Impact Of Macroeconomic Factors On UK Housing Market After Brexit

Importance of the Housing Sector in the UK Economy

From the macroeconomic perspective, the housing sector is one of the most important sectors  of the economy. Fluctations in housing prices, specifically and the real estate industry as a whole has a significant impact on some of the key variables of the macro economy including the Gross Dometic Product, inflation and employment. t  the real estate sector accounts for 20% of gross fixed capital formation and around 50% of the country’s fixed capital stock (ONS, 2017). Consequently, it is always under the radar of researchers, policy makers and investors. Over the past decades, the UK market, along with other developed countries, has experienced significant fluctuations in asset prices that arose due to the subprime mortgage crisis and local housing market bubbles. The ability to forecast future fluctuations in asset prices will allow politicians to take preventative measures and have a benevolent effect on economic stability.

Researchers, generally, tend to agree on the overall impact of macroeconomic factors on the housing market but they have not been able to identify a universal set of variables to predict future prices. Various past studies (Cristadoro et al., 2005; Giannone and Matheson, 2007; Gupta et al., 2009; Himmelberg et al., 2005; Brunnermeier and Julliard, 2008; El-Montasser et al., 2016 and Martins et al., 2015) applied numerous approaches to projecting property prices and their determinants. However, due to the large discrepancy in the results, attempts to create a unified forecasting model have unsuccessful, leaving a significant uncertainty about the future of the UK housing market. This uncertainty  following the Brexit referendum, the common parlance term used to signify cessation of the United Kingdom as a member of the European Union. . Currently, the UK housing market faces a great uncertainty  regarding future regulation, workforce supply, changes in migration patterns and expected economic growth.

The effect of different Brexit scenarios on house prices are being widely speculated. In general,, housing associations expect a  decline in prices in the future, while economists and business analysts  expect a moderate growth. However, these expectations are based solely on assumptions and have no empirical confirmation yet. This reveals a gap in the existing empirical literature. The effects of Brexit and their impact on the British economy remain poorly understood and there is even lesser clarity regarding the housing market in the context of Brexit.   This study aims at contributing to the knowledge by assessing the drivers of house prices in the UK and projecting future property prices for the next five years after Brexit.

The main aim of this dissertation is to explore =the extent to which housing prices (for house ownership) in the UK market will change after the country’s exit from the European Union in the next 5 years. Thus, the objectives of the study are a:

1. To identify a set of factors that drive house prices in the UK property market;
2. To forecast future house prices based on the explored set of determinants.

Limitations in Forecasting Future Prices of UK Housing Market

This dissertation focuses on the p given objectives by analysing future house prices using the method of vector autoregression (VAR) analysis. This method allows for identifying how house prices respond to the underlying macroeconomic and demographic shocks associated with Brexit and how this will be reflected in future dynamics of house prices. The analysis is performed based on the data for the period 1975-2016.

This study consists of six chapters and is structured as follows. Chapter 2 analyses theory and empirical evidences on fluctuations in house prices and their links to macroeconomic events. Chapter 3 consists of three subchapters, which in turn formulate hypotheses of this study, provide overview of econometric model applied and describe data used. Chapter 4 presents the results and interprets the main findings of the VAR analysis. Chapter 5 critiques the methods and practices used in this study and discusses potential implications of obtained results for policy makers. Finally, Chapter 6 draws main conclusions and limitations and develops recommendation for future researchers.

According to Keeny (1998), the real estate market, including the residential housing market, is significantly different from other markets for a number of reasons. The goods on the housing market have a dual character – they can be both a commodity and an investment asset. However, there are several other differences, which were distinguished by Quigley (1998) and Miles (1995). These differences included  the relatively high cost of supply to the housing market, its longevity and heterogeneity, fixed location, the possibility of attracting loans on the security of housing and the presence of a well-developed secondary market. These characteristics of the real estate market mean that it is a set of highly segmented and marginally connected markets (Iacoviello, 2000).  This allows researchers to apply different models for predicting future market movements and price fluctuations.

The s, Standard Model by Poterba (1984),  categorises the general housing market is  into two separate markets: one for secondary or newly constructed housing, which determines the price, and the second – for future projects, which determines the level of investment. At Equilibrium point there is a balance between suppliers of housing and buyers. This model assumed that future prices are determined by expected return on investments, as increase in return leads to increase in demands and vice versa.

The monetarist approach combines the two models viz: Modigliani’s life cycle model and Tobin’s investment theory. According to this approach, demand is determined by the lifetime resources of buyers. These resources consist of financial wealth, human capital and real assets.   As asset prices decline, lifetime resources are alsodeclining, leading to a general decline in (aggregate?) demand (Meltzer, 1995). It is important to note that this model is based on the assumption that the transfer process takes place in the asset market, where transaction and information costs are  lower than the costs of new investments or changes in production. Asset prices in this model are very sensitive to uncertainty about monetary policy (Iacoviello, 2000).

Effect of Brexit on UK Housing Market

Bernanke and Gertler (1995) suggested the “credit channel” approach. It pays attention to imperfections in the credit market, such as information, enforcement and incentives, questioning frictionless nature of it. Based on this, the approach assumes that credit can be more easily provided to market participants who have a healthy financial position or are able to provide guarantee. This leads to the fact that the cost of pledged property determines the ability of the participant to attract loan financing, hence, influencing its ability to produce and consume. Consequently, housing prices are subject to the influence of extensive macroeconomic forces (Iacoviello, 2000).

These models were studied in a variety of empirical research. Results of study by Englund and Ioannides (1997), which was based on a panel data of 15 OECD countries, showed a positive impact of GDP on housing prices. Raising interest rate was found to have a negative impact on prices, contesting the assumptions of the above-mentioned models that house prices repeat the dynamics of macroeconomic shocks.

Important contribution to the house priceforecasting were made by Himmelberg et al. (2005) and Brunnermeier and Julliard (2008). According to Himmelberg et al. (2005), decreasing long-term trends in real interest rates lead to lower mortgage rates, which, in turn, increase demand and push prices up. However, their analysis of 46 Metropolitan Areas in the US for the period 1980 and 2004 revealed that taxes, expected inflation and house price appreciation could influence house-pricing model. Moreover, changes in these fundamental variables could affect locations differently. In areas with inelastic supply, housing prices react to changes in interest rates more acute. These conclusions were supported in the recent studies by El-Montasser et al. (2016) and Martins et al. (2015).

Brunnermeier and Julliard (2008) based their study of house prices on the money illusion theory. Based on the work of Modigliani and Cohn (1979) they concluded that decreasing nominal interest rates and inflation have a positive effect on prices. However, the authors argued that declining trend in these variables leads to mispricing, because customers confuse the reduction of inflation with a decrease in the real interest rate and, thus, underestimate the real value of future payments for mortgages. Cameron et al. (2006) argued that this mispricing could justify a large and significant share of ups in the US and the UK housing market.

Rapach and Strauss (2007) studied the ability of selected macroeconomic variables, including housing rent-to-price ratio, consumer confidence, unemployment and inflation rates to forecast the growth in real estate prices for seven separate states in the US. Results showed that while this set of variables was able to provide an accurate forecast, the influence of single variable was statistically insignificant. The authors concluded that it is difficult to categorise a small set of variables that are best fit the forecasting model of the growth in house prices for a given territory and a period. These findings were supported by Campbell et al. (2009), who also found the aggregate influence of these variables on housing prices, but failed to identify a small set of variables.

Objectives and Methodology of the Study

Expanding the set of “traditional” macroeconomic variables, Sa (2015) analysed the impact of immigration on the housing market in the UK.  Sa looked at information for 159 local areas across the UK on housing prices and immigration andfound that net immigrant inflow has a significant and negative effect of prices. A 1% increase in immigrant inflow leads to decrease in housing prices by 1.6%. The negative relationship between housing prices and net inflow of immigrants was also confirmed in the later study by Wadsworth et al. (2016).

While researchers have come to a relative consensus on the main variables to incorporate the model, the choice of the model itself remains a contentious issue. There is a discrepancy in opinions on whether large cross section factor models are preferable to traditional econometric models, such as vector autoregression (VAR) models. Cristadoro et al. (2005), Giannone and Matheson (2007), Gupta et al. (2009) and Van Nieuwenhuyze (2006) showed that the use of sophisticated models with a large set of variables leads to an improvement in the forecasting of the effectiveness of macroeconomic determinants. On the other hand, results by Schumacher (2007) and Gosselin and Tkacz (2001) showed insignificant or minor improvements in forecasting accuracy. However, this study applies commonly used VAR model despite its limitations. The pros and cons of Vector Auto Regressive model will be discussed in the following chapters.

This section describes the theoretical relationship between Brexit consequences and expected changes in the UK housing market on the basis of which hypotheses are developed.

It is widely argued that the Brexit could contribute to a decline in housing prices and this trend could continue long after Brexit (Hunt and Wheeler, 2017). According to the research by the Royal Institution of Chartered Surveyors, it was expected that housing prices would fall across the UK shortly after the referendum. Thus, Hypothesis 1 proposes:

Hypothesis 1: Short-term Brexit consequences will lead to lower prices in the UK property market.

Nevertheless, the fall in prices is expected only during the first 12 months. This is in line with the Breinlich et al. (2016) prediction that net EU migration will stimulate the growth of housing prices, but with a slight delay. Hence, the second hypothesis is as follows:

Hypothesis 2: In the long term, the UK’s exit from the EU will be positively associated with a growth  in housing prices.

This study applies the vector autoregressive (VAR) framework for the econometric analysis. VAR is a dynamic time series model that allows for data, to determine the dynamic structure of the model (Bafoe-Bonnie, 1998).

Theoretical Models for Predicting Future Market Movements

The VAR approach has minimal theoretical requirements (assumptions?) for the model’s structure and uses a collective lag for each variable in all equations. This technique principally involves defining a set of exogenous and endogenous variables that are assumed to interrelate and, thus, should be included as part of the modelled economic system. In addition, it is necessary to determine the highest possible number of lags required to seize maximum of the effects that these variables have on each other (Bafoe-Bonnie, 1998).

Where,  represents a (n x 1) vector of forecasted endogenous variables; A₀ represents an (n x 1) vector of constant terms; A(L) represents a (n x n) polynomial matrix in the backshift operator L with the length of lag equal p, and  represent a (n x 1) vector of error terms.

To estimate the model, annual data for the period from 1975 to 2016 is used. The variables in the VAR model are the Net Migration, Interest Rates, Effective Exchange Rate, GDP per capita (GDPpc), unemployment rate (Unemployment), the total value of stock market as % of GDP (Stock Market) and house prices (HP). The time series plots of these variables are provided in the appendices section, Appendix A to Appendix G.

The statistic about the UK net migration was retrieved from the Migration Watch UK (MWUK, 2018). Information regarding interest rates and effective exchange rate is taken from the Bank of England (2018). GDP per capita, unemployment rate and the total value of stock market are sourced from the World Bank (2018) World Development Indicators (WDI). Data series on the UK housing prices are sourced from the Nationwide Building Society (2018). The descriptive statistics and the unit roots tests of the selected variables are presented in the following chapter.

Raw data cannot be used due to the presence of Unit Roots and other factors.   In order to determine e if a transformation is needed, the descriptive statistics and the Augmented Dickey Fuller (ADF) unit root tests have been run. The following table provides the descriptive statistics of the raw variables in levels, without any transformations.

Table 1: Descriptive Statistics in Levels

Variable	Obs	Mean	Std. Dev.	Min	Max
Net Migration	42	106,524	119,388	-79,000	332,000
Interest Rates	42	7.06	4.53	0.40	16.30
Effective Exchange Rate	37	94.71	9.50	79.94	116.08
GDPpc	42	20,299	4,742	12,653	26,925
Unemployment	42	7.36	2.27	4.00	11.51
Stock Market	40	52.66	40.02	5.83	132.25
HP	42	140,864	52,610	76,637	242,475

The average house prices in the UK, based on the data provided by Nationwide (2018), were £140,864 during the period from 1975 until 2016. The prices ranged from £76,637 to £242,475 with a standard deviation of £52,610. There has been a long-term positive trend in house prices. In light of Brexit, it  was hypothesied that house prices could react to changes in the net migration flows, unemployment, exchange rates, economic growth in the country and even stock market performance, since the latter is an alternative option for allocating wealth. In addition to this, house prices are expected to be driven by interest rates as they determine the cost of acquiring new property. The average official Bank of England short term interest rates in the UK were 7.06% in 1975-2016 but they fell as low as 0.4% as a response to the financial crisis. Effective Exchange Rate index had an average value of 94.71 and the average GDP per capita was £20,299. The UK had moderate unemployment rates with an average value of 7.36%. The lowest unemployment was observed in 1975 when it constituted only 4%. The value of all stock traded in the UK stock market and measured as percentage of GDP was on average 52.66% but in the periods of boom, this value exceeded 130%. This is one of the most volatile indicators in the dataset. Lastly, the average net migration to the UK was positive in 1975-2016, which indicates that more people  migrated to the UK as compared to the people the people that emigrated. Negative net migration was only observed in the late 1970s – early 1980s and early 1990s. Since 1994, net migration has been positive in the UK.

Empirical Studies on UK Housing Market and Macroeconomic Variables

The results of the ADF unit root tests applied to the raw data in levels show that none of the series are stationary.

Table 2: ADF Unit Root Tests

Variable	Level	1st Difference
Net Migration	-1.158	-5.471***
Interest Rates	-1.028	-5.161***
Effective Exchange Rate	-2.131	-4.604***
GDPpc	-0.903	-3.642***
Unemployment	-1.638	-3.81***
Stock Market	-1.52	-6.115***
HP	-0.36	-3.366**

ies are integrated of the first order because they have become stationary after conversion to first differences. It is valid to note that some series such as GDPpc, Effective Exchange Rate, and House Prices were differenced using the log transformation to achieve log normal distribution of variables and smoothen the series. However, other series that had negative values or were presented originally in the form of percentages could not be differenced using this method as it is mathematically impossible. Therefore, the first differences for such variables were estimated without log conversion. The following table presents the descriptive statistics for the transformed variables.

Variable	Observations	Mean	Std. Dev.	Min	Max
dNetMigration	41	7,049	42,229	-84,000	109,000
dInterestRates	41	-0.27	1.80	-4.04	4.62
dlogEER	36	-0.94	5.44	-12.93	13.95
dlogGDP	41	1.84	1.98	-5.18	5.40
dUnemployment	41	0.02	1.02	-2.01	3.60
dStockMarket	39	1.64	18.42	-57.81	38.52
dlogHP	41	2.17	7.93	-15.34	16.45

The prefix “d” means that the series is differenced. The prefix “dlog” means that the series is log differenced. Log differences are useful because they are equivalent to growth rates. Thus, the average growth rate in house prices in the UK was 2.17% per annum and the average growth rate in GDP per capita was 1.84% per annum. House prices were growing faster than the economy because of the periods of property booms in the country. This tendency was also observed in the stock market, which was growing in the long term. Interest rates on average were falling during the period from 1975 until 2016, and the British currency was losing its value at an average rate of -0.94% per annum. Net migration flows were rising on average by 7,049 people each year. Unemployment rate, at the same time, was also growing at an average rate of 0.02% per annum. Thus, one could assume that the growth in immigrants could possibly contribute to the gradual growth in unemployment. However, this assumption still needs to be confirmed using a VAR model because descriptive statistics do not show relationships between variables.  

Once the series differenced and turned into stationary form, the next step of the VAR analysis is to choose the optimal number of lags to include. Very few lags could lead to such problems as inefficiency of the estimated coefficients. The relationships would be unreliable.  Inclusion of too many lags, however, could also become a problem because too many degrees of freedom would be consumed and the model could be overfitted. In order to make a rational choice, information criteria have been used in Stata as shown in the following table.

The results of HQIC and SBIC suggest that 0 lags should be used, which is not acceptable for a VAR model. This choice was rejected. FPE suggests that 4 lags should be used but two more tests are consistent in making a suggestion that 3 lags are enough. Since annual data are used and AIC and LR provide consistent results, the final choice is 3 lags.

The VAR output is depicted in the table above.The results evidence positive autocorrelation in house price growth rates. This suggests that past price changes affect current price changes, reinforcing the latter. There is also evidence of a negative effect of lagged interest rates on house prices. This is consistent with the theory and expectations that higher interest rates increase the cost of acquiring property and discount rates used in valuation of property. Therefore, the relationship is negative. In line with expectations, net migration is found to be a positive driver of house prices but the significant effect is observed only at the third lag. This can be explained by the long time that it takes for new immigrants to finally buy a home in the country. Stock market, unemployment and exchange rates appear to have insignificant relationships with house prices. The effect of GDP per capita growth is rather ambiguous and requires further analysis, which can be done using the Granger causality test. When there are many lags, the coefficients in the VAR system could be difficult to interpret. Therefore, joint significance of the lagged values is detected by the Granger Causality Test reported in the next table.

The results confirm that in the equation where log differenced house prices are used as the dependent variable, the effect of other variables is jointly significant. Thus, the VAR model can be used for the forecasting of house prices. However, exchange rates, unemployment and stock market do not produce a significant influence on house prices. The effect of GDP per capita is also statistically insignificant at the conventional 5% level. Thus, the ambiguity in relation to the effect of GDP in the raw output of the VAR model has been resolved. House prices are ultimately driven by migration, interest rates and their own lagged values.

Impulse Response Functions (IRF) have been constructed to trace how house prices will react to one standard deviation shocks in net migration and interest rates.

In line with the original interpretation of the VAR, a positive shock in net migration will stimulate growth in house prices but this will not happen immediately. It will take up to 3 years until this effect can be fully reflected in growth in house prices. At the same time, a shock such as an increase in interest rates will lead to a much quicker response in the housing market. The growth in house prices will demonstrate a fall within the first year of the interest rate hike. However, a gradual correction would follow as market participants and households will adjust to the new conditions and monetary policy environment. Within 5 years, the growth in house prices would tend to stabilize completely from the effects of the shock (other things being the same) 

Before making a five year out of sample prediction of house, it is important to look at how well the VAR model produces in-sample forecasts. This can be done by comparing the predicted and actual values of house price growth rates during the evaluation period of 1990 until 2016.

The model provides accurate forecasts during the first decade between 1990 and 2000. After this, the forecast was able to predict general trend of long-term changes in the housing market, but it could not capture short term deviations. The deterioration of the quality of the forecast in the second and third decade can be explained by the dynamic nature of the prediction. The starting point of the forecast is the actual value of house price growth rates observed in 1990. However, the forecast for each subsequent year is made on the basis of previously forecasted values rather than actual values. The greater the distance of the forecast year from the starting point, the lower was the quality of predictions.  However, the estimations made by the model  had a good fit during the first ten years.  It is expected that it will also produce a rather accurate forecast for the five year ahead prediction. This out of sample forecast is provided in the next figure.

During the next five years, in spite of the Brexit referendum and possible economic consequences, the housing market in the UK is expected to grow each year at a rate of up to 4% by the end of the fifth year. To show how these predicted growth rates will be reflected in actual average house prices in the UK, an exponentiation procedure is performed to convert the log values into British pounds.

As the next figure demonstrates the house prices in the UK are not expected to experience a decline due to Brexit.

Moreover, the house prices are predicted to reach their historical highs. The next chapter will discuss these results comparing them to literature, critique the method and provide policy implications.

The results of VAR estimations showed that the house price dynamic could be forecasted by using a multivariate simultaneous equations framework. These results support previous research by Wheaton and Nechayev (2008). The authors argued that the monetary policy has a lagged negative effect on the housing market as higher interest rates increase acquiring and transaction costs and decrease net present value of property. This conclusion was proved in this study. Moreover, the statistical insignificance of GDPpc growth rate, unemployment and exchange rates, which was highlighted in the works of Wheaton and Nechayev (2008) and Gupta et al. (2011), was also confirmed.

While the majority of conclusions are in line with previous empirical literature on this topic, there is a little inconsistence regarding the role of net migration. In this study, it was found that net migration has a lagged significant positive effect on house prices. However, several authors, such as Breinlich et al. (2016) and Sa (2015), did not find a significant correlation between net migration and local house prices. They argued that the increase in migrants adds pressure on the UK housing market, but the main problem lies in the inefficient development of infrastructure and insufficient volume of housing construction. Thus, with migrants from the EU or without, this problem will remain relevant.

With regard to the forecast of future house price movements, the analysis showed that the house prices would continue to grow in the future, despite Brexit. Therefore, Hypothesis 1 of a short-term market decline should be rejected. This contradicts with recent research by Pedrick (2016), Royal Institution of Chartered Surveyors (Hunt and Wheeler, 2017) and Lea (2016), who expect short-term deterioration in house prices. On the other hand, Hypothesis 2 of a long-term growth is accepted, which is in line with the above-mentioned research.

The applied VAR method has several limitations that could affect the results of this study. This method usually applies an equal length of lags for all variables, which implies that the observer has to evaluate many parameters, including those that are statistically insignificant (Iacoviello, 2000). This problem of over-parameterisation could create multicollinearity and adversely affect degrees of freedom, which leads to inefficient estimates and increases the likelihood of prediction errors outside the sample (Iacoviello, 2000).

In addition to this, all errors in measuring or deflecting the model will also cause unexplained information left in the violation conditions, which makes interpreting impulse responses even more difficult. Nevertheless, this does not imply that these responses are unusable, but rather emphasises the importance of cautious analysis of empirical results that should be used when specifying the VAR model. According to Bjornland (2000), if it is necessary to draw conclusions about the characteristics of the fundamental information generating process based on the analysis of the impulse response of the estimated VAR The success of a model would be confirmed only if the model id harmonious enough to include competing model and should not change considerably with an expansion in the data used.

Nevertheless, the susceptibility of impulse responses to missed variables and all types of error specifications should warn the observer of excessive interpretation of data from VAR models.  Specialattention should be paid to checking for dynamic misspecifications when applying the VAR model. In addition, each model should be identified using tight or loose economic theories. By including other relevant variables, it can be checked whether impulse responses correspond to invariants to changes in model specifications (Bjornland, 2000). However, there are more cautions in applying the VAR model to house prices.

According to Iacoviello (2000), any simple VAR model  takes into account 100% variance of variables by unpredictable movements in endogenous variables. The search for exogenous turbulences is a way to search for technology, policy or beliefs-caused changes this variable. However, it is not able to provide a detailed analysis of the effect systematic policies or perfectly anticipated shocks. Moreover, it is also problematic to distinguish between the variables as the fundamental and non-fundamental determining factors of housing prices, for instance, speculative bubbles.

Another crucial issue is linearity. Housing prices and other variables could react differently to external shocks that are equal in magnitude, but are in the opposite direction. Iacoviello (2000) argued that that in the housing market, which is very speculative in nature, the reaction or the dynamic of house price to exogenous variables, such as interest rates are non-constant and non-linear over time.

It is widely believed that expansionary monetary policy is the cause of dramatic changes in asset prices (Iacoviello, 2002). In this study, it was found that such monetary policy has a negative impact on housing prices. Iacoviello (2002) argued that this is due to internal characteristics in the UK, such as low transaction costs, reviewable mortgage rate and high owner occupancy rate. Therefore, stable monetary policy could save the housing market from shocks on the one hand, enabling smooth regulation of prices through the interest rate on the other hand. At the same time, Brexit is not expected to have serious implications for housingprices. Thus, policy makers should focus on fundamental determinants of house prices to avoid excess volatility in the market.

Conclusion

This dissertation has explored what is expected to happen to the UK housing market in the next five years with the UK negotiating and implementing its withdrawal from the European Union, known as the Brexit. This study is based on quantitative research methods and applies VAR model to forecast future house prices in the UK. Estimations of VAR model are based on historical data for the period 1975-2016.

The first objective of this study was to identify the set of determinants that cause fluctuations in house prices. The review of empirical literature showed the great agreement among scientists on the importance of macroeconomic factors in predicting movements in the housing market. The most commonly used variables were real or nominal interest rate, effective exchange rate, and economic growth proxied by per capita GDP. However, it was concluded in various studies that while the large set of variables is able to provide a relatively accurate forecast, a small set or individual macroeconomic variables are not able to provide anacceptable prediction accuracy within VAR models. To overcome this deficiency, it was decided to include demographic variables in the model such as unemployment rate and net migration. It is assumed that they will also be significantly affected by the forthcoming withdrawal of the UK from the EU. The analysis showed that all selected determinants have significant influence on house prices in the country.

The second objective was to forecast the movement of house prices in the following 5 years. The analysis of existing research showed that researchers expect a short-term drop in prices for 12 months, followed by a prolonged recovery. Based on this, it was decided to consider a forecast for the next five years. The results of forecasting showed that despite the negative impact of Brexit, housing prices would continue to grow in the next 5 years.

Similar to other studies on this topic, the conclusions in this paper have some limitations. The results of the analyses of this study are based on aggregate data for the UK and cannot be applied to specific regions or metropolitan areas. For instance, house prices in Wales or Great London could have other dynamics in the future, different from the forecasts of this study. Other limitations relate to the VAR model. Among them are over-parameterisation, dynamic misspecifications and linearity. However, these limitations did not influence results of this study significantly.

Moreover, some findings in this paper are in conflict with previous research, which gives a favourable basis for further research. It was found that positive net migration has a positive effect on house prices, while Sa (2015) argued that a positive influence of inflow of migrants is offset by more negative influence of outflow of local population, which is usually more solvent. Along with the fact that the researchers did not manage to come up with a single model for forecasting house prices that would be universally applicable, further research on this topic will contribute to better understanding of the housing market.

References

Baffoe-Bonnie, J. (1998) The dynamic impact of macroeconomic aggregates on housing prices and stock of houses: a national and regional analysis. The Journal of Real Estate Finance and Economics. 17(2). pp. 179-197.

Bank of England (2018) Statistical Interactive Database – interest & exchange rates data. [Online] Available at: https://www.bankofengland.co.uk/boeapps/iadb/index.asp?first=yes&Section Required=I&HideNums=-1&ExtraInfo=true&Travel=NIxIRx, [Accessed: March 27, 2018].

Bernanke, B. and Gertler, M. (1995) Inside the Black Box: The Credit Channel of Monetary Policy Transmission. Journal of Economic Perspectives. 9(4). pp. 27-48.

Bjornland, H. (2000) VAR Models in Macroeconomic Research. Statistics Norway Research Department. 14. pp. 1-31.

Breinlich, H., Dhingra, S., Ottaviano, G., Sampson, T., Van Reenen, J. and Wadsworth, J. (2016) BREXIT 2016: policy analysis from the Centre for Economic Performance. CEP, LSE.

Brunnermeier, M. and Julliard, C. (2008) Money illusion and housing frenzies. The Review of Financial Studies. 21(1). pp. 135-180.

Cameron, G., Muellbauer, J. and Murphy, A. (2006) Was there a British house price bubble? Evidence from a regional panel. CEPR Discussion Paper No. 5619.

Campbell, S., Davis, M., Gallin, J. and Martin, R. (2009) What moves housing markets: A variance decomposition of the rent–price ratio. Journal of Urban Economics. 66(2). pp. 90-102.

Cristadoro, R., Forni, M., Reichlin, L. and Veronese, G. (2005) A Core Inflation Indicator for the Euro Area. Journal of Money, Credit and Banking. 37(3). pp. 539–560.

El-Montasser, G., Ajmi, A., Chang, T., Simo-Kengne, B., André, C. and Gupta, R. (2016) Cross-Country Evidence on the Causal Relationship between Policy Uncertainty and Housing Prices. Journal of Housing Research. 25(2). pp. 195-211.

Englund, P. and Ioannides, Y. (1997) House Price Dynamics: an International Empirical Perspective. Journal of Housing Economics. 6. pp. 119-136.

Giannone, D. and Matheson, T. (2007) A New Core Inflation Indicator for New Zealand. International Journal of Central Banking. 3(4). pp. 145-180.

Gosselin, M. and Tkacz, G. (2001) Evaluating Factor Models: An Application to Forecasting Inflation in Canada. Working Paper No. 18 Bank of Canada.

Gupta, R., Kabundi, A. and Miller S. (2009) Using Large Data Sets to Forecast Housing Prices: A Case Study of Twenty US States. Journal of housing research. 20(2). pp. 161-190.

Gupta, R., Kabundi, A. and Miller, S. (2011) Forecasting the US real house price index: Structural and non-structural models with and without fundamentals. Economic Modelling. 28(4). pp. 2013-2021.

Himmelberg, C., Mayer, C. and Sinai, T. (2005) Assessing High House Prices: Bubbles, Fundamentals and Misperceptions. Journal of Economic Perspectives. 19(4). pp. 67-92.

Hunt, A. and Wheeler, B. (2017).Brexit: All you need to know about the UK leaving the EU. BBC News. [Online] Available at: https://www.rexsresources.com/uploads/6/5/2/1/6521405/brexit_ _all_you_need_to_know_about_the_uk_leaving_the_eu_-_bbc_news.pdf, [Accessed: March 27, 2018].

Iacoviello, M. (2000) House prices and the macroeconomy in Europe: Results from a structural VAR analysis. ECB Working Paper, No. 18.

Iacoviello, M. (2002) House prices and business cycles in Europe: A VAR analysis. Working Papers in Economics. 81. pp. 1-43.

Kenny, G. (1998) The Housing Market and the Macroeconomy: Evidence from Ireland. Bank of Ireland Working Paper.

Lea, R. (2016) The Brexit vote has had no major economic impact so far, according to the ONS. Arbuthnot banking group. 26. pp. 1-14.

Martins, A., Serra, A., Martins, F. and Stevenson, S. (2015) EU housing markets: the role of institutional factors. Working Papers in Real Estate & Planning. 04/15. pp 1-42.

Meltzer, A. (1995) Monetary, Credit and (Other) Transmission Processes: A Monetarist Perspective. Journal of Economic Perspectives. 9(4). pp. 49-72.

Miles, D. (1995) Housing, financial markets and the wider economy. New York: John Wiley and Sons.

Modigliani, F. and Cohn, R. (1979) Inflation, Rational Valuation and the Market. Financial Analysts Journal. 37(3). pp. 24-44.

MWUK (2018) Net Migration Statistics. [Online] Available at: https://www.migrationwatchuk.org/statistics-net-migration-statistics, [Accessed: March 27, 2018].

Nationwide (2018) House Prices in the UK, Available at: https://www.nationwide.co.uk/about/house-price-index/download-data#tab:Downloaddata [Accessed on 24 March 2018].

Nationwide Building Society (2018) Download house price index data.  [Online]                                                                            Available at: https://www.nationwide.co.uk/about/house-price-index/download-data#tab:Downloaddata, [Accessed: March 27, 2018].

ONS (2017) An international comparison of gross fixed capital formation. [Online]                                                                            Available at: https://www.ons.gov.uk/economy/grossdomesticproductgdp/articles/aninternational comparisonofgrossfixedcapitalformation/2017-11-02, [Accessed: March 27, 2018].

Pedrick, L. (2016) The business impact of the Brexit vote: a view from housing associations. LSE Brexit.

Poterba, J. (1984) Tax Subsidies to Owner-Occupied Housing: An Asset Market Ap- proach. Quarterly Journal of Economics. 99. pp. 729-752.

Quigley, J. (1998) Housing Markets. in J. Eatwell, M. Milgate and P. Newman (eds.), The New Palgrave:  A Dictionary of Economics. pp. 3-20, London: Macmillan Press.

Rapach, D. and Strauss, J. (2007) Forecasting real housing price growth in the eighth district states. Federal Reserve Bank of St. Louis. Regional Economic Development. 3(2). pp. 33-42.

Sa, F. (2015) Immigration and House Prices in the UK. Economic Journal. 125. pp. 1393-1424.

Schumacher, C., (2007) Forecasting German GDP Using Alternative Factor Models Based on Large Datasets. Journal of Forecasting. 26(4). pp. 271–302.

Sims, C. (1980). Macroeconomics and Reality. Econometrica. 48(1). pp. 1-48.

Van Nieuwenhuyze, C. (2006) A Generalized Dynamic Factor Model for the Belgian Economy Identification of the Business Cycle and GDP Growth Forecasts. Journal of Business Cycle Measurement and Analysis. 2. pp. 213–248.

Wadsworth, J., Dhingra, S., Ottaviano, G. and Van Reenen, J. (2016) Brexit and the Impact of Immigration on the UK. CEP Brexit Analysis. (5). pp. 34-53.

Wheaton, W. and Nechayev, G. (2008) The 1998-2005 Housing “Bubble” and the current “Correction”: What’s Different This Time? Journal of real Estate research. 30(1). pp. 1-26.

World Bank (2018) World Development Indicators. [Online]                                                                            Available at: https://data.worldbank.org/data-catalog/world-development-indicators,                                                [Accessed: March 27, 2018].