity to Greater Sydney and the presence of these farming activities make the Central Coast region an ideal case study. Applying autoregressive distributed lag (ARDL) and dynamic ordinary least squares (DOLS) estimators, our main finding indicates a significant and positive relationship between house prices and farmland prices.

This study offers four contributions to the existing literature. Firstly, while prior studies have examined the impact of urban pressure (or urban sprawl or urban influence) on farmland values (e.g. Livanis et al., 2006; Guiling et al., 2009; Tsoodle et al., 2007; Chicoine, 1981; Plantinga et al., 2002; Kuethe et al., 2011; Roe et al., 2004) or the effect of house prices on farmland prices (e.g. Zhang and Nickerson, 2015), none have analysed the bi-directional relationship between these two markets. For instance, Zhang and Nickerson (2015), using a hedonic model, found that the residential housing bust of 2009–2010 negatively affected farmland values near urban areas in western Ohio. Secondly, most existing research relies on U.S. data (e.g. Oklahoma, Kansas, OH; Chicago) and a few other countries (e.g. Turkey (Bayramoglu and Gundogmus, 2008) and to the best of our knowledge, no study has explored the bi-directional link between house prices and farmland prices in farming regions of Australia in the long run. Thirdly, we contribute to the literature on the ripple- also referred to as spillover or diffusion-effects in property markets. Existing studies on ripple effects typically focus on linkages between different sub-housing markets within a country or region (e.g. for a review, see Ranjbar et al., 2022). Our study takes a slightly different approach by examining spillover effects between housing and farmland markets in farming regions, assessing whether price movements in one sector influence the other. Finally, to the best of our knowledge, our study is the first to examine the relationship between farmland prices and both strata and non-strata property markets in Australia.

In contrast, typical freestanding properties (e.g. detached houses) do not involve shared ownership of common areas; instead, owners have exclusive ownership and full responsibility for the land and structures on their individual title. Conducting separate analyses is valuable, given that strata and freestanding properties have slightly different market dynamics in the Australian housing sector (e.g. Lu et al., 2023).

Our results are particularly relevant to farmers, local councils and property investors. Farm real estate represents a major component of the farm sector’s balance sheet and a key asset in most farm household investment portfolios (Zhang and Nickerson, 2015). Changes in farm real estate values have significant implications for farm sector health, household well-being (Zhang and Nickerson, 2015) and overall profitability (Gholipour et al., 2025). Understanding the determinants of farmland price movements is therefore of critical interest to farmers and policymakers. Local councils are also implicated in these dynamics, as rising house prices can intensify incentives for the conversion of farmland to alternative uses in agricultural regions. Finally, property investors can leverage our results to identify farming regions with strong investment potential by monitoring housing market trends.

The rest of the article is organised as follows. Section 2 provides the conceptual background on the link between farmland prices and house prices. Section 3 describes the data and methodology. Section 4 presents the results and Section 5 concludes the study with implications.

Several studies have examined the relationship between urban pressure, house prices and farmland (or agricultural land) prices (e.g. Livanis et al., 2006; Guiling et al., 2009; Tsoodle et al., 2007; Chicoine, 1981; Plantinga et al., 2002; Roe et al., 2004; Bayramoglu and Gündoğmuş, 2008; Kuethe et al., 2011). Using a variety of data sets and empirical methodologies, these studies examine the one-way effects that agricultural and residential land markets exert on each other.

Regarding the positive effect of urban pressure on farmland values, the urban sprawl hypothesis is frequently cited. Livanis et al., 2006 provide both theoretical and empirical evidence identifying three key channels through which urban pressure affects farmland values: changes in nonfarm opportunities, the speculative effects of anticipated development and variations in net agricultural returns. More specifically, Kuethe et al. (2011) argue that farms located near urban areas benefit from improved access to markets and ports, leading to lower transportation costs. As a result, these farms may earn economic returns exceeding those of comparable farms located farther from urban centres. In addition, farmland at the urban fringe provides recreational opportunities and lifestyle amenities that are valued by nearby urban populations. At the same time, such land is increasingly exposed to development pressure, with its value often bid up by competing land uses, particularly residential and commercial development.

Linking to the house prices, it is argued that as urban areas expand, demand for housing in the urban fringe rises (McDonald, 2018), pushing up house prices in these locations. Consequently, residential development becomes more profitable than agricultural use. These expectations, in turn, drive farmland prices upward, especially in peri-urban areas ^[1].

Conversely, growth in farmland values can also exert upward pressure on house prices in adjacent rural areas. As farmland becomes more attractive for development, higher land acquisition costs are passed through to housing prices. In other words, increases in farmland values raise the cost of land, for example, Plantinga et al. (2002) with development potential, as expectations of rezoning or residential use are capitalised into land prices even before rezoning occurs. Moreover, rising farmland prices enhance farmers’ borrowing capacity through collateral effects (Greenville, 2024), thereby increasing the likelihood of their investment in the local housing market. Consequently, higher farmland prices can ultimately translate into higher housing demand and house prices.

In a broader context, our study falls within the literature on spillover or ripple effects in property markets. Previous research on the ripple effects has primarily focused on housing markets and has shown that changes in house prices in one location (including suburbs, councils/municipalities, cities, states/provinces or even countries) tend to ripple or spillover to adjacent or neighbouring areas over time (e.g. Gholipour et al., 2016; Bangura and Lee, 2023a). Ripple effects may also occur within a city’s housing market, where price movements in higher-value segments cascade into lower-priced areas (e.g. Bangura and Lee, 2020).

In light of stronger evidence regarding the effect of house prices on farmland values, we propose the following hypothesis:

We focus on the NSW Central Coast as the context of study, given the availability of data for farmland prices and house prices. In addition, located between Sydney and Newcastle, the NSW Central Coast Council has an area of about 1,681 km², including rural farmlands, making it the sixth-largest urban area in Australia (Central Coast Council, 2025). We collect annual data from 1992 to 2024 from various government departments and institutions in Australia. Data on price per hectare of NSW coastal farmland (FLP) are sourced from the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES, 2025), while the cash rate (CR) is obtained from the Reserve Bank of Australia (RBA, 2025). The estimated resident population (ERP) of the NSW Central Coast and the median regional annual income (RAI) are gathered from the Australian Bureau of Statistics ABS (2025). The median prices for non-strata and strata dwellings (HP) are sourced from the New South Wales Department of Communities and Justice (New South Wales Department of Communities and Justice [NSDWCJ], 2025). This sales data is originally derived from the information provided on the “Notice of Sale or Transfer of Land” form lodged with Land and Property Information NSW. The median RAI indicates the personal income of residents in regional NSW. The CR is the overnight lending rate between commercial banks, set by the RBA. The summary statistics of these variables are presented in Table 1.

We use the augmented Dickey-Fuller (ADF) test to verify stationarity in the variables and the ARDL bounds cointegration test. The ARDL is based on an error correction framework that can test for cointegration, even in small sample sizes and with mixed orders of stationarity in the variables (e.g. Bangura et al., 2025; Tursoy, 2019). The ARDL is designed with built-in features that could internally address potential confounding effects in the system (Pei et al., 2019). It also minimises the likelihood of endogeneity and autocorrelation in the model (e.g. Wen and Dai, 2020). As developed by Pesaran et al. (2001), the adopted ARDL models in our study become:

From equations (1) and (2), FLP denotes farmland price per hectare of NSW coastal area, HP denotes housing (non-strata and strata) prices, Δ denotes first-order difference, α₀ and ω₀ are the constant terms in equations (1) and (2), respectively, X and Y as well as M and N are the optimal lag length of equations (1) and (2), respectively. ε_t1 and η_t1 are the continuous independent and normally distributed random disturbance terms in equations (1) and (2), respectively, α_1i and α_2i are the short-run coefficients and β₃ and β₄ are the long-run coefficients in equation (1), while ω_1i and ω_2i are the short-run coefficients and π₃ and π₄ are the long-run coefficients in equation (2).

We use the Schwarz information criterion, Akaike information criterion and Hannan-Quinn information criterion (HQ) to determine the appropriate maximum lags of X and Y as well as M and N. In equation (1), the null hypothesis of no cointegration between the variables: H₀: β₃ = β₄= 0 is tested against the alternative that H₁: β₃ ≠ β₄≠ 0. This process is repeated in equation (2). The F-test is the joint significance of these coefficients and it is compared with the critical values of the upper and lower bounds provided by Pesaran et al. (2001). We reject the null hypothesis when the F-statistic is greater than the upper bound and we fail to reject the null hypothesis when the F-statistic is below the lower bound, suggesting the lack of cointegration between the variables. If the F-statistic lies between these two limits, the system is undefined (Pesaran et al., 2001). This test was previously applied by Bangura et al. (2025), Gounopoulos et al. (2019) and Lee et al. (2017).

Once cointegration is established between the farmland price per hectare of the NSW coastal area and each of the non-strata and strata housing prices, we used the Granger-causality and the error correction model (ECM) to determine the short and long-term causalities, respectively ^[2]. The ECMs are defined in equations (3) and (4):

Although the results of the ECM in equations (3) and (4) can generate the short and long-term coefficients without any loss of generality, our focus is on the long-term causality between the variable pair. From equations (3) and (4), the value of the error correction term (Ψ) is expected to be negative and significant, indicating long-run causality. The existence of causality means the independent variable holds valuable information for predicting the dependent variable.

After the analysis of the cointegration and causality, our next step is to examine how changes in key market variables, such as the Australian CR, ERP and the median annual regional income (MRI) of NSW Central Coast, influence the farmland price per hectare (FLP) of the NSW coastal area, as well as housing prices (HP) for non-strata and strata properties. The results from the DOLS model can reveal how shocks in these market variables impact the dominant markets – the housing market – and how this influence may transmit to farmland prices. This analysis could help explain the connection between housing prices and farmland prices per hectare through these variables. With FLP and HP as the dependent variables in separate estimations, the DOLS model becomes:

From equation (5), the coefficient of the Australian CR is expected to be negative (Lee and Park, 2022), while those of the ERP and the MRI of NSW Central Coast are hypothesised to be positive (Yanotti et al., 2024; Bangura and Lee, 2023b). Key diagnostic tests were conducted to check the validity of the model.

Our analysis of whether a long-term relationship exists between housing prices and farmland prices begins with a unit root test, followed by a cointegration test between the farmland prices and each of the non-strata and strata dwellings.

The results of the ADF unit root test for all logarithmic variables used in the study are shown in Table 2. Except for the on-level stationarity of the CR and the ERP of the Central Coast, all other variables are stationary in first difference at their respective levels. The information criterion indicates an optimal lag of two. The diagnostic tests strongly reject the presence of heteroskedasticity and serial correlation in the model, while the residuals are normally distributed. The combination of level and first-difference stationarity in the variables further supports the use of the ARDL bounds cointegration technique in the DOLS model. Although farmland, non-strata and strata prices are all first-difference stationary, we still use the ARDL bounds cointegration test because of its built-in features that minimise the likelihood of endogeneity and autocorrelation in the model. The results of the ARDL bounds cointegration test between the price per hectare for NSW Coastal farmland and the median prices for non-strata and strata in the Central Coast, respectively, are presented in Table 3.

From Table 3, clear evidence shows cointegration between the price per hectare for NSW Coastal farmland as the dependent variable and the median price for non-strata in the Central Coast as the independent variable. We reject the hypothesis of no cointegration in this model, as the F-statistic, 6.83, exceeds the upper bound of 6.76 even at the 1% significance level. However, there is no reciprocal cointegration when the variables are swapped. For the strata market, there is also a clear rejection of no cointegration when the median price for strata in the Central Coast is the independent variable. Similar to the non-strata market, the F-statistic, 15.30, is above the upper limit. These cointegration results suggest that a long-term relationship exists only when housing prices, both non-strata and strata, are used as explanatory variables. The findings show that the price per hectare of Coastal farmland and housing prices for non-strata and strata are interconnected and share a long-term relationship.

The results indicate that housing prices are positively and significantly associated with farmland prices. Any increase in housing prices is expected to also raise farmland prices in the Central Coast of NSW. This suggests that farmland in peri-urban areas functions as a forward-looking asset, with values shaped by anticipated future conversion to residential and commercial uses rather than by current agricultural returns.

To further investigate this relationship, we conduct a Granger causality and an ECM to determine short and long-term causalities. They are reported in Tables 4(a)-(b) and 5.

From Table 4a, there is evidence of Granger causality when the median price for non-strata dwellings in the Central Coast is the independent variable without any reciprocal effect. This indicates that prices of non-strata dwellings contain useful information that influences farmland prices. The unidirectional Granger causality results suggest that an increase in housing prices is likely to impact farmland prices in the coastal region of NSW. The strata market in Table 4b shows similar results in the Granger causality test, as prices in this market are also expected to influence farmland prices without a contemporaneous effect. The absence of reciprocity suggests that changes in housing market prices could trigger changes in farmland prices. It shows an expected spillover from the housing market to the farmland market in the region. This finding is consistent with the results of Zhang and Nickerson (2015), who showed that a negative shock to housing prices reduces farmland prices in nearby areas.

These results are intuitively appealing. Empirical evidence has shown that soaring residential prices in Sydney are affecting many households, who resort to migrating to alternative housing markets. As such, priced-out households in this housing market are likely to explore regional residential alternatives to seek relative affordability. Bangura and Lee (2023a), for instance, find strong spatial connectivity between coastal cities and Greater Sydney due to improved travelling times. A recent publication by the Regional Australia Institute (RAI, 2025) reveals that regional Australia is experiencing the largest city-to-regions migration in the 20 years outside of the COVID-19 pandemic. They find that about 40% of city dwellers are considering a move to the regions. RAI therefore calls for an urgent intervention to ensure liveability factors in regional areas keep pace with the growth in population and the resulting demand. In the March 2021 quarter, for instance, Greater Sydney recorded a net loss of 5,100 people to the rest of the state, including regional areas such as the Central Coast, which is more than the 3,100 who left for interstate destinations (Australian Bureau of Statistics[ABS], 2021).

In addition, Bangura and Lee (2023a) report that the Hunter region experienced a 200% rise in net internal migration by 2017. More specifically, Infrastructure Australia (2022) indicates that the M1 Pacific Motorway enhances traffic flow and reduces travelling times, with over 90,000 light and heavy vehicles passing between the Hunter, Central Coast and Sydney each day. The travel time savings, among other factors, are attracting households to move to the Central Coast, increasing housing demand. This ongoing migration to the region is contributing to rising housing prices, undoubtedly impacting farmland prices through various channels like urban expansion and speculation in the market.

The concept of urban expansion is a major driver of farmland conversion. With increasing house prices in metropolitan areas (Martin and Pawson, 2024; Bangura and Lee, 2023a), especially in Greater Sydney, prospective homebuyers or renters often seek relative affordability in the city’s outskirts, making the Central Coast a viable option. Developers and investors also seek out cheaper land in nearby communities with growth potential to build new homes. This pressure can lead to land-use changes as farmland is being converted to residential uses by the relevant state or local authorities. This is reiterated by NSWDPI (2020), as they highlighted that the expansion of residential and lifestyle development has incrementally pushed farming out of some areas, making it difficult for some farmers to operate. Saqib et al. (2024) and Francis et al. (2012) support this, as they noted that housing market expansion often requires accompanying services like transport, industry, businesses and schools, which further contribute to farmland conversion. For example, in the USA, about 500,000 hectares of food and fibre land are converted each year to non-agricultural uses. With a 1% annual population growth, this is expected to halve the hectares per person from today’s 0.6–0.3 by 2050. In Canada, despite higher land density, farmland is still being lost in fertile coastal areas (Francis et al., 2012). Leitner et al. (2023) also note that large-scale planned developments and infrastructure projects can cause widespread displacement of people and a reduction in agricultural fields and vegetable plots. This drive to expand housing and key services certainly impacts farmland prices.

There are speculations and expectations in the housing market, which have also led to farmland conversion. As house prices constitute both building value and land value, an increase in house prices generally signals an increase in land values too. Farmland owners may be motivated to sell to property buyers expecting the farmland to be rezoned. In general, the run-up of housing prices, changes in housing policies and investment strategies of institutional investors, as well as announced or planned government development projects, often generate speculative behaviour in the property market (Leitner et al., 2023; Leitner and Sheppard, 2022). Leitner et al. (2023), for instance, echo that the everyday financial, land and property speculative tendency among individuals and organisations is triggered by planned development in the peri-urban areas. These economic agents seek to enhance their wealth by speculating on the future value of key assets like land and property. This speculative conduit can impact housing prices, which are expected to spillover to farmlands.

As housing markets function within the wider economy, the importance of market fundamentals becomes evident. In this part of the analytical framework, we examine how key variables including the Australian CR, the ERP and the MRI of NSW Central Coast influence farmland price per hectare in the NSW coastal region, along with housing prices for both non-strata and strata properties using the DOLS framework. Before estimating the DOLS model, we test for the existence of a long-run relationship between each dependent variable and the vector of explanatory variables in equation (5) using the ARDL bounds cointegration test. The results are reported in Table 6. There is a clear rejection of no cointegration in each of the models – 5% significance level for the farmland price per hectare, 5% for the non-strata price and 10% for the strata price models. With the establishment of cointegration, we estimate the DOLS and the findings are presented in Tables 7–9. The results from Table 10 show that the models are a good fit.

From Tables 8 and 9, we find a negative and statistically significant relationship between the Australian CR and both non-strata and strata housing prices. The CR has a direct and strong influence on other interest rates, including deposit and mortgage lending rates (Australian Housing and Urban Research Institute (Australian Housing and Urban Research Institute [AHURI], 2022). Typically, a change in mortgage interest rates affects not only the maximum loan size a borrower can access but also, more importantly, their actual repayment amounts (Kearns, 2022). Consequently, a lower CR leads to reduced mortgage lending rates from financial institutions, making borrowing more attractive to both homebuyers and investors.

Based on the results in Tables 8 and 9, a percentage decrease in the CR is projected to increase the prices of non-strata and strata dwellings in NSW Central Coast by 4.1% and 2.9%, respectively, assuming all other factors remain constant. This outward shift in housing demand will drive up prices for both types of dwellings. Existing homeowners may also take advantage of lower lending rates to refinance or engage in sequential buying. These amplified effects of lending rate shocks on housing prices are closely linked to the growing reliance on borrowed funds for housing purchases (Lee and Park, 2022). Farm activities, by contrast, are often seasonal and may not respond directly to changes in the CR. However, through the asset-valuation channel, rising housing prices can still affect farmland prices. As established earlier, increases in housing prices are expected to spillover into farmland prices, which helps explain the negative relationship between the CR and farmland prices.

The ERP of NSW Central Coast has a positive and statistically significant relationship with the prices of both non-strata and strata dwellings. As the area’s population increases, whether through net internal migration or natural growth, it generates new housing demand, which eventually drives up prices. Similar findings were noted by Bangura and Lee (2023b) in their analysis of the determinants of home ownership in Greater Sydney. The earlier discussion highlights a surge in internal migration from metropolitan Sydney to regional areas of NSW, especially those closer to the city, such as the Central Coast. When people move into a particular region, housing prices tend to rise not only within that market and nearby areas but also in more distant locations (Yanotti et al., 2024). However, the ERP has an insignificant effect on farmland prices. This suggests that population growth in the local market does not necessarily create a demand–supply imbalance in farm products and therefore may not affect farmland prices, which helps explain the insignificance of this variable.

The median income is also an important factor influencing housing prices, but not farmland prices. It has a positive relationship with residential prices, signalling that households tend to buy or invest in properties when their income rises. As a result, an increase in income can boost housing prices through higher demand, greater borrowing capacity and increased competition in the housing market. When household income improves, it enhances their ability to access borrowed funds from lenders. This leads to an improved financial situation, which expands demand and subsequently impacts housing prices. Similar findings were documented in Goda et al. (2020) in their study on income inequality and housing prices in organisation for economic co-operation and development (OECD) countries. They find that the rise in OECD house prices is partly explained by a top-income-induced increase in housing demand, highlighting the interaction between income and housing prices. There is also a positive and significant relationship between income and farmland prices. Income growth could boost household savings, which might be channelled to further investment in farmlands and subsequently raise farmland prices.

In summary, housing prices and farmland prices are not disconnected. Our empirical findings show an existing long-term relationship between these markets. As housing prices become extremely high in metropolitan cities like Greater Sydney, priced-out households are likely to seek residential alternatives in nearby communities, such as the NSW Central Coast. This move expands demand for housing in the destination communities, which will push up prices. The twin demand for housing and the accompanying services could prompt the rezoning of farmlands to residential uses. The results of the impact of market fundamentals support this. As the Australian CR drops, it makes borrowing cheaper. When income improves, housing demand expands and population growth also adds to the already increasing demand for housing. The overall effect is a surge in demand for housing, which pushes up prices that diffuse to farmlands.

In this study, we contribute to the ongoing debate on whether house prices influence farmland prices or vice versa in Australia. Using annual data from the Central Coast region of NSW over the period 1992–2024 and applying time-series ARDL and DOLS estimators, we find robust evidence that changes in both strata and non-strata house prices are positively related to farmland prices.

This finding has several implications for various stakeholders. Farmers in the region should anticipate increases (or decreases) in the value of their farmland following positive (or negative) shocks to nearby residential property prices. Our results are also relevant for the farming industry and agricultural policymakers, as positive shocks to farmland prices (following a positive shock to house prices) may indirectly reduce the profitability of farming operations (Gholipour et al., 2025). This, in turn, can contribute to farmer exit and potentially affect the long-term sustainability of the agricultural sector and Australia’s food security. To preserve the Australian farming sector and reduce the likelihood of farmland-use conversion in agricultural regions, local councils should closely monitor house price dynamics. Housing booms in these areas may stimulate greater incentives for residential development, increasing pressure to convert farmland. In addition, property investors in agricultural land may also find that monitoring house price movements offers valuable insight into future changes in farmland values within the region.

Finally, we acknowledge the limitations of our study. Due to data constraints over the study period, we are unable to empirically test which channels explain the relationship between house prices and farmland prices. Accordingly, our empirical analysis provides indirect support for the theoretical framework rather than a direct test of its individual components. Future research may explore, for example, how rezoning expectations, speculative demand or development pressure mediate this relationship. Future studies could also extend our analysis to remote agricultural regions, as our findings for the Central Coast- uniquely positioned near Sydney (an urban area)- may not be generalisable to more remote agricultural regions. A further extension of our study would involve analysing the relationship between house prices and farmland prices in the Central Coast once the Sydney–Newcastle high-speed rail becomes operational. With a planned stop on the Central Coast, this infrastructure is expected to significantly influence migration flows, housing demand and activity within the farming sector.

The authors would like to thank the two reviewers for their constructive and useful comments on earlier version of this paper.