Written by
Karen Høvring Maiken Ludt Parmo
Master’s thesis, May 2016
Copenhagen Business School
MSc in Finance and Strategic Management
Supervisor
Jørgen Bo Andersen
Characters: 203 002 (113 pages)
AN EMPIRICAL ANALYSIS OF THE NORWEGIAN HOUSING MARKET
Are there indications pointing towards a housing bubble in the Norwegian market?
Abstract
In this thesis, we aim to provide whether there are bubble tendencies in the Norwegian housing market, or if the price growth can be explained by fundamental factors. Moreover, we will investigate if our provided models underline the formation of the housing bubble in Denmark that occurred during the years before the financial crisis in 2008. After approximately 40 years with a relatively synchronized development in house prices in Norway and Denmark, Danish house prices evolved differently after 2008. Norwegian house prices picked up relatively fast, whereas Danish house prices continued to fall – which subsequently has proven to be a result of a bubble burst in Denmark. Accordingly, we argue that our findings for Norway from the provided models will be more reliable if the similar models show vital signs of the bubble formation in Denmark before the burst. Additionally, Norway and Denmark are two quite similar countries in terms of demographics, governance and traditions, which we believe make the analyses more comparable and valid.
To get an understanding of how bubbles are formed and cracks we have adopted Hyman Minsky's model. When considering factors that drive house prices, we have been looking at macroeconomic variables in light of Jacobsen and Naug’s article “what drives housing prices?”. Case & Shiller’s seven criteria for a housing bubble are discussed in relation to the Norwegian housing market today, and to the Danish market prior to the financial crisis, while the empirical analysis is conducted on the basis of a Hodrick-Prescott filter and the Price-to-Rent method. Throughout the thesis we have been aware of the fact that we will be unable to find definite and concrete answers to our research question, but we will emphasize if house prices are overvalued based on our fundamental and empirical analyses. It is both challenging and problematic to reveal a housing bubble with certainty before a potential bubble burst, since what will happen ahead of time is unknown.
The results which emerge in this thesis gives us reason to argue for overpriced house prices in the Norwegian housing market, which provides indications of the existence of a housing bubble.
Table of contest
1 INTRODUCTION ... 5
1.1 Introduction and problem statement ... 5
1.2 Structure of the paper ... 8
1.3 Delimitations ... 10
2 METHODOLOGY ... 13
3 HOUSING PRICES AND HOUSING BUBBLES ... 15
3.1 Definition of a housing bubble ... 15
3.2 Why housing prices are rising, housing bubbles occur and burst ... 17
3.2.1 Hyman Minsky´s model ... 17
3.2.2 Case and Shiller ... 20
3.3 Demand and supply for the housing market ... 21
3.3.1 Demand ... 23
3.3.2 Supply ... 27
3.4 Price determination and equilibrium in the short- and long-term ... 29
3.4.1 The short-term ... 30
3.4.2 The long-term ... 31
3.5 Fundamental factors (J&N) ... 33
3.5.1 Interest rate ... 35
3.5.2 Unemployment ... 35
3.5.3 Real income ... 35
3.5.4 New constructions ... 36
4 HISTORICAL DEVELOPMENT ... 37
4.1 The Norwegian housing market ... 37
4.2 The Danish housing market ... 46
5 COMPARATIVE FUNDAMENTAL ANALYSIS ... 50
5.1 Development in real house prices ... 50
5.2 Development in Gross Domestic Product (GDP) ... 51
5.3 Development in key interest rate ... 52
5.4 Development in interest rate (lending rate) ... 54
5.5 Development in disposable income ... 55
5.6 Development in the unemployment rate ... 56
5.7 Development in new constructions ... 58
5.8 Conclusion comparative fundamental analysis ... 60
6 ANALYSIS OF CASE AND SHILLER´S HOUSING BUBBLE CRITERIA ... 62
6.1 Norway ... 62
6.1.1 Conclusion of Case and Shiller´s criteria in Norway ... 70
6.2 Denmark ... 71
6.2.1 Conclusion of Case and Shiller’s criteria in Denmark ... 76
6.3 Other factors ... 77
6.3.1 Developments in the debt level ... 77
6.3.2 Number of unsold homes ... 80
6.3.3 Financing ... 81
7 EMPIRICAL ANALYSES ... 82
7.1 Hodrick-Prescott filter ... 82
7.1.1 Theoretical framework ... 82
7.1.2 Data Material ... 84
7.1.3 Empirical testing ... 85
7.2 Price/Rent ... 92
7.2.1 Theoretical framework ... 92
7.2.2 Data Material ... 97
7.2.3 Empirical testing ... 100
7.3 Conclusion Empirical Analysis ... 107
8 FINAL CONCLUSION ... 109
Bibliography ... 114
APPENDIX ... 120
Table of figures
Figure 1.1: Real house price index for Norway and Denmark (1992=100) ... 6
Figure 3.1 Minsky´s model for bubble formation ... 18
Figure 3.2: Price determination in the short-term ... 28
Figure 3.3: Housing supply in the short- and long-term ... 32
Figure 5.1: Development in real house prices Denmark and Norway 1992-2015 (Index 1992=100) ... 50
Figure 5.2: Development in GDP in Norway and Denmark 1980-2015 (Index 1980=100) ... 51
Figure 5.3: Development in key interest rate Norway and Denmark 1994 -2015 ... 53
Figure 5.4: Developments in interest rate Norway and Denmark 1994-2015 ... 54
Figure 5.5: Development in real disposable income in Norway and Denmark 1992-2015 (1992=100) ... 55
Figure 5.6: Development in unemployment rate in Norway and Denmark 1980-2015 ... 56
Figure 5.7: Development in unemployment rate and real house prices in Norway 1980-2015 (1980=100) ... 57
Figure 5.8: Development in unemployment rate and real house prices in Denmark 1992-2014 (1992=100) 58 Figure 5.9: Development in new constructions Norway 1983-2015 ... 59
Figure 5.10: Development in new constructions Denmark 1980-2015 ... 59
Figure 6.1: Amount of commenced dwellings in Norway 1983-2015 ... 64
Figure 6.2: Construction costs and house price index Norway 1978-2016 (2000=100) ... 64
Figure 6.3: Media that concerns house prices in Norway (1987-2015) ... 65
Figure 6.4: Media that concerns housing bubbles in Norway (2002-2015) ... 66
Figure 6.5: House price- and disposable income index Norway (1980=100) ... 67
Figure 6.6: Forced sales in Norway 2008-2015 ... 69
Figure 6.7: Started dwellings in Denmark 1978-2016 (Index, 2000=100) ... 72
Figure 6.8: Construction costs- and house price index Denmark (2003=100) ... 72
Figure 6.9: Media attention that concerns house market in Denmark (1990,2001,2003,2006) ... 73
Figure 6.10: Media attention that concerns house prices in Denmark (1990,2001,2003,2006) ... 73
Figure 6.11: House price- and disposable income index Denmark (2000=100) ... 74
Figure 6.12: Forced sales in Denmark 2000-2015 ... 75
Figure 6.13: Development in household´s foreign and domestic debt in Norway 1985-2016 ... 77
Figure 6.14: Development in total credit growth and house prices in Norway 1985-2015 (1985=100) ... 78
Figure 6.15: Development in debt/disposable income ratio in Norway 1996-2015 ... 79
Figure 6.16: Unsold homes in Norway 2011-2016 ... 80
Figure 7.1: Time series from 1980-2015 for real house price with HP-filter, Norway (1980=100). Shows lambda value of 100 and 2 500. ... 85
Figure 7.2: Time series from 1819-2015 for real house price index with HP-filter, Norway (1980=100). Shows lambda value of 100 and 2 500 ... 87
Figure 7.3: Time series from 1819-2015 for real house price index with HP-filter, Norway (1980=100). Shows lambda value of 2 500 ... 88
Figure 7.4: Gap from trend 1980-2015 in Norway. Shows lambda value of 2 500 ... 89
Figure 7.5: Gap from trend 1819-2015 in Norway. Shows lambda value of 2 500 ... 89
Figure 7.6: Time series from 1992-2015 for real house price index with HP-filter, Denmark (1992=100). Shows lambda value of 100. ... 90
Figure 7.7: Time series from 1992-2015 for real house price index with HP-filter, Denmark (1992=100). Shows lambda value of 100 and 2 500. ... 91
Figure 7.8: Gap from trend 1992-2015 in Denmark. Shows lambda value equal to 2 500. ... 92
Figure 7.9: P/R rates in Norway 1871-2015 ... 101
Figure 7.10: P/R with and exponential trend line 1980-2016 in Norway ... 102
Figure 7.11: Real- and fundamental P/R rations in Norway ... 105
List of tables Table 5.1: Conclusion comparative fundamental analysis ... 60
Table 6.1: Conclusion Case and Shiller Norway ... 70
Table 6.2: Conclusion Case and Shiller Denmark ... 76
Table 7.1: Real P/R ratios in selected years in Norway ... 103
Table 7.2: Calculations of fundamental P/R ratios in Norway ... 105
1 INTRODUCTION
1.1 Introduction and problem statement
The last 24 years have appeared to be a continuous expansion period in the Norwegian housing market, with exception from fluctuations during the financial crisis. Norwegian house prices dropped in the wake of the financial crisis in 2008, but in 2009, Norway experienced a period of recovery. This brought prices far above the peak level, which have had a significant impact for most households, since housing is one of the largest consumer components, as well as the largest investment over a lifetime for a majority of us. Housing investments, in addition to expected value appreciation, is often regarded as the most important saving method. Hence, developments in house prices are of a big concern for a majority of households, and have major impact on personal finances. We therefore find it interesting to investigate the underlying factors of the excessive growth in Norwegian house prices.
Norway has had a significant growth in house prices since the 1990s. The last recession in house prices occurred in 2008, as a result of the extensive financial crisis that affected the world economy.
Today, house prices in Norway are higher than ever. In sharp contrast to developments in Norway, there has been a significant decline in Danish house prices from 2007 until late 2011.
The house price development in Denmark was similar to the Norwegian market until 2006. During the financial crisis, house prices dropped in real terms by 22 % in Denmark, and by 7.73 % in Norway. Moreover, Norwegian house prices have increased by 27 % in real terms since 2008, while the Danish housing market has not fully recovered. Figure 1.1 below illustrates the development in Norwegian and Danish real house price index from 1992 to 2015:
Figure 1.1: Real house price index for Norway and Denmark (1992=100)
As figure 1.1 illustrates, developments in real house prices in Denmark and Norway had a similar trend until the financial crisis. At this point, Norwegian and Danish house prices started to move in different directions. Danish house prices had a significant increase during the beginning of the 2000s, and later it has been proven that Denmark experienced a housing bubble that burst in 2005/2006 (Dam et.al., 2011). When the bubble burst, there were subsequent dramatic price drops, as the prices were far above their fundamental level. As the housing bubble in Denmark by now is a fact, we know the underlying explanation of the different direction in the development of house prices in Denmark and Norway after the financial crisis.
It is interesting to investigate whether the Norwegian economy has indications of a housing bubble today or if house prices can be supported by underlying fundamental economic conditions. To emphasize the importance of potential deviations from fundamental factors when investigating housing bubbles, we will additionally examine the Danish housing market prior to the financial crisis. Moreover, we will compare the Norwegian economy today, with factors that were present (or absent) in the Danish economy prior to the bubble burst – in order to get a deeper understanding of the mechanisms of a housing bubble. Denmark has about the same population as Norway, and has
many similarities regarding demographics, governance and traditions. This makes Denmark an interesting basis for comparison.
Problem statement
Based on this, we have chosen the following problem statement:
To answer this question, we will investigate two sub-questions;
• Can the high growth in house prices in Norway be seen as abnormal, or can it be explained by underlying fundamental factors?
• Can we prove the Danish housing bubble prior to the financial crisis? If this is the case, we can argue that our applied models and theories may be valid for the Norwegian market as well.
There is now an ongoing heated debate about whether the rise in house prices can be justified by distinctive fundamentals, or whether the growth is due to other factor s. Simplified, there exist two main opinions regarding the Norwegian housing market. These opinions are no theories themselves, but they are based on the perspectives and theories regarding how housing bubbles occur.
Opinion 1: People deny the existence of a housing bubble and explain the high house prices as a result of real fundamental factors. This is the dominant opinion among Norwegian economists.
Opinion 2: People argue that the growth in house prices cannot be explained by fundamental factors. This group support that a housing bubble is the explanation behind the high growth in house prices.
“Are there indications pointing towards a housing bubble in the Norwegian market?”
The housing market is complex and driven by economic conditions, as well as expectations and speculations in several variables. It is therefore difficult to say for sure whether there is a housing bubble, before the potential bubble burst. For that reason, we will look into indications pointing towards whether or not a housing bubble exists in the Norwegian market today. We will not look into whether there are housing bubble tendencies in the Danish market today, but analyze characteristics of the market when they were experiencing a housing bubble in 2005/2006.
Furthermore, we will compare our findings in order to get a deeper understanding of a possible bubble formation in the Norwegian housing market.
It is often difficult to determine whether or not there exists a housing bubble until it eventually burst. When we are looking at fundamental factors in the housing market, we consider macroeconomic variables such as the GDP, key interest rates, interest rates (lending rates), unemployment levels, income level and the level of new constructions. A bubble in a national context will inflict serious economic repercussions in the real economy, as for example the financial crisis did in 2008. In the stock market, professional investors and stockbrokers are overrepresented.
They have a strong understanding of the concept of risk and what risk entails. In contradiction, the housing market is overrepresented by players who do not have a good understanding of risk. There is a weak understanding of the concept, and the danger associated with a decline in house price is not known among many actors in the housing market. It is therefore important to reveal indications of a housing bubble before it eventually burst, to curb the economic consequences of a potential crack.
In order to answer our problem statement, we will use well-known housing bubble theories and frameworks. We consider the housing bubble theories as more reliable if they confirm the fact that Denmark was in a housing bubble prior to the financial crisis. We will in the next section present how the thesis is structured, and present the applied models.
1.2 Structure of the paper
Our problem statement will be answered using empirical data and analytical assessments of fluctuations in both the Norwegian and the Danish housing market. The paper is structured as
follows:
We will start our thesis by introducing the methodological approach to inform the reader of the background behind our applied methods and frameworks, in order to obtain reliability and validity throughout the thesis.
In the beginning of chapter 3 we will provide a definition of a housing bubble, by using different definitions given by known economists. Further on, we will discuss two theories and hypothesis regarding why house prices are rising, house bubbles occur and burst. These theories comprise:
• Hyman Minsky’s model
• Case and Shiller
Moreover, the chapter addresses housing market theory, namely the supply and demand side of the housing market. We will explain how prices are determined in the short- and long-term. Finally, we will explain fundamental factors1 in the housing market, where particularly Jacobsen and Naug (2004) is an essential platform for our work.
The Norwegian housing market development is addressed in chapter 4 with a detailed explanation of previous housing bubbles, in light of Minsky’s model and the current situation today. Similarly, we will provide a general explanation for Denmark.
The comparative fundamental analysis in chapter 5 is made with the purpose to investigate whether the developments in house prices can be explained by fundamental factors or if the growth seems to be abnormal. More specific, we will focus on today’s situation for Norway compared to the period in the mid-2000s for Denmark, when they were experienced a housing bubble. We will look at developments in real house prices, the GDP and the key interest rate in Norway and Denmark. In addition, we will analyze the fundamental factors in light of Jacobsen and Naug’s article “What drives housing prices?”.
1 By fundamental factors we will look at economic variables as interest rates, unemployment, real income and new constructions
Chapter 6 addresses Case and Shiller’s seven criteria for a housing bubble based on the article “Is there a bubble in the Housing Market”. We will discuss whether these criteria are met in both the Norwegian housing market today and the Danish housing market prior to the financial crisis.
Additionally, we analyze three other factors we find highly relevant for developments in house prices in Norway.
The empirical analysis in chapter 7 examines the HP-filter and the Price-to-Rent method. We will first introduce the theoretical aspects of the HP-filter. Further on, the HP-filter is applied on the Norwegian and the Danish housing market to determine whether house prices deviate from trend.
As a supplement regarding the Norwegian house market, we will do a P/R analysis in order to investigate the growth in house prices relative to the growth in rental prices. The theoretical aspect of the P/R model will be introduced prior to the analysis.
To provide a comprehensive understanding regarding the Norwegian housing market, we believe the analyses and theories provided will enable us to derive a conclusion. All together this will lead to our final conclusion, which we will present in chapter 8. Here we will gather all threads together.
To make it easier to distinguish between Norwegian and Danish figures, graphs and tables, Norway will consistently be illustrated in blue, whereas Denmark will be illustrated in red. This is done in order to provide the reader with a clear and systematic overview throughout the paper.
1.3 Delimitations
We will in this section go deeper into the thesis delimitations, i.e. what we will limit ourselves to, in addition to what we will restricts ourselves from, in order to provide the reader with a clear understanding of the scope of the paper.
Overall delimitations
We will restrict us from developing any new models or frameworks. The aim of this paper is not to develop a new model for predicting house prices, but rather to provide a descriptive explanation of
or estimate new methods to detect possible housing bubbles. We will limit the paper by using familiar theories to analyze house price developments.
We will limit ourselves to mainly focus on the Norwegian market. As our problem statement indicates, our main focus will be the Norwegian housing market. Analyses regarding the Danish market are made in order to confirm whether our provided models and frameworks can be used to prove the previous housing bubble. If this is the case, the models will be more reliable when it comes to findings regarding the Norwegian market.
We will limit ourselves to look at both countries as a whole, even though a housing bubble most likely will occur in primarily big cities. However, due to difficulties finding satisfactory data material for separate cities, the analysis of the housing market in both Norway and Denmark will be made for the countries as a whole. Therefore, we are looking at figures for fundamental macroeconomic variables for both countries as a whole, and have not focused on geographical variations in the housing market, although this will have a huge impact on the price level.
We will restrict ourselves from distinguishing between different types of housing. The study is limited to examining housing as a whole. Hence, no distinction is made between different housing types, such as detached villas, apartments or townhouses. The terms “dwelling”, “housing” and
“house” is thus referring to all kinds of housing throughout the paper.
Delimitations in section 6: Case and Shiller
We will limit ourselves to only analyze “other factors” for the Norwegian housing market. We will analyze the following factors for Norway: developments in debt level, number of unsold homes, and financing. We find these factors as highly relevant for the current situation in Norway characterized by high growth in house prices.
Delimitations in section 7: Empirical analyses
We will restrict ourselves from conducting a P/R analysis for Denmark. We will not conduct a P/R analysis for Denmark due to limited data access, as we do not have any yearly data of the average
rental price per square meter. We did our best to find usable figures, without any luck. We have searched the statistical ten-year overviews provided by Statistics Denmark from 1964 to 2015, but they have only collected average annual rent for 1981, 1985, 1992, 1995 and 2000. As the latter was the latest figure we could collect, we were not able to draw any conclusions regarding developments in rental market prices during the years prior to the bubble burst. One possible solution could have been to plot in the five figures and to draw a trend line from 1981-2000, and further look at the continued trend - but we did not find this as reasonable to do. Furthermore, the Danish rental market has been, and still is, highly regulated. It would therefore not be reliable to calculate average annual rents forward in time based on these statistics. Additionally, we contacted Statistics Denmark2 and Denmark's National Bank, but they could not assist us. Hence, we will not develop a P/R analysis for Denmark for two reasons:
i. It will not make sense to calculate the P/R ratios for five separate years, as fluctuations will not appear. Moreover, we do not have the annual average rental prices for the period prior- and during the housing bubble in 2005/2006.
ii. It will not make sense to calculate the P/R ratios due to the highly regulated rental market
Hence, we believe that it is sensible to disregard the Danish P/R development in this thesis.
We will limit ourselves to only consider the years from 1992 to 2015 in the HP analysis for Denmark. The aim of the HP-analysis for Denmark is to investigate whether the model can prove the formation and burst of the housing bubble prior to the financial crisis. As we have data ex-ante and ex-post this period, we regard it as reasonable to not go further back in time. Again, we want to further underline that Denmark is not our main focus.
The data collection for this thesis was completed on March 1st 2016. Published material after this date will not be considered.
2 METHODOLOGY
In the previous chapter we presented the theoretical basis to provide the reader of an overview of which theoretical frameworks and models that is included in the thesis, in order to answer the problem statement in the best possible way. In this chapter, we will present the methodological approach. Moreover, we will emphasize and discuss the thesis’ applied methods, in addition to the implementation of data material, in order to obtain reliable information and discuss the thesis reliability and validity.
The thesis uses an ontological framework, and the approach is post-positivistic. According to Guba (1990), this ontology assumes the reality to be objective, but grants that the reality can be apprehended only imperfectly and probabilistically. Hence, our findings are most likely true, but always subjected to falsification (Guba and Lincoln, 1994). Additionally, the approach is post- positivistic as we strive to ensure that the knowledge produced in this study can be used by others to undertake similar studies.
Moreover, we are using a deductive approach in the thesis. According to Johannessen et. al. (2011), the researcher tests general statements by using empirical data in a deductive approach. We are using house price models derived from well-known theories, with basis in the statement “are there tendencies of being a housing bubble in Norway”. Hence, we are testing our hypothesis by using specific theory, models and obtained empirical data. We do not attempt to verify or disprove theories empirically, but our paper is based on existing theories, frameworks and methods that further are combined in order to answer the thesis’ problem statement. The method we are using is descriptive, as the purpose of this thesis not is to develop or create a new model for assessing house prices, but rather investigate the problem statement by using existing models.
The problem statement will be approached by using empirical data and an analytical assessment of movements and fluctuations in the Norwegian housing market, seen in context to movements in the Danish housing market. Our findings will be discussed by using familiar analysis tools. Both qualitative and quantitative analyses will be applied to assess developments in the Norwegian and Danish housing markets. Our fundamental analysis is based on quantitative and qualitative
assessments, while the empirical analysis is primarily based on quantitative data. Hence, we apply theoretical frameworks that again are based on secondary data (Riis, 2005). The data are mainly based on journals, reports, articles from newspapers, publications from statistics banks and online resources. The academic areas of interest provided are quite wide and offers numerous of academic articles online. The time series and data sets for Norway are mainly obtained from The Norwegian Central Bank and Statistics Norway. Statistics Denmark is the main source for Denmark.
To achieve the best possible quality of the data provided in the assessment, it is crucial to achieve valid and reliable data material. When the report is largely quantitative, it is important to focus on validity and reliability when working in conjunction with preparations for data collection.
Reliability is related to how trustworthy the data is. The thesis will mainly use time series and data sets obtained from The Central Bank of Norway and Statistics Norway, in addition to Statistics Denmark. These are considered to be very reliable sources. The data are mainly processed using Excel. Excel can to a certain extent uncover errors in the data, for example by pointing to values which differ significantly from remaining data, so one can verify that the values are correct. It could however be debated how reliable the figures published by The Norwegian Central Bank are, as they go back to 1819. Many factors have changed significantly over the past 200 years. It has for example been sold homes with different standards in different periods, which makes it difficult to get the same basis for comparison in house prices. In addition, there are difficulties related to calculating the consumer price index. For example, a kilo of meat from a cow today, is difficult to compare with a kilo of meat from a cow in 1819, as the primary task of a cow before was to produce milk, while today's meat production is the primary task (Eitrheim et. al., 2003). Despite this, it is assumed that the data is reliable.
The validity of the thesis refers to the validity of the data relative to the problem to be answered. To achieve a high degree of validity, it is essential that the obtained data actually is relevant to the problem. All data used in the thesis have all been necessary to answer the research question regarding whether Norway have tendencies of a housing bubble (Grønmo, 2007).
3 HOUSING PRICES AND HOUSING BUBBLES
The purpose of this chapter is to present the theoretical background of the characteristics in the housing market. We will first provide the reader with knowledge of the term “housing bubble”.
Further, we will address theories of why housing bubbles occur and crack, namely Hyman Minsky´s model and Case and Shiller´s seven criteria. Moreover, we will provide an overview of supply and demand in the housing market. Next, we will present price determination and equilibrium in the short- and long-term. Lastly, fundamental factors are presented in the light of Jacobsen and Naug´s model.
3.1 Definition of a housing bubble
“Housing bubble” has become a more discussed subject over the last years. In the 1960s economists discussed whether market forces could prevent housing bubbles to occur, and that a bubble eventually have to burst (Stiglitz, 1990). There exist several definitions of a housing bubble, and we will in this section list some theoretical definitions. We will further derive our own suitable definition.
“If the reason that the price is high today is only because investors believe that the selling price will be high tomorrow – when “fundamental” factors do not seem to justify such a price – then a bubble exist” Stiglitz (1990).
Case and Shiller (2003) said “A tendency to view housing as an investment is a defining characteristic of a housing bubble”. A characteristic is that homebuyers do not believe the prices to fall so it is better to buy now than later. Furthermore, “a housing bubble exists when homebuyers buy houses that would normally be considered too expensive and is now an acceptable purchase because they believe they will be compensated by a further increase in price” (Case and Shiller, 2003). In other words, a housing bubble can be explained as a situation where an increase in house prices is affected by households’ expectations that house prices will increase. Hence, a bubble exists when house prices cannot be explained by fundamental factors.
Moreover, Kidleberger (1987) said: “A bubble may be defined loosely as a sharp rise in the price of an asset or a range of assets in a continuous process, with the initial rise generating expectations of further rises and attracting new buyers – generally speculators interested in profits from trading rather than in its use or earning capacity. The rise is then followed by a reversal of expectations and a sharp decline in price, often resulting in severe financial crises, which is when then bubble bursts”.
An interpretation of the definitions above implies that a bubble does exist when more people are viewing housing as an investment, and they are buying houses since they believe in a further price increase. This will increase the housing demand, which consequently will push the prices up. Hence, a suitable and adapted definition of a housing bubble on the basis of the above definitions is:
A housing bubble emerges when there is a positive change in house prices, which cannot be supported by changes in fundamental factors of the housing market. When we see a positive and greater discrepancy between fundamental- and actual value a housing bubble exists.
Calculations of fundamental values of housing can be made by basing the calculation on specific macroeconomics- and other explanatory factors. The market value of housing usually includes an element that takes expected price increase into account. Several factors may create expectations regarding house price growth. Households may expect that future ability to pay will be improved as a result of shift in macroeconomic conditions, such as remaining low unemployment, growth in income, and decreased interest rate. This will affect households’ future expectations of house prices.
The deviation from the fundamental value has a self-reinforcing effect, where homebuyers demand will expand due to positive expectations of future house prices. Homebuyers, who previously did not considered housing as a potential purchase, will consider it as acceptable if they have expectations of future rise in house prices. Buyers will now expect that future gains are possible with the assumption that the house will be sold to an even higher price in the future. Hence, the aspect of expectations has a central role when we identify a housing bubble, or indications for a housing bubble. However, housing bubbles are very difficult to predict, as there is no accurate and objective way to measure the fundamental factors.
3.2 Why housing prices are rising, housing bubbles occur and burst
Hyman Minsky’s model and theory presented in this part explains the different phases ex-ante and ex-post of a housing bubble burst. According to Levitin and Watcher (2012), there are several competing theories for factors causing housing bubbles:
”Some theories are demand-side theories, meaning that the housing bubble was causedby a growth in consumer demand for housing, which pushed up housing prices. Othersare supply-side theories, meaning that the housing bubble was caused by a growth in thesupply of housing finance, thereby enabling consumers to make more heavily leveragedbids for housing and bid up home prices.”
As Levitin and Watcher conclude, a strong demand will contribute to form a bubble in the market, but the increase in price may also stem from large supply, and easy access of credit. However, it is not enough to only explain why prices increase; it is just as important to. Even though bubbles are hard to determine, it is possible to recognize them by identifying distinctive characteristics for the presence of a housing bubble. We will particularly support us to the following theories:
• Hyman Minsky’s model
• Case & Shiller
3.2.1 Hyman Minsky´s model
Hyman Minsky published an article named “The Financial Instability Hypothesis” in 1978 where he introduced a model that describes various phases of an economy, and these phases’ distinctive traits.
In the context of the model, the economy exceeds its natural growth path and moves towards a bubble. Although the model is made up for financial crises, it can easily be transferred to- and used for the housing market. The model assumes that instability in the supply of credit is the main driver of a crisis. According to Minsky, access to credit creates crises. During the expansion phase in the economy, there is great optimism among investors, and they are risk averse. Similarly, the lenders also possess the same growth expectations. This means that lenders have a greater willingness to lend money during expansion phases. The opposite will occur in economic recessions. Interest expenses will rise, the lending policy becomes stricter and borrowing becomes more difficult. These
have a tendency to form the basis to reinvigorate bubbles and amplify crises. An illustration of Minsky’s phases is presented in the figure below.
Figure 3.1 Minsky´s model for bubble formation
Displacement
Displacement is represented by an exogenous macroeconomic shock. This is characterized by a process being initiated, which further causes a trend where the economy moves away from the normal growth path. This creates expectations around the possibility of making profits in the longer term. Hence, higher investment levels will occur. A result of this could be higher interest rates to reduce further overheating in the housing market. By fear of a possible appreciation of the currency, money supply will increase to avoid a further increase in interest rates. Increased money supply will justify lower interest rates and greater willingness to invest.
i. Overtrading
As a result of increased expectations in the previous phase, the economic activity increase. Hence, the overtrading phase is characterized by a high level of activity and growth, and the market has positive expectations regarding future profitability. As a consequence of increased activity, prices will rise, which in turn will lead to increased demand and further increased activity. This phase is a spiral that builds up in line with the market activity. The operators in the market will begin to overestimate values resulting from the increased expectations, and prices will exceed their initial
level. This is largely driven by psychology among the market operators. Speculations, psychology and irrational behavior are therefore the main characteristics of this phase, as people are worried to miss out on gains.
ii. Monetary expansion
This phase do often occur simultaneously with the overtrading phase, and is characterized by high optimism and irrational expectations for future profits. An increase in demand for money and credit is consistently in the monetary expansion phase. Banks and financial institutions see substantial opportunities to earn money and the supply of credit increases. The banks are conducting a freer lending policy, and optimism is further reinforced. Additionally, high demand for loans will increase competition among banks to capture and retain customers. As a consequence, the activity level increases.
iii. Revulsion
As illustrated in figure 3.1 this phase is the turning point in the economic recovery. The market operators discovered that prices are overvalued and that they cannot be supported by fundamental factors in the economy. The recession that occurs in the revulsion phase will spread uncertainty in the market. When people realize that the price growth will discontinue, they want to exit the market.
Demand for housing will decrease, the banks’ lending policies will be stricter, and the fall in money supply will push the interest rates up. Consequently, house prices will fall. The phase will therefore lead to liquidity problems, increased bankruptcies and low economic recovery.
iv. Discredit
The model’s final phase is characterized by lower values in turn of lower prices, while the money- and credit amount are reduced. Consequently, the demand for housing decreases. There is considerable pessimism in the market, and expectations of future profits are reduced significantly.
This results in prices becoming lower than their initial values.
3.2.2 Case and Shiller
Case and Shiller started their empirical analysis of the housing market in the 1980s and are considered pioneers in analyzing and determining housing bubbles. They argue that one should look at the stability of the relationship between income and other fundamental factors and home prices over time (Case and Shiller, 2003). In 2003 they conducted a study for the period from 1995 to 2002 in which they examined whether different fundamental factors could explain a high increase in house prices.
They claim that there are a number of indicative criteria that should be met to be able to say something about the existence of a housing bubble. According to Case and Shiller (2003), the following criteria must be present:
• Pressure of being a home-owner
If the majority of people own their own home, some people might consider owner-occupied housing as a “requirement”, and thus enter the house market.
• Widespread comprehension that it is profitable to own housing
Whether or not we are in a housing bubble can be seen in relation with people considering housing as an investment. If the buyer seeks to own the home only for the purpose of generating returns, it will be considered to be an investment.
• Widespread expectations of increase in house prices
An important factor for how house prices will further develop is expectations about the housing market. Households might expect further increase in prices. Consequently, households will buy houses that would normally be too expensive. The purchase will now be considered as acceptable, because households now believe they will be compensated by a further price increase in the future. If expectations of house price increase are the main motive for a housing purchase, the house market will not be stable in the long-term.
• House prices receive much attention in media and private conversations
The media affects developments in house prices, and articles published in the media usually affect what people are talking about in private contexts.
• House prices increase more than private income
House prices are not driven by costs if the increase in income is higher than the increase in house prices.
• Limited understanding of risk attached to the investment
Weak understanding of risk indicates that there might be a housing bubble. Among other, this contains weak understandings about changes in the interest rate.
• Simplified opinion regarding mechanics of the housing market dominates This is highly related to the factor above.
These seven market characteristics, among other factors, will be investigated throughout the thesis to examine whether or not there are indications of being a bubble in the Norwegian housing market.
Similarly, we will investigate if these characteristics underline the identified housing bubble in the Danish housing market prior to the financial crisis.
3.3 Demand and supply for the housing market
Housing investments are for a majority of households the largest investment they make in their lifetime. Housing procurement differs from procurement other assets because it is both a purchase of the consumer good “to live somewhere” and an investment in the home as a capital object. In this thesis we are focusing on the private housing market. A home is defined as an essential good, which we consider as both a consumer durable and an asset. The majority of households have a larger part of their wealth tied in their home (NOU, 2002).
Several factors influence the determination of prices in the housing market. Houses that are transacted in the market differ along a number of dimensions, such as space, location, standard and
type of ownership. Additionally, homebuyers’ motives vary from purely residential consumption to pure financial investments. In the latter case, housing is a capital object considered in relation to other investments based on the risk-adjusted returns from rental income and potential appreciation of value. Regardless of the motive behind the procurement, significant risk is associated with such a purchase, due to uncertain future disposable income streams and consumer preferences. The greatest uncertainty is still associated with the expected growth in house prices (Nordvik, 1993). However, considering a purchase with a long-term horizon will increase the ability to carry short-term price risk. Another factor that separates the housing market from many other markets is that the housing supply is given in the short-term, as explained in section 3.3.2. Regulatory and housing construction takes time, and adaption of capacity within the construction industry requires sustained high profitability.
Jacobsen and Naug (2004) argue that the most important explanatory factors for house prices are interest rates, housing construction, unemployment and household income. What actually determines house prices are housing supply and housing demand. In the short-term, housing supply is moderately stable since new constructions take time and house prices are thus determined by housing demand (Jacobsen and Naug, 2004). However, housing supply will affect housing prices over a long-term perspective. Construction costs will be important, but the price of land will also be significant for the amount of resources that should be used for new housing constructions.
Demographics, migration and centralization are also factors that will affect house prices in the long- term.
The housing market consists of partially integrated markets, such as the ownership- and rental market. We will use the ownership market as a starting point to explain the market's overall behavior. This is because the integration between the markets in the long-term affects the development and price level in the submarkets to directly influence each other. Simultaneously, the ownership market constitutes approximately 84 % of the total market in Norway (Statistics Norway, 2015), and 61 % in Denmark (Statistics Denmark, 2015). Neither the Norwegian, nor the Danish housing market is homogeneous, as housing is not identical. Nevertheless, as we mentioned in the delimitation section, we assume that all homes are equal, with the purpose to make the explanation
most intuitive. The basic scenario can be expanded to include more submarkets with different types of dwellings.
The rest of this section will explain the rationale behind demand and supply closer. In 3.3.1 we will look at housing demand, while 3.3.2 investigates the supply.
3.3.1 Demand
The demand side consists of all potential homebuyers, and we assume all homebuyers to have different willingness to pay. Each point on the demand curve3 represents an individual homeowner’s maximum willingness to pay, namely the marginal willingness to pay for housing. Total demand at any point on the curve is determined by the number of homebuyers who have a marginal willingness to pay equal to, or higher than, the buyer in that current point. Their ability to pay is in turn determined by factors, such as disposable income and assets, wealth, future expectations of income and expenses, interest rates and other factors affecting the housing cost. However, the willingness to pay is also affected by the value a buyer is putting into housing as a consumption and investment good, relative to other goods. Households with equal ability to pay may therefore have varying willingness to pay, and households with equal willingness to pay may have different ability to pay.
Households’ expectations can however change in the short-term and can thus contribute to significant fluctuations in their willingness to pay in a short-term perspective (Kongsrud, 2000).
In a long-term perspective, demographic change will be an essential driver of housing demand.
Housing demand consists of household’s demand for owner-occupied housing and for housing as an investment object. Numerous of buyers invest in housing if they believe they will get a positive outcome by renting or selling the object. Furthermore, housing demand is a function of the willingness to pay for different prices, and depends on different variables. We will now look at disposable income, the housing cost, credit limits, unemployment and demographics.
3 See figure 3.2
Income
Income affects housing prices in several ways. A real income target has to be used to explain the development in real house prices. Income may be measured as a gross- or net value where total income refers to a gross value, and disposable income is a net value. Disposable income measures the development in households’ budget, and is intuitively the vital factor for house prices. For a given tax level, income growth may contribute to increased house prices. However, this would not be the case if the growth is offset by higher taxes. To capture both of these effects we have decided to use disposable income.
The housing cost
More or less explicit, most models are using the housing cost as an explanatory factor for housing demand (Jacobsen and Naug (2004), NOU (2002), The National Bank of Denmark (2003) and Wagner (2005)). In line with Wagner (2005), the housing cost can be defined as:
Real interest expense after tax + Operating and maintenance costs - Housing tax
- Increase of housing value
= The cost of housing
The housing cost of owner-occupied housing equals the value of goods the owner misses out by owning the house in a given period (NOU, 2002). In other words, it may be understood as an opportunity cost. The housing cost is a more intuitive term than a figure we can estimate precisely.
The main component of the housing cost is interest expenses, as most homes are fully or partially leveraged. Hence, a higher real interest rate results in a higher housing cost, as the housing cost will increase when interest expenses on the loan increases. This effect will decrease or disappear if the loan has fixed interest rate for parts of the maturity. Additionally, higher interest rates will result in an increase in the opportunity cost of the equity tied up in the home. The consideration of housing as an investment will to a large extent depend directly on expected returns relative to assets as fixed
income securities. Low borrowing costs and alternative returns will tempt investors to invest in the housing market when interest rate cuts are not fully reflected in reduced housing rental costs. Needs for the consumer good “to live somewhere” will not change significantly by a rise in interest rates.
Tax deductions for interest expenses will decrease the interest- and housing cost, and thus stimulate housing demand. A higher deduction rate increases the stimulus in the same way as higher inflation for a given interest rate level.
Operating and maintenance costs are associated with maintenance and administration of housing on a daily basis, and include insurance and municipal taxes (NOU, 2002). Intuitively, housing demand is stimulated if operating and maintenance costs decrease, and conversely if they rise. These costs are difficult to estimate. However, they may be approximated by a fixed percentage of the housing stock. We assume this to be reasonable as development in operating and maintenance costs hardly explain a significant part of the development in house prices.
The tax benefit of owner-occupied housing differs from the tax benefit of debt financing of house purchasing4. The tax benefit of owner-occupied housing (tax on housing capital) is based on the assessed value (which usually is lower than the market value) of the house and relates to the fact that housing is taxed less than other assets in the calculation of income and wealth (NOU2002: 2, Lunde 1999). Housing is favorable taxed in Norway, which has a positive effect on housing demand and is a strong incentive for housing – making it an even more attractive investment. As a result, it is more beneficial to invest in housing than in other financial assets.
Increase in housing value is the most problematic component of the housing cost since it is difficult to estimate (Jacobsen and Naug (2004), Wagner (2005) and NOU (2002)). A potential increase in the housing value is unknown at the time of the purchase. Consequently, the assessment of this component is only based on expectations. Obviously, changes in interest rates, increase in income, and similar factors regarding the increase in housing value is difficult to estimate. Hence, historical
4 The Norwegian Tax Administration Act (1999) § 6-40 states that paid interest on debt may be deducted by a rate equal to 27 %. Hence, the tax is reduced by 27 % of interest expenses – reinforcing the benefit of owner-occupied housing.
This interest deduction applies to all kinds of debt, but the largest part of a majority of household’s debt is related to housing.
price inflation can be perceived as the best estimate of future housing price inflation, meaning that expectations are adaptive (Nordvik, 1993). Nordvik (1993) finds adaptive expectations to be the best approach for the formation of expectations for Norwegian households. Accordingly, the expected value of the housing costs can contribute to a housing bubble, where prices are significantly higher than what can be explained by fundamental factors (Case and Shiller, 2003).
Credit limits
The relationship between the housing cost and housing prices is less apparent in cases where there are effective credit constraints (Jacobsen and Naug, 2004). Credit limits set a ceiling on households’
purchasing power, and are limiting the housing price development. However, banks always consider the solvency for each and every borrower, and credit to households is limited in all situations of banks’ assessments. House price growth is affected in cases where banks’ lending policy change practices, as new types of loans and the possibility of interest-only periods may have increased households’ ability to distribute their consumption over a lifetime. This especially applies to first- time buyers.
Unemployment
Unemployment is an indicator of future prospects, and affects the amount of potential homebuyers.
Low unemployment improves wage settlements for everyone in the workforce. Simultaneously, larger workforce means more people being creditworthy. However, it may be problematic to include unemployment in a model where the income also is an explanatory variable, since developments in income and unemployment probably is closely correlated.
Demographics
Demographics impacts housing prices in several ways. The size of the total population relative to the housing stock is essential in the long-term. Increased percentage of people in the establishment phase is pushing prices up, especially for smaller dwellings. Reduced size of the average household has a similar effect. Moving patterns and centralizing tendencies may provide regional differences and thus affect average prices. For instance, increase in housing prices in cities are often greater
than the corresponding price growth in less densely populated areas, because housing supply in cities is less elastic than supply throughout the country. Some demographic factors may act simultaneously, and mutually reinforce each other;
• Settlement patterns of immigrants show a clear centralization trend
• The trend of smaller (and consequently a higher number) household is strongest in the cities
• Young people in the establishment phase are to a greater extent drawn to cities for conducting education and to work
All of these factors contribute to push prices up in urban areas.
Summary
In this section, we have discussed the relationship between explanatory variables for housing demand and -prices. Fluctuations in elements of the housing cost and other variables create cycles around the rising trend. The housing cost consists of interest expenses after tax, operational and maintenance costs, property taxes, and the expected appreciation of the residence. Changes in credit restrictions, unemployment and demographic factors may also affect housing prices. We will now introduce the supply side of the housing market.
3.3.2 Supply
Figure 3.2 illustrates determination of prices in the short-term. It shows the equilibrium where a group of homebuyers equal to the line segment of AB are left without a house.
Figure 3.2: Price determination in the short-term
Hendry (1984) stated that housing supply is perfectly inelastic in the short-term5, which means that it is relatively fixed. Consequently, the short-term housing supply is not largely affected by a rise in demand or in housing prices. The most obvious reason is that new constructions are time consuming. Even though housing supply is changing somewhat as a result of reconstruction of commercial buildings and natural attrition of housing, a good approximation is to assume the offer to be constant in the short-term. Thus, the supply curve becomes vertical, as illustrated in figure 3.2.
In this section we will discuss factors affecting housing supply in a theoretical and empirical perspective. The housing stock in a given period depends on the stock of existing homes from former periods, in addition to new constructions subtracted the housing falling out of the market.
However, new constructions will in turn depend on prices of existing homes, as new and existing homes are close substitutes. Housing investments will also depend on other factors, which in turn may be influenced by the willingness to pay. One such factor is the construction cost of new homes.
We will now look at explanatory variables on the supply side, namely construction- and property costs.
Construction cost
The factor price of labor and materials, housing requirements and productivity in the construction industry are fundamental factors affecting construction costs.
Property cost
The cost of land and property reprocessing determine the property cost. There is ample supply of land and potential property areas in most places in both Denmark and Norway, at least outside of urban areas. However, public infrastructure, in addition to necessary regulations, is just as vital as the property area itself. Cost of land outside urban areas are mainly determined by the alternative value of the land, as farmland or similarly.
Scarcity of land provides different prices in urban areas than in less densely populated areas. Access of land determines housing prices more than the construction cost itself in metropolitan areas.
Extensions of the housing stock must take place on vacant lots, in the periphery, via densification, or by reversal of commercial property. Other key factors are building processing, infrastructure in terms of public transportation and communication technology.
Housing supply is stimulated by higher prices on existing homes and lower construction costs.
Housing investments and the capacity in the construction industry are adjusting slowly. The construction cost may deviate from normal levels over longer periods, controlled by factor prices and the productivity in the industry. Slow adjustments make housing supply inelastic. Property access is the most important limiting factor for housing supply in urban areas, while it is a small problem in less densely populated areas where the alternative value of the area is less.
3.4 Price determination and equilibrium in the short- and long-term
Figure 3.2 illustrates a positive market price, as there is a shortage of housing. In equilibrium, all homes for sale will be bought6. The line segment of AB shows potential homebuyers with a lower marginal willingness to pay than the market price. These homebuyers will thus not able to buy a
6 The equilibrium price will equal the point of intersection where the supply and demand meet in a perfect market without frictions
home in this market. They will have to seek other markets to find homes priced below their reservation price.
The demand curve illustrates, for any price level, how many homebuyers that have a willingness to pay equal to, or above, the actual price level. In practice there are not only one, but several homebuyers with a willingness to pay equal to each and every price level. Hence, the slope of the demand curve does not necessarily have to be minus one. The crucial factor of the slope is determined by the number of homebuyers with a willingness to pay equal to every single price level, and the slope may be exponential. Similarly, the slope of the supply curve is determined by the housing stocks’ price sensitivity. Higher price levels will result in more new constructions over time, which in turn will increase the total housing stock, and in the long-term we expect the slope of the supply curve to be positive. We will now address price determination in the short- and long- term.
3.4.1 The short-term
As mentioned above, the short-term housing supply is inelastic, and it is reasonable to assume the housing supply to be unchanged. This implies that prices in the short-term are determined by factors on the demand side. Hence, drivers of a price increase in the short-term may be explained by increased disposable income or expectations of increased disposable income, a potential tax advantage of home ownership, a lower interest rate level, an expected increase in housing prices, or lower operating- and maintenance costs. This requires everyone to have access to credit. In practice, these effects will turn out differently, depending on whether people have debts, wealth or if they are existing homeowners (Kommunal og Regionaldepartementet, 2002).
The market price will be equal to the short-term demand, and if prices are lower than the market price, the number of demanded homes will exceed the number of supplied homes. In that case we will experience a housing shortage. People with the highest willingness to pay are expected to make the highest bids in order to ensure the house. Accordingly, prices will be pushed up, until there is market clearance. Hence, the market price will equal the marginal demanders’ willingness to pay. In
other words, only people with a capacity to pay a minimum price, equivalent to the market price, will be able to buy.
3.4.2 The long-term
It would not be reasonable to assume housing supply to be given in the longer term, as the commencement will pick up when housing prices exceed construction costs. Further, this will in the longer-term result in a higher housing supply, which in turn will curb the price increase. New constructions will continue until the marginal cost of constructing one more unit equals the market price - at this time it will no longer be profitable.
In the long-term, housing supply will adapt to the demand. If the number of new constructions exceeds the resignation, the housing stock will increase. Hence, the short-term level of supply will shift to the right, as shown in figure 3.3. This will result in a long-term rising supply curve.
Additionally, it is conceivable that other factors than the demand components may result in shifts in the demand. Among others, demographic factors and long-lasting production changes between sectors may play substantial roles. In addition, the supply side will in the long-term be limited by the quantity of building plots available (Skak, 2011).
We have in figure 3.3 illustrated a positive shift in the demand curve. The new demand curve is named D’. As shown, this will cause a new price of P’’ in the long-term. The long-term price is lower than the short-term price P’. A more elastic supply can more rapidly adjust to the demand, and in turn, the long-term price increase will be lower.
Firstly, we get a new price in the intersection between supply in the short-term and the new demand curve (b). As illustrated, the short-term price increase is independent of the supply elasticity, as it is believed that the supply is given in the short-term. However, the housing stock will in the long-term be adjusted up to meet the increase in demand. This is illustrated by a rightward shift in the short- term supply curve, S’. If the housing stock increases more than the demand, the reservation price will be lower, and thus prices will fall. On the other hand, we will get the opposite effect if the demand side changes more rapidly than the increase in the housing stock. The price will increase to
P’, given by the new equilibrium in b. Ultimately, figure 3.3 illustrates how the market price has increased as a consequence in a shift in the demand- and supply side, which is shown in c. The long- term price is named P’’.
Figure 3.3: Housing supply in the short- and long-term
Income is the most central driver for housing demand, and an increase in income provides a higher disposable income if everything else is held constant. Households will afford more expensive housing, and the willingness to pay for housing will increase. This means that the housing demand increases, which is illustrated in the figure above as a rightward shift in the demand curve from D to D’. This will drive prices up, and existing homeowners will come fortunate out of the situation.
They will obtain increased wealth, as their home will become more valuable. People selling their homes will obtain an economic gain on the sale. Conversely, people who are trying to enter the house market will have to pay a higher price for the exact same standard than what they would have done before the increase in average real income occurred.
Additionally, a reduction in the interest rate will increase the willingness- and the ability to pay, as the housing cost will be reduced. People with existing debts may be granted a new mortgage loan, provided that they are granted credit. Housing prices will be pushed up, and the price increase will
