Real estate related linkages in highly developed European
countries
University of Groningen Faculty of Economics and Business
Master thesis International Economics and Business
Author:
Sil Spiegelaar – S3029808
s.spiegelaar@student.rug.nl
Supervisors:
Prof. dr. D.J. Bezemer (University of Groningen, supervisor) Prof. dr. B. Los (University of Groningen, co-assessor) M. Klok (ING, supervisor)
2
Abstract
The recovery of the Dutch economy, which started in the second quarter of 2013, has been largely contributed to the recovery of the housing market. In order to analyse the importance of the real estate and the construction sector, linkages are calculated using an input-output analysis for the period 2000-2014. Furthermore, in order to assess the development of these linkages, correlation between real estate related linkages and housing prices/transaction volumes are determined. The results indicate that both sectors have strong linkages to the rest of the economy, moreover the strength of the linkages increased throughout the analysed period. Additionally, real estate linkages are correlated with transaction volumes and construction linkages are correlated with housing prices.
3 CONTENT 1. INTRODUCTION ... 4 2. LITERATURE REVIEW ... 7 2.1 Input-output model ... 7 2.2 Linkages ... 10
2.3 Role of the real estate related sectors in the economy ... 11
2.4 Correlation between property prices and linkages ... 13
2.5 Hypotheses ... 15
3. METHODOLOGY & DATA ... 16
3.1 Measurements of linkages ... 16
3.2 Traditional measurements versus hypothetical extraction ... 17
3.3 Total linkage indicator ... 18
3.4 Linkages based on labour and value added ... 19
3.5 Correlation between linkages and property prices... 20
3.6 Data for input-output tables ... 20
3.6.1 Data source ... 20
3.6.2 Data limitations ... 22
3.7 Data on property prices ... 23
4. EMPIRICAL RESULTS ... 24
4.1 Real estate activities linkages ... 24
4.2 Construction linkages ... 27
4.3 Combined linkages ... 30
5. CONCLUSION & LIMITATIONS ... 33
4
1. INTRODUCTION
The Dutch economy began recovering in the second quarter of 2013. Real GDP has been rising again and the housing market is showing a similar trend. Average prices of existing homes started rising after the second quarter of 2013. Similarly, housing transactions, housing investment and construction sector activity also saw a turnaround in mid-2013 (De Nederlandsche Bank, 2017). De Nederlandsche bank (2017), the central bank of the Netherlands, calculated how much the housing market contributed to the recovery of the Dutch economy by constructing an alternative scenario where housing prices did not rise. By their calculations, a little over one quarter of the cyclical upswing of GDP until the end of the third quarter in 2017 can be contributed to the housing market. The influence on consumption is even bigger. Without the housing market recovery, private consumption would have increased by 60% less. Furthermore, the reduction of the unemployment rate would have been lower by half. According to De Nederlandsche Bank (2017) the housing market contributed in two ways to the recovery of the Dutch economy. First, realised house prices, due to the increased value of owned homes and consumer confidence, have had an upward effect on private consumption and GDP. Second, housing investment – consisting of new building, renovations, and transaction costs of home sales – also gave a boost to production.
The influence of real estate prices, and especially housing prices, is a widely researched and debated topic. The general consensus is that rising housing prices increase private consumption in one of three ways. The direct wealth effect, an increase in housing prices increased housing wealth. In turn, this increases private consumption because households adjust their lifetime plan. Second, housing capital as collateral, a house is an asset that can be used as collateral, an increase in house prices increases the value of the collateral. This extra collateral can be used to borrow and in turn consume more. Finally, correlation is
influenced by common factors, there is no causal relationship between housing prices and
consumption, but rather they are both affected by a common factor (Campbell & Cocco, 2007; Cas, Quigley & Shiller, 2005; Windsor, Jääskelä, & Finlay, 2015; Browning, Gortz, & Leth-Petersen, 2013).
5
opportunities in upstream and downstream sectors and induce private investment in these sectors (Holz, 2011).
This study will address the role of real estate in the Dutch national economy by calculating linkages for the sector for real estate activities and the construction sector. Since I-O tables focus on inter-sector trading and display all flows of goods and services, it is considered a main tool to determine linkages between sectors (Song, & Liu, 2007). That is why, in order to analyse the connectedness of real estate related sectors, this study will use an input-output analysis to determine linkages between real estate related sectors and the rest of the economy. In order to get a clear picture of the relative size of these linkages, they will be compared to countries that are similar to the Netherlands. The countries that will be compared to the Netherlands are Austria, Belgium, Denmark, Ireland and Sweden. These countries have been chosen because of their comparable size (population), comparable income (GDP per capita), and because they are all members of the European Union. These features increase the comparability between these nations. Appendix A contains a comparison of European countries in size and income. Furthermore, the evolution of the housing price and the number of houses sold is depicted for the selected nations.
Changes in home prices and trading volumes seem to have an economic impact on builders, brokers, lenders, and appraisers (Clayton, Miller & Peng, 2010). That is why the correlation between property prices and real estate and construction linkages are analysed, as
well as the correlation between the number of houses sold and real estate linkage. This will
give an indication whether real estate related linkages become more important to the national economy when property prices and housing transactions increase. Through an input-output analysis this study will try to answer the following question:
“Based on economic linkages, can the Dutch real estate related sectors be considered key sectors and how does this relate to countries similar in size?”
Although the calculations of the De Nederlandsche Bank (2017) are mostly aimed at the housing market, which is to be considered residential real estate, this study will take into account the real estate related sectors. These real estate related sectors are the sectors for real
estate activities and construction. The United Nations (2008) defines the sector for real estate
activities as: “lessors, agents and/or brokers acting in one or more of the following: selling or buying real estate, renting real estate, providing other real estate services such as appraising real estate or acting as real estate escrow agents. Also included is the building of structures, combined with maintaining ownership or leasing of such structures.” The construction sector is defined as: “general construction and specialized construction activities for buildings and civil engineering works. It includes new work, repair, additions and alterations, the erection of prefabricated buildings or structured on the site and also construction of a temporary nature. Also included is the repair of buildings and engineering works” (United Nations, 2008). In the rest of this study, the combination of the construction sector and the sector for real estate activities will be indicated with real estate related sectors.
6
construction sector for the same set of developed countries. They both found strong linkages, however the data that was used in both studies originated from before 2000. This study will add to existing literature by providing an analysis with far more recent data. Moreover, although there are quite a few studies that examine linkages of the construction sector, not many combine linkages of real estate activities and the construction sector. In this study real estate linkages and construction linkages will be combined into one linkage. Not only will this illustrate the importance of real estate related sectors, it will also facilitate in broader understanding of the effects that the real estate sector has on a national economy. Finally, this study will add to existing literature by analysing the correlation of both real estate and construction linkages with housing prices and transaction volumes.
Results indicate that both the sector for real estate activities and the construction sector can be considered key sectors in all six analysed countries. Moreover, after the 2008-crisis the strength of the real estate related linkages even started to increase for the Dutch economy, which indicates that these sectors became even more important to the national economy. Correlation coefficients between real estate linkages and the number of housing transactions indicate a positive and significant relation. Furthermore, construction linkages and property prices also had a positive significant correlation. This indicates that with higher levels of activity in the property market construction linkages and real estate linkages will increase. The correlation coefficients for real estate linkages and housing prices, however, were not significant. This indicates that real estate linkages do not increase when housing prices increase.
7
2. LITERATURE REVIEW 2.1 Input-output model
In order to understand the concept of sectoral linkages, it is first important to understand the concept of the input-output (I-O) table. Leontief (1937) described the I-O model as a model that is primarily concerned with interdependence between economic agents. It captures the activity of a group of sectors that both produce products –outputs- and consume products – inputs- in the process of producing each sector‟s own output. The I-O analysis captures the flows of products from each industrial sector, considered as a producer, to each of the sectors, itself and others, considered as consumers (Miller & Blair, 2009).
Figure 1 shows a standardized I-O transaction table. The rows of the table labelled
producers represent the output of the corresponding sector throughout the economy. The
columns labelled producers as consumers describe the composition of inputs which are required by the corresponding sector to produce its output. These inter-industry exchanges of
products are indicated by the shaded part in figure 1. The column labelled final demand
records the sale by each sector to a specific final demand category, such as private consumption, government consumption, and exports. In figure 1 these are private consumption and government consumption. The rows labelled imports give the imports specified for each sector. The row value added accounts for other inputs to production, such as labour, depreciation of capital, and indirect business taxes. Finally, the sum of all value added gives the gross domestic product (Miller & Blair, 2009).
Sectors use both domestically produced inputs and imported inputs. This study focusses on the linkages within a single country. That is why only domestically supplied inputs should be considered, since it is the impact on the domestic economy that is of concern. Disregarding this issue has important empirical implications and may bias the results. When the distinction between imported and domestic inputs is neglected multiplier effects of a sector might be overestimated (Dietzenbacher, Albino & Kühtz, 2005; Reis & Rua, 2006). That is why this study will only focus on linkages within the shaded part of figure 1 and linkages for imported inputs will not be considered.
Note that the standardized I-O table in figure 1 is highly aggregated. Sectors can be disaggregated in as many sectors as data allows. For example, one could aggregate into the
manufacturing sector, or separate to a steel manufacturing sector and when data allows it, one
could even categorize as small as a steel nails and spikes sector. The same holds for the value added rows and the final demand columns (Miller & Blair, 2009).
8
Figure 1, standardized I-O table. Source: Miller & Blair, 2009.
To be able to transform the physical production process into monetary terms, several assumptions have to be made. The first assumption is that each sector only produces one good or service. While most sectors will produce more than one product, this assumption is very convenient. Each sector now has a one-to-one correspondence between sectors and products. Moreover, because each sector only produces a single product it will now have one price. The second assumption is that production is proportional. As mentioned above, the input coefficients (A) are constant. This means that economies of scale are non-existent and that substitution is not possible when the output of sector changes. The third assumption is that all prices are equal to one. Although this may seem to be a highly rigorous assumption, it is not. It means that the unit of measurement is chosen is such a way that the unit of measurement is the quantity that can be bought with one euro. For example, when the wage rate is 60 euros per hour, the I-O table will show the wage rate as 1 euro per minute (Dietzenbacher, 2017).
The inter-industry exchange (figure 1) depicts the flows between a pair of sectors, where the sector on the rows (producer) is sector i and the column sector (producer as consumer) is sector j. If we assume that the economy consists out of n sectors, the shaded part
is then a n x n matrix called zij that gives the intermediate deliveries (from each sector i to
sector j). In the zij matrix, sector j‟s demand for inputs from each sector is related to the total
production by sector j (Miller & Blair, 2009). Additionally, there are also purchasers who are external to the industrial sectors that constitute the producers in the economy. These are for example households, governments, and end users outside of the economy, i.e. exports. The demand of these units is not based on the amount that a specific sector produces but is rather based on the unit‟s own needs, like government purchases or consumption by private households. Since the demand of these external units tends to be for their own use and not as an intermediate input for a production process, these products are generally referred to as final
demand and denoted as fi the final demand of sector i‟s product (Miller & Blair, 2009). Total
output (xi) is then given by:
∑ (2.1)
Private Government Exports
consumption consumption Agriculture Manufacturing Services Agriculture Manufacturing Services Labour Other Final Demand
Gross domestic product
V a lu e a d d e d Im p o rt s To ta l o u tp u t Total output Labour compensation
Other value added Producers as consumers
Agriculture Manufacturing Service
9
Where is the number of sectors in the economy. The elements of equation (2.1) can also be rewritten to show the form of the vectors and matrices:
[ ] , [
] , and [ ] (2.2)
Here and throughout this paper lower-case bold letters represent a (column) vector, as in f for final demand and x for total output, and an upper-case bold letter for a matrix, as in Z for the inter-industry sales. A special summation vector is i, which is a vector of 1‟s. This is used to create, post-multiplication, a column vector whose elements are the row sums of the matrix (Miller & Blair, 2009). Finally, equation (2.2) can then be rewritten as:
(2.3)
A fundamental assumption of the inter-industry flows from sector i to sector j is that this depends entirely on the total output of sector j (Miller & Blair, 2009). This relation can be illustrated as follows. Take for example the steel sector (i) and car manufacturers (j). The input of steel bought by car manufacturers depends on the amount of cars that are being
produced, which gives the relation zij. The ratio of steel input to car output is then given by:
(2.4)
Calculating the ratios between each sector will result in an input coefficient matrix of n x n, which is called A and only examines domestically created inputs. Note that the input
coefficient aij is constant for the period that the I-O model examines. This means that there are
no economies of scale and thus that the production process operates under constant returns to scale (Miller & Blair, 2009).
Based on equation (2.4), multiplying the entire input coefficient matrix (A) with the gross
output of each corresponding sector (xj) would result in the inter-industry exchanges of
products (Z). This can also be written as:
(2.5)
Let I be an identify matrix, which is a matrix that consists of ones on the diagonal and zeros everywhere else. I can be used to turn a vector into a diagonalized matrix. For any given final demand vector f, the gross output that is necessary to satisfy f is given by:
(2.6)
10
(2.7)
The Leontief inverse illustrates the dependence of the gross outputs on the values of each of
the final demands. Hence, the Leontief inverse (lij) can be interpreted as the extra input from
sector i that is needed to satisfy an (extra) final demand of one euro for product j (Miller & Blair, 2009).
2.2 Linkages
Linkages are generally divided into three groups, depending on the direction of their interdependence. The first group consists of the backward linkages. If sector j increases its output, it will also increase its demand for inputs that are used by sector j. These linkages identify how a sector depends on other sectors for their input, the sectors that supply input to sector j are called “upstream” (Miller & Blair, 2009). When dependence is high, the sector has a high economic pull. This indicates that higher output of the specific sector will have a large effect on the other sectors that supply input to that sector (Pietroforte, & Gregori, 2003). The second group consists of the forward linkages. Additionally, if sector j increases its output it will also mean that sector j has more output available for sectors that use products of sector j as input. So, there will be increased supplies from sector j as a seller. These linkages identify how a particular sector distributes its outputs to the remaining economy, the sectors that use products from sector j are called “downstream” (Miller & Blair, 2009). When these measurements are high, dependence of other sectors on the selected sector is high (Pietroforte, & Gregori, 2003). Finally, the last group are the total linkages. Total linkages capture all the linkages a sector has, in all directions of interdependence. It captures both backward and forward linkages, as well as inter-sectoral linkages i.e. a sector delivering intermediate inputs to itself (Holz, 2011).
11
2.3 Role of the real estate related sectors in the economy
Real estate plays an integral role in most developed economies and forms a critical part of the economic system. Residential real estate is often the greatest source of wealth and savings for families, while commercial real estate creates jobs, spaces for retail, offices, and manufacturing (Campbell, & Cocco, 2007; Boshoff, & Seymore, 2016). The construction sector, in turn, has always been considered to be an essential contributor to the process of economic growth. On the demand side the construction sector purchases large amounts of intermediate products from other sectors, while on the supply side it provides basic infrastructure that is used by other sectors (Zheng, Chau, Hui, 2012).
In a highly influential series of papers, Bon (1992; 2000) analysed the evolution of the construction sector, based on the stage of economic development of a vast amount of countries from all continents. In the early stages of development, the share of total output for the construction sector increases but ultimately declines in relative terms. At some point, in highly industrialized countries, the decline is not only in relative terms but also in absolute terms. This implies that the total construction output follows an inverted U-shape. In the initial development stage, the share of construction output in GNP increases but when a country becomes more developed, the share of construction output will start to decline. At some point the basic major infrastructure of a country is put in place and the need for new construction gradually becomes less (Bon, 2000; Giang & Pheng, 2011). The inverted U-shape relation between the share of construction in GNP and GNP per capita was further confirmed by Crosthwaite (2000) and Yiu, Lu, Leung, and Jin (2004). This finding also implies that the repair and maintenance part of the construction sector is more important in developed countries, whereas the new build and development part of the construction sector is a more important part for developing countries (Giang & Pheng, 2011).
This resulted in policy implications for developing countries that focussed on expanding the capacity of the construction sector. Development strategies mainly centred on removing constraints of the production factors for the construction sector, such as labour materials, capital, and technology. Furthermore, governments tried to influence construction activities through fiscal policies related to government spending and taxes on public construction works. Through this influence, the construction sector was expected to encourage employment and facilitate economic growth (Giang & Pheng, 2011).
12
A long run equilibrium exists between real estate prices and construction output. Zheng, Chau, and Hui (2012) show, with data from China, that higher property prices will lead to more new construction, which will ultimately generate greater construction output. This in turn, will lead to higher levels of demand for sectors from which the construction sector draws intermediate input. When demand for a product increases, not only will the price of that good increase, it will also increase the price of goods and services that are linked through intermediate demand. Since the construction sector‟s nature of operation is to assemble a large number of different products purchased from a large number of industries, a higher demand for construction will also influence a large number of other sectors. Resources, intermediate inputs and services which are used by the construction sector will also see an increase in demand and thus an increase in output when demand for construction activities rises (Chan, 2002).
The construction sector uses a wide variety of materials and services which are provided by other sectors. This indicates that the construction sector can stimulate the expansion of these industries through strong backward linkages. Furthermore, the sector for real estate activities shows strong forward linkages. These strong forward linkages stem from the fact that all other sectors of the economy use the flow of goods and services generated by the sector for real estate activities (Giang & Pheng, 2011). As a result, the supply of real estate activities can raise the profitability of production and increase output and employment of other sectors. These strong linkages for the real estate sector and the construction sector are confirmed by several studies (Song and Liu, 2007; Pietroforte and Gregori, 2003; Song et al., 2006).
Song and Liu (2007) analysed and compared the linkages of the sector for real estate related activities to the rest of the economy for seven developed countries (Australia, Canada, Denmark, France, Japan, Netherlands, and USA) from 1970 until 1998. They used the hypothetical extraction method (see section 3) and found an increasing trend for the strength of real estate linkages, which confirmed an increasing role of the real estate sector with higher economic development. The main reason for the increasing real estate linkages seems to be the rising real estate prices in these countries (Song, Liu & Langston, 2005). The backward linkage indicators for the real estate sector are all scattered at a low value. This suggests a strong sectoral independence and thus a weak economic pull by the real estate sector. However, forward linkages are at a high value, which indicates a weak sectoral independence and thus a strong economic push of the real estate sector (Song & Liu, 2007). This indicates that for these seven countries real estate activities is a suitable sector for policies in order to facilitate an economic push, which could lead to higher economic growth.
13
Pietroforte and Gregori (2003) analysed linkages in the construction sector based on traditional methods (see section 3) for the period 1968-1990, while Song et al. (2006) used the HEM for the period 1971-1998 for eight highly developed countries (Australia, Canada, Denmark, France, Germany, Japan, the Netherlands, and USA). The authors find that the construction sector has strong backward linkages and weak forward linkages. Furthermore, the construction sector displays a declining trend in the strength of the linkages. Hence, the construction sector becomes less important when countries become more developed, which is in line with the inverted U-shape by Bon (2000). Additionally, Pietroforte and Gregori (2003) indicate that there is a difference between the examined countries. When split in to two separate groups, the examined countries display varying degrees by which the construction and manufacturing sectors are interconnected. Japan, Germany, the Netherlands, and Denmark have relatively high linkages, while the remaining countries have lower linkages. The authors contribute lower linkage in the last group to the de-industrialization of these countries, with a shift from manufacturing towards services, which lowers the linkage between these sectors. This is also illustrated by a decline in linkages between the construction sector and manufacturing sectors, while linkages between the construction sector and the service sectors show an increasing trend (Pietroforte & Gregori, 2003). However, data for this study consists of the period 1968 until 1990. So, it is likely that Japan, Germany, the Netherlands, and Denmark now also have lower interconnectedness between the manufacturing sectors and the construction sector due to de-industrialization.
2.4 Correlation between property prices and linkages
A well-known pattern in the housing market is that prices and trading volume seem to correlate with each other. Transaction volumes in existing property are more intense and property has less time on the market before sale when property prices are rising compared to falling property prices. This correlation is generally interpreted as price changes causing changes in trading volumes (Clayton et al., 2010).
This causal relation is contributed to one of the three following factors. The first factor is equity constraints, falling prices reduce homeowners‟ home equity. This leads to homeowners who cannot afford a down payment on their new home from the sale proceeds of their current home, moreover they will also need additional funds to repay the existing mortgage. This results in constrained households who cannot move, which in turn decreases market demand (Chan, 2001; Engelhardt, 2003; Clayton et al., 2010). The second factor is
nominal loss aversion. Homeowners are less willing to sell in falling markets, since they are
14
While this may hint that property prices and transaction volumes are correlated and thus real estate linkages would be either correlated to both or correlated to neither, Clayton et al. (2010) show, by testing Granger causality, that only a decrease in property prices affects trade volumes and that an increase in property prices does not affect trade volumes. The fact that falling property prices decrease trade volumes is mainly caused due to equity constraints and nominal loss aversion, which causes a fraction of the seller not being able to move (Clayton et al., 2010; Steiner, 1995).
In contrast to Clayton‟s (2010) findings, Hort (2000) and Andrew and Meen (2005) find that property prices and transaction volumes respond differently to shocks in fundamentals of the housing market. Hort (2000) shows that, based on a VAR-model with Swedish data, that an interest shock has an immediate effect on transaction volumes, but prices only decline gradually. Moreover, Andrew and Meen (2003) use data from the UK to determine the adjustment of housing prices and transaction volumes to fundamentals. Their results indicate that, while a shock to fundamentals of the housing market drives sales and prices in the same direction, the sales effect peaks after about a year while prices decline for more than two years after the initial shock. Andrew and Meen (2003) contribute this to the mismatches in the valuations of properties by buyers and sellers. If sellers adjust their reservation prices only slowly in response to a negative demand shock, then the volume will be affected before prices.
Property prices may influence output in the real estate sector through brokerage fees. Generally, the brokerages fees that are calculated are a certain percentage of the selling price of the property (Martin & Munneke, 2010). Hence, when property prices increase the value of the brokerage fee will also be higher than without the higher level in property prices. When the brokerage fee is higher, total gross output will also increase in the real estate sector. This, in turn, might influence the strength of the linkage in the real estate sector.
Since prices and transaction volumes seem to respond differently to shocks and prices and transaction volumes seem to not always be correlated, it seems appropriate to test both the correlation between real estate linkages and property prices and real estate linkages and transaction volumes.
As with the real estate market, the market for construction activity is also believed to be correlated with property price levels. It is widely believed that construction activity depends positively on the ratio of property prices to construction costs. This means that,
ceteris paribus, an increase in property prices may lead to an increase in construction activity
(Hoffman, 2004; Zheng et al., 2012). It is generally believed that two transmission channels between property prices and construction activities exist. The first channel is that of real
estate corporation. An increase in property prices will lead to the expansion of real estate
corporation, which in turn will stimulate the expansion of real estate development and general construction activity. The second channel is that of urban renewal. When the expected housing price exceeds the cost of replacing or constructing the underlying asset, the urban renewal projects increase. This, in turn, will lead to higher construction output (Zheng et al., 2012).
15
corresponding sectors. They indeed find a long-run relationship between property price levels and construction output for both housing and retail construction output. Moreover, they find that property price levels have a stronger impact on housing construction than on retail construction. This, again, indicates that there might be correlation between the strength of the construction linkage and the property price levels.
2.5 Hypotheses
Earlier studies suggested that the sector for real estate activities is an important sector to the economy. Moreover, when economies further develop, the importance of the real estate sector increases. The construction sector, despite its decreasing trend, still fulfils an important role in the national economy. This leads to the first and second hypotheses:
H1: The sector for real estate activities will have a strong total linkage with the rest of
the economy.
H2: The construction sector will have a strong total linkage with the rest of the
economy.
Note that the concept of strong linkages will be defined in section 3.3 through equation (3.9).
Furthermore, transaction volumes and property prices may influence the size of the real estate linkages. With rising property prices and higher transaction volumes property markets become more active, which would increase real estate activities. A higher level of transaction volumes influences output in the sector for real estate activities directly, while housing prices might influence output in the real estate sector through higher brokerage fees. Since housing prices and transaction volumes are not always correlated or respond differently to shocks, it seems appropriate to test both individually. This leads to the third and fourth hypothesis:
H3: The strength of the total linkages for the real estate activities sector is positively
correlated with property price levels.
H4: The strength of the total linkages for the real estate activities sector is positively
correlated with the number of housing transactions.
Finally, construction output and property price seem to be correlated as well. When housing prices increase to higher level, investment in new construction increases. This also leads to expectations that construction linkages and property price levels might be related. This results in fifth and final hypothesis:
H5: The strength of the total linkages for the construction sector is positively
correlated with property price levels.
16
3. METHODOLOGY & DATA 3.1 Measurements of linkages
There are several measurements of linkages, which are generally divided into two broad groups: the traditional methods and the hypothetical extraction method (HEM) (Song & Liu, 2007). Two types of measurements for the backward linkages are most common within the traditional method. The simplest form is by taking the sum of the jth column in the input coefficient matrix (A). Since the input coefficient only measures the direct domestically needed input for one extra output, it only measures the direct effects and not the indirect effects. This means that this linkage only take into account the value intermediate inputs for sector j but not the inputs that are needed for the intermediate inputs and so on. That is why this linkage is called the direct backward linkage (Miller & Lahr, 2001):
∑ (3.1)
The second measurement for the backward linkage within the traditional method is called the
total backward linkage. It is calculated by taking the column sum of the Leontief matrix (L)
for sector j. By taking the column sum of the Leontief matrix for sector j, the output multiplier for sector j is calculated. This output multiplier is defined as the total value of production in all sectors of the economy that is necessary in order to satisfy one euro‟s worth of final demand for sector j (Miller & Blair, 2009). This means that the total backward linkage captures both the direct and indirect effects:
∑ (3.2)
Early measurements of the forward linkages were also based on the Leontief model. The row sums of the input coefficient matrix (A) were taken for the direct forward linkage and the row sums of the Leontief matrix (L) were taken for the total forward linkage. However, both of these measurements have been viewed with scepticism. Using the Leontief model to measure forward linkages for sector j is essentially to measure the rest of the economies backward linkage impact on sector j. One would not be measuring the forward linkage of sector j but rather the backward linkage of the rest of the economy on sector j (Cai & Leung, 2004). That is why suggestions were made to base forward linkages on the Ghosh model, which is an alternative to the Leontief model (Miller & Lahr, 2001). The difference between the Leontief
inverse and the Ghosh inverse is essentially that the Leontief inverse element (lij) is
interpreted as the extra input from sector i that is needed to satisfy an extra final demand of
one euro for product j. While the Ghosh inverse element (gij) can be interpreted as measuring
the total value of production that comes about in sector j per unit of primary input in sector i (Miller & Blair, 2009). This results in the following two formulas for the direct forward
linkage and the total forward linkage respectively:
17
∑ (3.4)
The second group of measurements is the hypothetical extraction method (HEM). The objective of HEM is to depict how much the total output within an economy would change if a particular sector were removed form that economy (Miller & Blair, 2009). This involves deleting the sector that is to be examined from the I-O table. This can be done by setting the rows and columns in the input coefficient matrix (A) of sector j to zero. After the removal of the sector, ceteris paribus, the vector of sector gross output that satisfies the given vector of sector final demands is then calculated again. The difference before and after the removal, in terms of gross output volume indicates the importance of the extracted sector relative to the rest of the economy (Dietzenbacher & Lahr, 2013). When sectors are large and highly interconnected, i.e. key sectors, the extraction of that sector will affect many other sectors and thus it will also greatly affect total gross output (Dietzenbacher & Lahr, 2013). Several forms of HEM exist. Miller and Lahr (2001) analysed seven different methods and concluded that the total linkage based on HEM is an appropriate measure to determine key sectors and that, in order to determine key sectors, there is no need to further separate in backward and forward linkages. To better explain HEM, recall equation (2.5). This equation can also be written in the following way in a two-sector economy, where sector 1 is the sector that has to be extracted and sector 2 consists out of the remaining sectors in the economy:
* + [
] * + [ ] (3.5)
Removing sector j from the economy involves setting the rows and columns of sector j to zero. It is then assumed that, due to the fixed-input-coefficient assumption, imports will substitute sector j‟s sales to the rest of the economy perfectly, so that the disappearance of the sales will not affect the rest of the economy‟s production (Cai & Leung, 2004). In the two sector economy, where sector j is sector 1, this would result in the following equation (Miller & Lahr, 2001):
[̅̅̅̅̅̅] [
] [
̅̅̅
̅̅̅] [ ] (3.6)
Where ̅ indicates that this is gross output after the extraction. Linkages can then be measured by taking the difference in gross output before and after the extraction, i.e. the difference between equations (3.5) and (3.6).
3.2 Traditional measurements versus hypothetical extraction
18
The traditional measurements suffer from several drawbacks. First, both the direct forward and the direct backward linkage only measure the direct effect and neglect the indirect effects (Miller & Blair, 2009). While the total linkage does incorporate the indirect effects, the problem with this measure is that is assumes uniform final demand changes: it does not bring out the relative importance of various sectors in the economy (Cai, & Leung, 2004). Furthermore, Oosterhaven (1988) shows that the results gained by the forward linkages based on the Ghosh model are nonsensical. This can be explained by the following example, which was given by Dietzenbacher (1997). Suppose that the value added in sector j is increased by one unit. Using the Ghosh model, this would result in an increase of the output in each sector. Hence, in any sector other than the jth, the production is increased without any increase in the value-added terms. Additionally, Dietzenbacher (1997) shows that results are indeed nonsensical unless the Ghosh model is interpreted as a price model. However, this model is based on price changes whereas quantities all remain the same. Since price effects are not what we want to measure in this study, the Ghosh model cannot be used as a price model, hence traditional methods based on the Ghosh model cannot be used in this study.
Additionally, Andreosso-O‟Callaghan and Yue (2004) compare the traditional and HEM in a study on intersectoral linkages and key sectors in China. They find that HEM is more suited to identify key sectors than traditional methods are, because HEM takes into account the relative magnitude of each sector‟s relative effect on overall output. For this reason and the reasons mentioned above, this study will use the hypothetical extraction method do determine linkages and assess whether the real estate sector can be considered a key sector.
3.3 Total linkage indicator
Based on the HEM, the sector of real estate activities and the construction sector should be extracted from the economy in order to determine their linkages with the rest of the economy. Based on equations (3.5) and (3.6) linkages can be calculated by computing the difference in gross output after extraction. Sector 1 is the sector that is to be extracted, while sector 2 consists out of all the remaining sectors in the economy. This can also be rewritten as:
̅ (3.7)
Tj is then the total linkage and is the difference between gross output before ( ) and after
the extraction of sector j ( ̅ ) in all other sectors in the economy. Again, i is a special
summation vector of 1‟s. Note that the first term, , does not include the original output
from the extracted sector j. Since this original output is omitted, this measures sector j‟s linkage to the remaining sectors in the economy, i.e. its linkage to the rest of the economy (Miller & Blair, 2009). If the original output would not be omitted, a sector with a large output on itself, without using inputs from other sectors, might be falsely identified as a sector with strong linkages to the rest of the economy. In order to normalize this difference this can also be rewritten to a percentage decrease in total output:
̅ ( ̅
19
The interpretation of this formula is: the higher the percentage decrease, the higher the linkages of sector j to the rest of the economy. However, this tells us nothing about the relative strength of the linkage compared to other sectors. In order to also interpret whether a sector has a strong or a weak linkage, additional normalization must follow. This is done by indicating values relative to the average:
̃ ̅⁄∑ ̅ (3.9)
This means that sectors with “strong” (above average) total linkages have indices that are greater than one and those with “weak” (below average) total linkages have indices that are smaller than one (Miller & Blair, 2009). This normalized indicator will be used in order to determine whether the real estate related sectors have strong linkages to the rest of the economy.
The magnitude of the absolute backward linkages is determined by two factors. First, the size of sector j and second its dependence on other sectors and the dependence of other sectors on sector j (Dietzenbacher & Van Der Linden, 1997). In order to obtain a measure of the relative size of the loss, the loss of output in all sectors after extraction is related to the original output of sector j. This measures the gross output value outside sector j created by a one-unit increase in sector j‟s gross output value (Holz, 2011). This is given by the following formula:
̿ ̅ (3.10)
3.4 Linkages based on labour and value added
Linkages will be based on changes in output, which can be considered the standard measure of linkages. Furthermore, the literature has shown that the real estate related sectors are labour-intensive sectors that create relatively high amounts of value added. That is why additional measurements of linkages will also be based on employment and gross value added as shown by Dietzenbacher and Lahr (2013). Changes in gross output due to hypothetical extraction can be translated into employment, for example hours worked or persons employed, using employment coefficients. These coefficients could be persons/hours worked per euro‟s worth of each sector‟s output (Miller & Blair, 2009). Let employment coefficients
be denoted by and total employment:
[ ] (3.11)
Where is total employment in each sector. Changes in output, as measured with equation
(3.7), can then be converted to changes in employment in the following way:
20
The same methodology can be used to measure changes in value added. One just has to replace the indicator for labour for an indicator that measures gross value added.
These linkages can then be normalized in the same way as was done for the change in output with equation (3.8), (3.9), and (3.10). Again, sectors with strong linkages will have indices greater than one and sector with weak linkages will have indices smaller than one. The largest employment impact or largest value added impact do not necessarily have to be in the same sector as the largest change in output. It depends on a combination of changes in output and the size of the employment coefficient (Miller & Blair, 2009).
3.5 Correlation between linkages and property prices
Correlation is a method of assessing a possible two-way linear association between two continuous variables. A correlation coefficient is a useful way to summarize the relationship between two variables with a single number that falls between -1 and +1. The closer the correlation coefficient is to +1, the stronger the positive correlation is and vice versa for -1. There are several correlation coefficients that can be used in order to determine this relationship. Most widely known are the Pearson‟s product moment correlation coefficient and the Spearman rank correlation. The correct usage of the correlation coefficient depends on the type of variable that is being studied. Pearson‟s correlation coefficient is used when both variables are normally distributed, while Spearman‟s correlation coefficient is used when one or both variables are skewed or ordinal (Welkowitz, Cohen & Brooke, 2011). The Shapiro-Wilk test for normality shows that both variables are normally distributed. Furthermore, since I-O data and data on housing prices and number of houses sold are not ordinal, this study will use the Pearson‟s product moment correlation coefficient.
3.6 Data for input-output tables 3.6.1 Data source
Input-output tables are available from different sources and are composed in different ways. Most developed countries have their own statistical offices who compile the input-output tables for their country. For example, in the Netherlands, Statistics Netherlands (CBS) compiles an annual input-output table for only the Dutch economy. Other organizations, like the Organisation for Economic Co-operation and Development (OECD), compile input-output tables for a group of countries. The level of aggregation matters in the identification of sectors with high linkages. With high sectoral aggregation, linkages will also be highly aggregated and as a result, will be less reliable. For example, if the real estate sector is aggregated into a service sector along with other service-based industries, linkages for this aggregated service sector might be high. However, it is not certain that linkages are high due to the real estate sector or any other sector that is aggregated into this big service sector. In general, the input-output tables of national statistic offices are more detailed, in that they split the economy into a larger number of sectors. For example, CBS splits the entire Dutch economy into 76 sectors while the OECD splits the economy of the analysed countries into only 34 sectors.
21
countries, since national statistical offices aggregate at different levels (Song & Liu, 2007). That is why this study uses input-output tables from the World Input Output Database (WIOD). This is an organization that makes harmonized input-output tables for 43 countries, which are split into 56 sectors classified according to the International Standard Industrial Classification revision 4 (Timmer, Los, Stehrer & De Vries, 2016). This is more aggregated than the input-output tables by the statistical offices in, for example, the Netherlands and Belgium, but it is less aggregated than tables by the OECD. Additionally, input-output tables by the WIOD are harmonized between countries, which increases the comparability of linkage indicators between countries. Furthermore, the WIOD only uses data that is publicly available. This way, the users of WIOD are able to trace the construction process. Moreover, officially published data are more reliable as National Statistical Institutes have adopted thorough checking and validation procedures compared to data generated on an ad-hoc basis for specific research purposes (Dietzenbacher, Los, Stehrer, Timmer & De Vries, 2013).
Only domestically supplied inputs should be considered, since the linkages within the domestic economy are of concern (see also section 2.1). That is why, to assess domestic linkages, one should only consider domestic inputs and leave imports out of the analysis (Dietzenbacher, et al., 2005; Reis & Rua, 2006). The WIOD splits imported and domestically supplied input within the inter-industry exchange of products (Z). The WIOD is not the only organization that does this; however, the fact that they also have harmonized input-output tables, which are not highly aggregated, justifies the choice for the input-output tables compiled by the WIOD.
The WIOD has two sets of input-output tables, world input-output tables and national input-output tables. Both have the same number of sectors and cover the years 2000 until 2014. However, the world input-output tables also illustrate all trade between countries, i.e. imports/exports are separated for each country (Timmer et al., 2016). Since it was already established that only domestic inputs should be taken into account, the national input-output tables will be used, which only split inputs between domestic and imported, but does not further split imported inputs to their country of origin. Data in these annual tables is denoted in millions of US dollars.
Additionally, WIOD offers socio-economic accounts which contain industry-level data on employment, capital stocks, gross output, and value added. Since the industry classification is the same as for the world input-output tables, this data can be used to determine linkages based on employment and value added. There are several variables that
can be used as employment coefficients ( ), such as labour compensation, number of
22
3.6.2 Data limitations
The I-O tables by WIOD are classified according to the International Standard Industrial Classification revision 4 (ISIC Rev. 4). The ISIC Rev. 4 is the international reference classification of productive activities, it provides a set of activity categories that can be utilized for the collection and reporting of statistics. Economic activities are split into 21 broad sections but can be split further into detailed classes (United Nations, 2008). Two of the broad sections are at the centre of this study: construction and real estate activities. The WIOD tables, like most other I-O tables, do not split the construction and the real estate activities into smaller classes. This means that data for these two groups is more aggregated than would be preferable for this study.
The construction sector, which is section F in the ISIC Rev.4, includes general construction and specialized construction for buildings and civil engineering work. Additionally, it also includes repair work, additions and alterations, the erection of prefabricated buildings or structures, and construction of a temporary nature. The construction sector can be disaggregated into three divisions: construction of buildings (division 41), civil engineering works (division 42), and specialized construction activities (division 43). For this study the main interest is in division 41, construction of buildings. This division contains the construction of residential or non-residential buildings, which is mainly the real estate to which this study is aimed. Division 43, specialized construction of buildings, contains the construction of part of buildings and civil engineering projects such as demolition, site preparation, and electrical, plumbing and other installation. Part of this division can be contributed to real estate as defined in this study. However, the construction sector in the I-O tables also contains division 42, which is civil engineering work. This contains the construction of motorways, streets, bridges, tunnels, railways, airfields, harbours, sewerage systems etc. (United Nations, 2008). As one might notice, this is not the part of the real estate that is referred to in this study. Unfortunately, there are not major sources for I-O tables that provide tables where construction is further disaggregated. Consequently, calculated linkages for construction in this study will also contain the construction of civil engineering, which might lead to overestimation of linkages for the construction sector related to real estate calculated in this study.
23
3.7 Data on property prices
To measure the property prices, the dataset „Analytical house price indicator‟ by the OECD will be used. The OECD provides real house price indices, which are nominal house prices deflated using the private consumption deflator. Other organizations, like the Bank for International Settlements (BIS), also provide indices for housing prices; however these statistics do not cover the entire period for some countries (like Denmark and Belgium) for which linkages will be determined. The OECD covers the entire period, 2000 until 2014, on an annual base. Furthermore, the housing price indices cover both existing and new dwellings of all types. This means that both transactions in existing real estate as newly constructed real estate can be taken into account (OECD, 2018).
Unfortunately, there is no dataset available that also covers commercial property prices for the specified countries and period. This means that, while linkages have been determined for all real estate, property prices will only be measured through housing price indices. This might influence the correlation between property prices and real estate related linkages.
24
4. EMPIRICAL RESULTS 4.1 Real estate activities linkages
This section will handle the empirical results for the different linkage indicators. The percentage decrease after extraction, the normalized indicator, and the relative indicator will be discussed. First real estate linkages are handled, than construction linkages, and finally both sectors will be combined into one sector.
Table 1 displays the total linkages for the real estate sector for changes in output (based on equation 3.9), for hours worked and for changes in gross value added (both based on equation 3.12) for a few selected years. Full tables for all years can be found in Appendix D. Recall that the strength of the linkage is based on the average of all other sectors in the same economy. Sectors with strong linkages have indices above one and sectors with weak linkages have indices below one. Furthermore, table 2 illustrates the change after extraction of the real estate sector in percentages (based on equation 3.8) and, in parentheses, the rank of this change compared to other sectors within the economy.
All countries have indices above one for all three indicators (output, hours worked, and value added). This indicates that the sector for real estate activities can be considered a sector with strong linkages. However, there are big differences between countries. While Sweden and the Netherlands have very high values for all three linkage indicators, Belgium barely has indices above one. This is also illustrated by the ranks in table 2, the real estate sector is the sector with the biggest change in Sweden for the entire period, while in Belgium the real estate sector is ranked outside of the top ten for the first five analysed years.
At the start of the financial crisis in 2007/2008, Ireland is the only country that shows a small decline in the strength of its real estate activities linkage. Moreover, most countries actually show an increase in the strength of the real estate linkages after the start of the financial crisis, the Netherlands in particular shows a strong increase. Since housing prices started decreasing in the Netherlands, and other analysed countries (see Appendix A), after the start of the crisis, it seems that housing prices and the strength of real estate linkages are not correlated. This is confirmed by table 10, which illustrated the correlation coefficients of the calculated linkage indicators and housing prices. The correlation coefficients between housing prices and all three real estate linkage indicators (output, hours worked, and value added) are not significant. Additionally, the scatterplots in Appendix C illustrate that there is also no nonlinear relationship between real estate linkages and housing prices. This might be due to the different reactions transaction volumes and housing prices exhibit after a demand shock as explained in section 2. It could also indicate that higher brokerage fees due to higher price levels do not significantly increase gross output in the real estate sector.
25
the rest of the economy. This in turn would affect the size of the extraction from the input-output tables and increase the strength of the linkage.
Value Value Value
added added added
2000 2.665 2.321 2.172 3.273 3.274 3.452 1.289 1.127 1.144 2002 3.974 3.429 3.258 3.326 3.373 3.510 1.498 1.301 1.341 2004 3.882 3.173 3.202 3.338 3.339 3.552 1.845 1.543 1.625 2006 3.388 2.659 2.737 3.330 3.414 3.471 1.505 1.229 1.314 2008 2.891 2.135 2.295 3.127 3.208 3.197 1.445 1.150 1.250 2010 5.706 3.675 6.479 3.553 3.532 3.530 2.317 1.962 2.380 2012 6.263 4.039 7.301 3.552 3.628 3.548 2.480 2.146 2.573 2014 5.854 3.670 6.926 3.740 3.741 3.762 2.617 2.209 2.705
Value Value Value
added added added
2000 2.852 3.025 2.860 2.590 2.408 2.585 4.148 4.144 4.350 2002 2.945 3.093 2.959 2.903 2.736 2.903 4.513 4.562 4.732 2004 3.186 3.231 3.179 3.817 3.700 4.035 4.927 4.788 5.202 2006 3.296 3.371 3.242 2.983 2.895 3.242 4.526 4.545 4.710 2008 3.404 3.342 3.655 1.997 2.109 2.129 4.324 4.439 4.491 2010 3.770 3.510 3.943 3.958 3.359 4.605 5.442 5.402 5.708 2012 3.913 3.592 4.230 3.156 2.956 3.485 5.518 5.562 5.793 2014 4.385 3.859 4.710 3.218 3.218 3.633 5.472 5.270 5.770
Normalization is based on eq (3.9). < 1 is a weak linkage, > 1 is a strong linkage Table 1, normalized link ages for real estate activities
Denmark Ireland Sweden
Output Labour Output Labour Output Labour
Normalized linkages for real estate activities
Netherlands Austria Belgium
Output Labour Output Labour Output Labour
2000 1.4 (5) 1.3 (5) 1.2 (5) 1.6 (3) 1.3 (3) 1.6 (3) 0.7 (13) 0.6 (19) 0.6 (15) 2005 2.0 (2) 1.7 (5) 1.7 (5) 1.6 (3) 1.4 (3) 1.7 (3) 0.8 (12) 0.7 (13) 0.7 (14) 2010 2.7 (1) 1.7 (4) 3.3 (1) 1.7 (3) 1.5 (3) 1.7 (3) 1.1 (5) 0.9 (8) 1.1 (7) 2014 2.8 (1) 1.7 (4) 3.4 (1) 1.8 (2) 1.5 (3) 1.8 (2) 1.2 (5) 0.9 (7) 1.2 (5) 2000 0.7 (4) 0.6 (5) 0.6 (4) 1.0 (5) 1.1 (7) 1.1 (6) 1.1 (1) 2.5 (1) 2.5 (1) 2005 0.8 (4) 0.7 (5) 0.7 (3) 1.4 (4) 1.5 (4) 1.6 (4) 1.6 (1) 2.9 (1) 3.1 (1) 2010 1.1 (3) 0.9 (4) 1.1 (3) 1.5 (1) 1.3 (4) 1.8 (1) 1.8 (1) 3.1 (1) 3.2 (1) 2014 1.2 (3) 0.9 (3) 1.2 (1) 0.9 (4) 0.9 (6) 1.2 (4) 1.2 (1) 3.1 (1) 3.2 (1)
Percentage change and rank for real estate activities
Output Labour Output Labour Value Output
Netherlands added Value added Belgium added Austria Labour Value
Percentage change in output, hours worked, or value added after hypothetical extraction. Rank of the sector is in parentheses, based on size of percentage decrease.
26
Table 3 depicts the relative linkages based on equation (3.10) for a few years, tables for all years can be found in Appendix E. The first column depicts the relative linkages based on output for each country. These indicators can be interpreted as the output created outside the real estate activities sector by a one-unit increase in the real estate activities sector‟s gross output (Holz, 2011). For example, in the year 2000 the Dutch relative output coefficient for the real estate sector is 0.247. This implies that a one dollar change in output of the real estate sector comes with a 0.247 dollar change in output in all other sectors. For the relative output indicators the results are mixed. The Netherlands showed a strong increase, especially after the start of the 2008-crisis, while Ireland illustrated a more volatile pattern with an increase and until 2010 and a strong decrease after 2010. The relative output linkages of the other countries remained fairly stable in the analysed pattern. The strong increase in the relative linkage for the Netherlands illustrates that the total extraction did not only become larger because of the growing output in the real estate sector itself. The increase is also because the relative linkage increased, which also increased input from other sectors for each unit of output in the real estate sector.
The second column in table 3 indicates the relative linkage indicator based on hours worked related to output of the real estate sector. Because this indicator was relatively small, it is based on hundred units of output. Thus, the number of hours worked created outside the real estate sector by a hundred-unit increase in the real estate activities sector‟s gross output. All analysed countries illustrate a decline pattern here. This might be an indication that the inputs used by the real estate sector are becoming less labour-intensive.
27
4.2 Construction linkages
Table 4 illustrates the strength of the total linkages for the construction sector, again based on changes in output, hours worked, and value added (see Appendix D for all years), while table 5 depicts the percentage decrease and the rank of the change compared to all sectors in the economy. The indices for all three indicators are well above one, which indicates that the construction sector is a highly linkages sector. This is underpinned by table 4, which shows that the construction sector is the highest ranked sector based on percentage change in most countries.
In the analysed period, construction linkages for all three indicators are increasing slightly for Belgium and Sweden, while the Netherlands, Austria, Denmark, and Sweden show a decline after the start of the financial crisis. Especially Ireland shows a strong decrease after 2008, which is when the Irish property bubble burst. Construction output took a major hit when the Irish property bubble burst and housing prices decreased sharply (also see appendix A). This might illustrate a relationship between the strength of construction linkages and property prices. Furthermore, construction linkages for the Netherlands also decreased after the start of the crisis but more gradually. This is in line with the housing prices in the Netherlands, which also gradually decreased (see also appendix A).
This relationship between construction linkage and property prices is confirmed by table 10, which depicts the correlation coefficients between housing prices and the three indicators of the construction linkages. Indicators based on output, hours worked, and value added have a correlation coefficient of 0.524, 0.528, and 0.544 respectively and they have a significance level of 1%. This gives an indication that there is a strong positive relationship
Value Value Value
added added added
2000 0.247 0.302 0.106 0.234 0.337 0.135 0.128 0.113 0.051 2002 0.359 0.393 0.156 0.233 0.313 0.131 0.143 0.118 0.057 2004 0.317 0.242 0.105 0.227 0.214 0.095 0.168 0.096 0.052 2006 0.273 0.183 0.088 0.226 0.192 0.089 0.143 0.073 0.042 2008 0.240 0.124 0.065 0.229 0.154 0.073 0.135 0.055 0.033 2010 0.439 0.182 0.192 0.239 0.168 0.083 0.190 0.084 0.064 2012 0.488 0.200 0.224 0.235 0.166 0.084 0.200 0.089 0.069 2014 0.441 0.161 0.193 0.228 0.149 0.080 0.199 0.082 0.067
Value Value Value
added added added
2000 0.212 0.285 0.848 0.227 0.324 0.111 0.286 0.374 1.285 2002 0.214 0.261 0.822 0.223 0.279 0.105 0.303 0.396 1.471 2004 0.220 0.186 0.651 0.295 0.251 0.109 0.332 0.290 1.258 2006 0.241 0.181 0.669 0.270 0.200 0.099 0.317 0.258 1.154 2008 0.259 0.148 0.654 0.202 0.120 0.055 0.310 0.215 0.976 2010 0.245 0.145 0.696 0.318 0.164 0.125 0.366 0.268 1.324 2012 0.256 0.142 0.762 0.199 0.119 0.087 0.366 0.248 1.248 2014 0.275 0.144 0.788 0.173 0.101 0.073 0.359 0.230 1.261
Based on output, hours worked, value added change after extraction related to the total output in the combined sectors. Denmark and Sweden value added is in DKK and SEK respectively. Labour indicators are based on 100 units of output, output and value added indicators are based on 1 unit of output. Table 3, relative real estate link ages
Netherlands Austria Belgium
Output Labour Output Labour Output Labour
Output Labour Output Labour Output Labour
Denmark Ireland Sweden
