0
extraction induced earthquakes:
The architectural design of a software valuation mechanism
MSc DD
Operations
& Supply ChainManagement - University of Groningen & Newcastle
ANNE-MARIJN KAMPS
STUDENT NUMBER S2415615/ B3059601
EMAIL A.KAMPS@STUDENT.RUG.NL
FIRST SUPERVISOR DR. H. BALSTERS SECOND SUPERVISOR DR. C. HICKS
DATE 22-12-2015
1 ABSTRACT
This research investigated the effect of induced earthquakes on the property value. Even though natural gas extraction is important for the Dutch economic prosperity, it caused induced earthquakes impacting the real estate value. The NAM, responsible for extracting the gas, had to provide property damage compensation. Immaterial damage due to earthquakes was difficult to define. The NAM did offer a compensation process for property depreciation. However, this process was determined to be unsatisfactory according to literature. The goal of this study was therefore, to establish the design of an alternative property compensation process capable to determine depreciation due to induced earthquakes according to the stakeholders’ needs. A design science philosophy was adopted, because it offers guidance to construct an artifact according to the stakeholders’ needs. A case strategy was applied by incorporating 22 cases (based on 20 interviews and two focus group sessions) to identify the needs of experts and affected inhabitants for an alternative process. Main results of the research identified five critical success factors/needs: fairness, objectivity, correctness, transparency and efficiency (the systems’ goal remained the same). Moreover, results identified that difficulties in addressing the problems that prevent an alternative compensation process from functioning according to the needs, relate to: calculation methods, stakeholders and information. An important insight of this research was the identification and comparison of applicable calculation methods to define depreciation/determine applicability. The hedonic- and hybrid method could be used to appraise regular properties and the comparison method for extraordinary properties. Besides, depreciation characteristics for the hybrid method were identified. Moreover, alternative process’ stakeholders were identified: the affected inhabitant, legal expenses insurance, WAG and an appraisal expert. With the results the solution: an alternative compensation process, was designed and validated. The comparison of calculation methods to determine depreciation offered new insights for the real estate knowledge base. However further quantitative research is necessary to verify applicability. Besides new insights were offered for design science. For conducting explorative research in order to design an artifact, combining the regulative cycle of van Strien (1997) (a design approach) with a case study strategy has proved to offer the means to carefully outline an explorative research. Finally, in order to design a software system, process modelling was only one vital part, data design and interaction design still need to be researched.
Monitoring real-estate value affected by gas
extraction induced earthquakes:
The architectural design of a software valuation mechanism
MSc DD
Operations
& Supply ChainManagement - University of Groningen & Newcastle
Master thesis
By
2
ACKNOWLEDGEMENT
The topic of induced earthquakes and property value is one of the most highlighted current issues at the time of writing in Groningen. Because of all of the parties involved, limited relevant literature and the context which is currently in a state of constant change, researching this subject has proven to be one of the most challenging projects conducted.
First I would like to express my sincere gratitude to my supervisors Dr. H. Balsters and Prof. C. Hicks for their excellent guidance, tips, feedback and encouragement at various stages of my research. Besides, I would like to thank Prof. Dr. J.P Elhorst for his keen interest and the time he took to discuss the state of research related to mathematical models to determine depreciation. Also, Prof. Dr. Ir. G.R.W de Kam, with his expertise on housing and land markets, thank you for raising important issues and providing me with helpful feedback and information.
Without the cooperation of all the inhabitants of Groningen who participated in this study, designing and validating an appropriate solution, would not have been possible. I would therefore, like to thank all the participants for the time they took to express their opinion related to the topic during the interviews. Without their input, designing a mutual acceptable solution would not have been possible. Besides, I would also like to thank the participants who accompanied the focus group sessions held at the University in Groningen. The validation of the solution offered in this report is very important and would not have been feasible without them.
I would also like to thank all of the experts who participated. Realtors and appraisers, thank you for the inside knowledge on property taxation and tips on designing an objective compensation method/process. Also thank you for taking the time to validate the outcome of the research. Finally, I really appreciated the input of the following important stakeholders and experts: Nationaal Coördinator Groningen, municipality Loppersum, the WAG, engineering experts and de Haan law firm. Their input and recommendation proved to be important and provided value-adding insights.
Finally, I really appreciated the personal support, ideas and assistance of my partner, parents, brother, family, friends and colleagues. With the help and support of everyone mentioned above, I was able to finalize my Msc. Thesis.
Thank you,
3
TABLE OF CONTENT
1. Introduction ... 7
2. Theoretical framework ... 11
2.1 Real estate literature ... 11
2.1.1 Real estate appraisal in general ... 11
2.1.2 Real estate appraisal methods ... 12
2.1.3 Inapplicable appraisal methods ... 13
2.1.4 Applicable appraisal methods ... 13
2.2 Modelling literature ... 17
2.2.1 Conceptual modelling in software engineering ... 17
2.2.2 Modelling from a process perspective within software engineering ... 18
2.2.3 Business Process Model & Notation (BPMN) ... 18
3. Methodology ... 20 3.1 Research philosophy ... 20 3.2 Role of theory ... 20 3.3 Time horizon ... 21 3.4 Research strategy ... 21 3.4.1 The approach ... 21 3.4.2 The strategy... 22 4. Results ... 26 4.1 Design problem ... 26
4.1.1 Problem context & scope ... 26
4.1.2 Stakeholder analysis ... 27
4.1.3 Problems/critique and needs of the stakeholders ... 28
4.2 Diagnosis/analysis ... 30
4.2.1 Calculation methods to apply ... 30
4.2.2 Stakeholders to involve ... 33 4.2.3 Information to acquire ... 34 4.3 Design solution ... 35 4.4 Design validation ... 40 5. Conclusion ... 43 6. References ... 45
Appendix I Description of the Case ... 53
Appendix I.I Parties related to natural gas extraction ... 53
Appendix I.II Stakeholders related to compensation process of induced earthquakes ... 53
4
Appendix I.IV Research related to the sale compensation proces ... 56
Appendix I.V Research related to the algorithm expressing property valuation ... 58
Appendix II Sale compensation process as proposed by Bakker ... 59
Appendix III The interview protocols ... 61
Appendix III.I Interview protocol realtors ... 61
Appendix III.II Interview protocol affected inhabitants ... 63
Appendix III.III Interview Protocol WAG ... 64
Appendix III.IV Interview protocol NCG ... 65
Appendix III.IV Interview protocol municipality Loppersum ... 66
Appendix IV. The focus Group sessions ... 67
Appendix IV.I Focus group protocol ... 67
Appendix IV.II The focus group script/ handout ... 67
Appendix V research Transcripts... 77
LIST OF TABLES
Table 1 "Inapplicable appraisal methods" ... 13Table 2 "The case selection" ... 22
Table 3 "Method, goal, source & instrument" ... 23
Table 4 "Quality aspects of the research" ... 24
Table 5 "Stakeholders identified" ... 27
.Table 6 "Problem identification current compensation process" ... 28
Table 7 "Variables to account for the effect of induced earthquakes" ... 32
Table 8 "Pros & Cons Calculation methods ... 33
Table 9 "Stakeholders to involve" ... 34
Table 10 "Information sources" ... 35
Table 11 "Validation remarks" ... 40
Table 12 "Stakeholders, goals & CSF within the sale compensation process" ... 57
5
LIST OF FIGURES
Figure 1 "Relation between gas extraction and occurrence of earthquakes" ... 7
Figure 2 "Software Design Framework" ... 8
Figure 3 "Types of value” adapted from (Lank, et al., 2001) ... 11
Figure 4 "Appraisal methods" ... 12
Figure 5 "Value estimation process Marktpositie" adapted from (Op 't Veld, et al., 2009) ... 15
Figure 6 "The development process of conceptual & computational models" adapted from (Dieste, et al., 2001) ... 18
Figure 7 "Elements of BPMN" adapted from (Thomas, 2015) ... 19
Figure 8 "Saunders research onion"... 20
Figure 9 "Steps to design a BPMN" ... 21
Figure 10 "Deductive & Inductive coding Atlas.ti" ... 25
Figure 11"Codes & families Atlas.ti" ... 25
Figure 12 "Analysing codes & quotes" ... 25
Figure 13 "Alternative compensation process aggregated" ... 36
Figure 14 "Alternative compensation process" ... 37
Figure 15 "Property appraisal processes" ... 38
Figure 16 "Hedonic pricing process" ... 38
Figure 17 "Hybrid pricing process" ... 39
Figure 18 "Price comparison process" ... 39
Figure 19 "WOZ check for Hybrid method" ... 41
Figure 20 "Including: Expert validation" ... 42
Figure 21 "Stakeholders natural gas extraction" adapted from (Dutch Safety Board, 2015) ... 53
Figure 22 "The stakeholders" ... 55
Figure 23 "Sale compensation process" adapted from (NAM, 2015) ... 56
6
ABBREVIATIONS
A
ANN Artificial Neural Network ... 12 ARIMA Autoregressive Integrated Moving Average ... 12
B
BPD Business Process Diagram ... 17 BPMN Business Process Model & Notation ... 9
C
CM Conceptual/ descriptive Model ... 16 CpM Computational-/ prescriptive Model ... 16 CSF Critical Success Factors ... 25
G
GIS Geographical Information Systems... 15
I
IT Information Technology ... 16
M
MIDAS The NVM database system ... 14
N
NAM Dutch Petroleum Company ... 7 NCG Nationaal Coördinator Groningen ... 26, 53 NVM Nederlandse Vereniging van Makelaars, the Dutch Realtors' Association ... 14
P
PFD Process Flow Diagram ... 17 PGV Peak Ground Velocity ... 15
S
SSoM State Supervision of Mines ... 7
U
UML AD Unified Modelling Language Activity Diagram ... 17
W
7 0 10 20 30 40 50 60 70 80 90 100 110 120 1 9 8 6 1 9 8 8 1 9 9 0 1 9 9 2 1 9 9 4 1 9 9 6 1 9 9 8 2 00 0 2 0 0 2 2 0 0 4 2 0 0 6 2 0 0 8 2 0 1 0 2 0 1 2 2 0 1 4 N um be r o f e ar thq ua ke s Year
Occurrence of earthquakes in the province of Groningen
3,5 and higher* 3 till 3,5* 2,5 till 3* 2 till 2,5* 1,5 till 2* 1 till 1,5 ML 0,1 till 1* * Richter scale
Figure 1 "Relation between gas extraction and occurrence of earthquakes"
1. INTRODUCTION
The extraction of natural gas is an important source of Dutch economic prosperity. 15 billion euros was realised in 2013. This accounted for approximate 10% of the total government revenues collected by tax, partnerships with gas extraction companies and other investors (Gas in the Netherlands, 2014). However, despite the economic benefits of gas extraction, negative consequences were present. Segall, et al. (1994) demonstrated the presence of a relationship between gas extraction and the occurrence of earthquakes. Gas extraction and occurrence of earthquakes in the Netherlands received much national and international publicity (Holligan, 2013; Kossen, 2015). Groningen, one of the northern provinces, is located above Western Europe’s largest natural gas field (Oil & Gas portal, 2015). Gas extraction started in 1963 and after two decades (1986), the first earthquake with a magnitude of 2.7
(Richter scale) was noticed (Dutch Safety Board, 2015). Figure 1 represents the relation between gas extraction and occurrence of earthquakes. Noticeably, both the number of earthquakes and the magnitude of earthquakes has risen since 1986. An important concern related to this trend was the impact of these earthquakes on property value. The Dutch Petroleum Company (NAM) has a monopoly for extracting natural gas from the Groningen gas field. The Dutch government decided that the NAM should provide compensation to victims experiencing earthquake damage to their properties. In general property damage was defined as: “physical damage to or
8
Figure 2 "Software Design Framework"
There is a significant social importance to investigate the matter of induced earthquakes and the relation to property value in more detail. Hagoort (2015) determined that 1,100 earthquakes were still going to occur, even if the government will stop the gas extraction (NRC, 2015; Hagoort, 2015). Besides, an investigation of the State Supervision of Mines (SSoM) concluded that the maximum magnitude of an earthquake cannot be determined (SSoM, 2013). Results indicated that a magnitude of 6 on the Richter scale could be possible which would cause major damage to properties in the earthquake zone. The fear and anxiety among inhabitants of the earthquake zone were also researched by prof. G. de Kam (2014). The research objective was to investigate the views of the inhabitants regarding the effects of earthquakes on their living comfort and house values in Groningen. Fear and anxiety dramatically shifted between 2009 and 2013. People indicated that their living comfort and house values were affected. Half of the inhabitants who participated even indicated that they wanted to move outside the earthquake zone, just to feel safer. However, moving is often not feasible. This due to the public image, especially in the eastern part of Groningen. Sales dropped from 1,531 properties in 2007 to 647 properties in 2013 which is a decrease of 58% in comparison to the average national decrease of 41% (Ellenbroek, 2014). Local realtors (real estate agents) indicated, that especially the non-regional market diminished for the earthquake zone both in the number of sales and transaction prices. The combination of problems mentioned above has created the situation where properties were not sold easily or were unsalable (DVHN, 2014). People affected by this situation cannot apply for compensation because the NAM only offers this option after a property is sold (NAMPlatform, 2014).
The practical contribution of this research was to offer an objective and independent alternative for the current NAM compensation process. However, legal ground would be necessary to realise such an alternative process. An important breakthrough on legal ground was established on the 2nd of September 2015. The court in Assen ruled that the NAM should compensate property owners even if they did not sell or are not going to sell their property (De Haan, 2015). This verdict raises the opportunity to develop and implement an alternative compensation process in contrast to the current compensation process offered by the NAM. The lawsuit was led by de WAG, a foundation for affected individuals and housing cooperatives represented by law firm: De Haan. However, the NAM appealed and the verdict proceeds in December 2015. So, for now, the current process offered by the NAM was the only viable option for affected parties to be compensated for property depreciation due to induced earthquakes. But research is necessary to offer the affected parties an independent and objective compensation process if/when the WAG will be successful during the appeal. The purpose of this research was to shape the design of an information system that could provide compensation for property
depreciation related to induced earthquakes. Zachman (1987) presented a classical paper discussing a framework of an information systems’ architecture. He indicated that the architecture should consist of a: process, data and networking. Hevner & Chatterjee (2010, p. 88) expanded this framework, which is presented in Figure 2. Bakker (2015) focussed on the process, stakeholders and their critical success factors. His study was, therefore, part of the software architecture design. Peetsold (2015) investigated an algorithm that could be used to determine the depreciation of
Research scope: Bakker (2015)
Scope of this research
Research scope: Koster & van Ommeren
9 properties affected by earthquakes and was therefore related to: the algorithm design. Figure 2 also outlines the scope of this research. The design of Bakker (2015) was evaluated and an alternative process proposed (software architecture design). Besides, an analyses of different methods together with their strengths and weaknesses to determine depreciation shall be outlined (algorithm design).
The relevant literature covering the topic of natural gas induced earthquakes and property valuation was scarce. This was mainly because the setting under which earthquakes occurred in the Netherlands were not common. Three main differences between the situation in the Netherlands compared to other countries were: the earthquakes that take place were induced, so they were caused by human activity. Second, property owners in the earthquake zone did not share in the profit of gas extraction directly. Lastly, the area is densely populated (Koster & van Ommeren, 2015). After the strongest earthquake due to gas extraction in 2012, the Dutch government initiated a series of investigations. Francke & Lee (2014) conducted research on behalf of the government that determined that properties will not statistically suffer from a decrease in property value by induced earthquakes. However, Koster & van Ommeren (2015) concluded the opposite in their research, even if properties damaged by earthquakes are repaired, immaterial damage will cause a depreciation of the property value. The University of Groningen also started to conduct a series of studies to determine an appropriate design of a valuation mechanism. The research of Koster & van Ommeren (2015) was verified with available data up to 2014. Besides the current compensation process for the depreciation of properties offered by the NAM was explored.
There was a gap in literature when considering the different methods available to determine depreciation of properties due to induced earthquakes. An overview of different methods able to determine the depreciation in relation to induced earthquakes would make a valuable contribution to the field of real-estate literature. A comparison of different methods together with the strength and weaknesses was analysed. This gave inside information on how the compensation process should be arranged. This because, certain methods could require the compensation process to be executed differently. Some methods could require the use of experts e.g. realtors to determine the depreciation/compensation. The calculation methods applied could, therefore, influence the parties involved in the compensation process and accordingly, change the process itself. Besides, some methods could be more suitable for certain situations than other methods. Another theoretical contribution will be to design science within the field of information systems. The research demonstrated the application/appropriateness of Business Process Model & Notation (BPMN) as a modelling language to develop the architectural blueprint of the compensation process. Moreover, insight was presented in the appropriateness of combining the regulative cycle of van Strien (1997) with a multiple case study, to improve exploratory research in design science.
The current problem and gap in literature described above resulted in the following research question:
How to design an alternative process blueprint of a property appraisal
mechanism capable determine induced earthquake depreciation
according to the stakeholders’ needs?
11
Figure 3 "Types of value” adapted from (Lank, et al., 2001)
2. THEORETICAL FRAMEWORK
As mentioned in the introduction, the theoretical field of property appraisal related to induced earthquakes was limited. Theory related to real estate was therefore, explored to offer valuable insight. First, real estate valuation, in general, is outlined. Next, different appraisal methods available are discussed. Moreover, inapplicable methods and the link of the applicable methods to property valuation in relation to induced earthquakes is reviewed.
In contrast, to literature covering property appraisal, a part of the theory section outlines modelling related literature. This was necessary to develop the actual artifact during the “design solution” stage. The second section of the theoretical framework, therefore, considers conceptual modelling in software engineering, how software engineering deals with process modelling and BPMN as the selected process modelling technique. 2.1 REAL ESTATE LITERATURE
2.1.1 REAL ESTATE APPRAISA L IN GENERAL
A lot of research has already been conducted on general real estate topics. A central theme within this type of research was the valuation of real estate. But before the term valuation is addressed, a more specific definition of real estate is necessary. Real estate was defined by Gaddy & Hart (2003, p. 12) as: “the earth’s surface extending downward to the centre of the earth and upward into space, including all things permanently
attached to it by nature or by people”. To evaluate real estate, the liabilities and benefits from owning were often translated to a quantitative value. However, the value of real estate was often debated in real estate literature (Floyd & Allen, 2002). Is it the worth to an individual investor, or society etc.? An appraisal could be conducted to assess the value of real estate. Figure 3 outlines the different property values and highlights the two types of values analysed within this research (market- and depreciated value) Lank, et al. (2001) indicated that the market value was the most frequently sought value. They defined it as, “the probable price a property will bring in a competitive and open market, offered by an informed seller and allowing a reasonable time to find a purchaser” (Lank, et al., 2001, p. 234). This term is closely related to market price that was explained as a synonym for the selling price. Generally, in ideal circumstances, the market price should reflect the market value.
12
Figure 4 "Appraisal methods"
the appraisal institute (1994) indicated, the sales comparison method to be the most applied approach during an appraisal. However, there were more property valuation approaches. The next section covers different appraisal methods available according to literature and presents a link to induced earthquakes.
2.1.2 REAL ESTATE APPRAISA L METHODS
Pagourtzi et al. (2003) conducted a research on the appraisal of real estate. Their research presented a review of the different valuation methods available to perform an appraisal. Methods could either be traditional or advanced. An overview of the different methods within the two categories is presented in Figure 4. Not all of the mentioned appraisal methods were applicable/favourable for this research. The next section will elaborate on the applicable and less applicable methods and explain why certain approaches were favourable/ less favourable for this research.
13 2.1.3 INAPPLICABLE APPRAISAL METHODS
The appraisal methods that were not further considered during this research are highlighted red in Figure 4. The traditional inapplicable methods should be applied when there are no likely buyers for the current state of the property or when the objective is to establish the rental value for a property (in either way, the comparison method is not optional), see Table 1 for details (Armatys, et al., 2009).
Table 1 "Inapplicable appraisal methods"
Inapplicable appraisal method
Explanation Investment/income
method.
The investment method was characterised by analysing the potential rental income for a certain property. This rental income should afterwards be translated to the market value that an investor would pay for the property. Profit method. Used to determine the property rental value, instead of property capital value. The property is viewed as an unit of
production. The market value will be determined by analysing the potential revenue that property could produce. So, this is related to the worth of a property for a business market.
Development/residual method.
Determines the market value by assessing the object in a redeveloped form. This is performed by starting with the situation where the current use of a property and land is not optimal. A redevelopment analysis should be conducted. The redeveloped form can afterwards be assessed by the comparison or investment method mentioned above. In the last step, the development costs should be deducted.
Contractor’s/cost method.
This method could be applied when a property is intrinsically linked to the business carried out in the property. Often used for special properties that e.g. are designed to conduct certain manufacturing processes. For these type of properties there is often no existing market, so a (opportunity) costs logic should be applied as an indicator for the market price. It is based on the assumption that the buyer will not pay more for the property than the costs of buying a similar new property. Armatys, et al. (2009) indicated that this method should be used as a last resort. The market value should not be determined by the cost of production but rather by supply and demand forces. Artificial Neural
Networks (ANNs).
Just like the hedonic method, the ANN is an example of a mass appraisal method. This artificial intelligence model is designed to replicate the learning processes which also occurs in human brains. It first needs training with a certain dataset of a property market. Afterwards, it could be implemented to estimate market value for the properties of that market. However, one important drawback of this method is the “black box” nature (Kauko & d'Amato, 2008, p. 184). This implies that these models have problems in explaining how input data are eventually transformed to an output: the market value (model primary components: hidden layer(s)). However, Tay and Ho (1991/92) do indicate that ANN’s are more capable in determining the market value for an extraordinary property in comparison to the hedonic method. A mean absolute error of 3,9% for the ANN in comparison to 7,5% for the hedonic method. This is due to outlier values (extraordinary properties)
Autoregressive Integrated Moving Average (ARIMA).
ARIMA models are an example of time series based models that also can be applied in situations where the data is not stationary (increasing/ decreasing/ fluctuating market value for properties over time) (Pagourtzi, et al., 2003). These models predict future market values of properties by determining the trend of past realised property transaction values (historical data) and projecting that trend to future values.
For the advanced methods, Artificial Neural Networks (ANN’s) and Autoregressive Integrated Moving Average (ARIMA) as explained in Table 1, were also not further explored. ANN’s were considered to function like a “black box”. So, it is not transparent in how a certain property valuation is realised. Besides, the ARIMA method was not considered because it is less applicable for valuation on a micro level: valuation per property. Finding a small amount of comparable properties to a certain property could already be difficult. To establish an ARIMA model, it is necessary to have large sets of comparable properties sold over a period of time.
2.1.4 APPLICABLE APPRAISAL METHODS 2.1.4.1 THE COMPARISON METHOD & THE NAM
14 transaction values of similar real estate. Schram (2006) indicated two factors that were of critical importance when applying this approach:
1. It is necessary to have a sufficient number of transactions. Besides, transactions should consist of similar and comparable real estate (at least three to five or more similar properties sold in the last six months (Stapleton & Williams, 2004, p. 367)).
2. The market price for comparable properties must be adjusted in order to meet the characteristics of the subject property.
To select similar properties, different similarity functions were available. A similarity function is a function that can quantify how similar two properties are, based on certain characteristics. But as Op ‘t Veld, et al. (2009) pointed out, calculation of similarity is often based on the similarity per characteristic in order to account for differences in the measurement level of a characteristic (e.g. characteristics with an ordinal value need different similarity functions than those consisting of nominal values). After the similarity functions were calculated per characteristic an overall similarity score could be calculated.
The current compensation process offered by the NAM applied a variant of this appraisal method (NAM, 2014; Bakker, 2015). The goal of this variant was to determine the amount of compensation due to property depreciation caused by induced earthquakes. To estimate the amount of compensation, the subject property that was sold in the earthquake zone was first compared with similar properties within the earthquake zone. This, to determine whether there was a difference between the subject property and other properties experiencing depreciation due to induced earthquakes. Next, areas outside the earthquake zone were selected for providing comparable properties that do not suffer from induced earthquake depreciation. 100 reference areas were matched by analysing 150 area characteristics. The reference areas were selected on postal code level to ensure similarity. From the reference areas, similar sold properties related to the subject property were selected based on 37 property characteristics. During the final step, the following was determined. Is there a higher price decline of similar properties (related to the subject property) in the earthquake zone in comparison with similar properties (related to the subject property) in the reference areas? If this holds true, a difference in percentage times the selling price of the subject property was paid to the selling party.
However, there is a risk in applying the current method as proposed by the NAM. Schram (2006, p. 199) highlighted the following regarding location as an element of comparison:
“Ideally, comparable sales should be located in the same neighbourhood as the subject property. A neighbourhood can be defined as an area where market forces operate in a similar way on all properties in the
area. If comparable sales from the subject property neighbourhood are not available, the appraiser may consider sales from nearby neighbourhoods that are subject to similar market influences.”
The comparison method that was applied by the NAM does not follow the instruction above, to compare with nearby neighbourhoods. However, the method does try to control this risk, by selecting neighbourhoods with similar neighbourhood characteristics.
2.1.4.2 HEDONIC PRICING MODEL
Another type of advanced model is the hedonic pricing model. The general theory behind this model explained that a good (e.g. a property) can be valued based on the attributes or characteristics which it possesses (Callan & Thoma, 2007). So for real estate, this implied that the market price of a property could be determined by characteristics such as, the location of the property, size of the property and other aspects influencing the willingness to pay a certain price.
15 the transaction price (Monson, 2009). It is however necessary to have a sufficient sample size in order built a reliable model. Hidano, (2002) indicated that a sample of at least: 100 data points, for evaluating a certain minor feature, should be sufficient. Statistics explained the effect of a certain characteristic on the property price as the correlation (Bruce, 2015). The basic form of regression analysis, is the least squared method. With this method the correlation between two variables could be determined by the influence of the independent variable (X) on the dependent variable (Y). One could for example, by using regression analysis, determine what the effect is of a certain amount of rooms within a house (X) on the transaction price of that house (Y). By collecting historical information on the transaction price of properties sold and the amount of rooms every property has, a trend could be determined. This could result in e.g. more rooms increase the transaction price. The outcome of the mathematics performed during the regression analysis, presents the correlation coefficient of the independent variable. This is explained as a constant number which indicates a measure of some characteristic. The coefficient always fluctuates between: +1 (a perfect positive correlation) and -1 (a perfect negative correlation). 0 indicates that no correlation exists between the independent and the dependent variable (Bruce, 2015). For hedonic modelling within real estate, a multiple regression analysis was often conducted. So, the effect of several independent variables (X’s) on the transaction price (Y) was determined simultaneously. It is important to note that, every value that a characteristic can obtain, will receive a certain correlation coefficient (e.g. type of property could be: detached, terraced or semi-detached, so every option will receive a coefficient).
Monson (2009) indicated the following benefits of applying the hedonic model in comparison to other appraisal methods:
1. Establishment of a property valuation in absence of a market (the less favourable traditional methods explained above do not have to be applied).
2. The ability to quantify, quantitative but also qualitative terms (nice location, nearby facilities etc.). 3. One could value a property for non-revenue generating circumstances.
2.1.4.3 HEDONIC MODEL APPLIED IN THE NETHERLANDS: “MARKTPOSITIE” To develop a hedonic model, the coefficients generated during the
regression analysis served as an input. A real life example of a hedonic valuation model web application was presented in the research of Op ‘t Veld, et al. (2009). The NVM realtors (a Dutch realtor’s association, see section: Appendix I.II Stakeholders related to compensation process of induced earthquakes) which cooperated during this research, had the opportunity to apply this web based application. For every transaction that the NVM realtor closed, they made sure that 69 property characteristics were updated in their database system: MIDAS. This database contains all sold properties from 1985. For realtors within the earthquake zone, around 85% were connected to the NVM. Annually 200,000 houses were offered nationwide by the NVM and the database already exceeded 3 billion houses in total (Op 't Veld, et al., 2009). For the year 2014, the database contained 4,851 sold properties within the earthquake zone (Peetsold, 2015).
The steps conducted to determine the value of a property by the web application are presented in Figure 5. The 69 characteristics of a property are first determined. Afterwards, like explained at the comparison method, the similarity of the subject property is compared with similar closed transactions from the Midas Database. After the
similarity functions are applied, the similar properties are ranked. Next, Figure 5 "Value estimation process
Marktpositie" adapted from (Op 't Veld, et al., 2009) Midas Database: Houses sold Estimated market value Characteristics of the house
Find similar houses
Apply similarity function
Rank similar houses according to similarity
Check: autocorrelation/ heteroscedasticity
Hedonic regression
16 the set of similar properties are analysed on two factors. The first is heteroscedasticity. When this occurs, it implies that extreme outliers are present in certain property characteristics which create biased results. So, either the characteristic(s) need(s) to be deleted or the property itself should be deleted. Also, autocorrelation is an important unwanted factor. This indicates the occurrence of multi-collinearity. Certain independent variables are highly correlated if multicollinearity occurs (correlation coefficient > 0.60). To resolve autocorrelation, the less important independent variables need to be excluded from further analysis to prevent biased results. With the final set of properties and their characteristics, the hedonic regression could be applied to establish the regression coefficients per characteristic. Finally, the value of the property can be estimated by multiplying the target property characteristics with the established coefficients.
2.1.4.4 HEDONIC PRICING MODEL AND INDUCED EARTHQUAKES
An interesting research related to induced earthquakes and hedonic property valuation was conducted by Koster & van Ommeren (2015). The outcome concluded that a consequential effect of gas extraction was the depress of property value even when the owners were fully compensated for the material damage caused by the induced earthquakes. When an earthquake with a magnitude of 2.2 or stronger on the Richter scale occurred, the market value would drop in an area up to 10 kilometres from the epicentre1. This translated according to Koster & van Ommeren (2015) in a value decrease of 1.2% or on average €3,000.- per property every time an earthquake appeared. In reaction to the outcome of their research, Peetsold (2015) conducted the same research but included the last known available NVM property data (over 2014). He concluded that the influence of earthquakes on property value was significantly negative. Moreover, he also transformed the outcomes of Koster & van Ommeren (2015), which presented depreciation per square meter to the depreciation of an entire property. This adjustment still offered the ability to forecast property value and the effect of induced earthquakes. Peetsolds (2015) explained that an earthquake with a PGV2 > 0.5cm/s, will cause a value decrease of €3,787.- euro per property.
2.1.4.5 SPATIAL ANALYSIS
The introduction of spatial analysis was important for property valuation in relation to induced earthquakes. This because it could improve the capability to predict the property value when applying certain appraisal methods. Spatial analysis related to real estate could be used to determine geographical influences on the value of a property (Pagourtzi, et al., 2003). Today more sophisticated techniques are available to determine the geographical influences by using a Geographical Information System (GIS). GIS offers the ability to determine how locational factors affect property value. Wyatt & Ralphs (2003) indicated that by applying GIS, valuers can select comparable properties based on similar locational factors instead of having to select properties in the proximity of a certain subject property.
Spatial analysis could be beneficial for developing a more detailed analysis in how induced earthquakes affect the market value of a property. The research of Koster & van Ommeren (2015) was an example where spatial analysis was combined with a hedonic pricing method. By analysing the available information on the location and the magnitude of earthquakes, the researchers were able to determine, how the induced earthquakes affected the property value. Koster & van Ommeren (2015) however, only based their analysis on a number of earthquakes with a certain magnitude.
1The vertical projection of the focal point (hypocentre, were the fault line originates and the earthquake actually occurs at a
certain depth) to the ground surface (Filiatrault, et al., 2013).
2A measure of the earthquake intensity. This measure was selected because it provides the highest correlation with
17 2.1.4.6 FUZZY LOGIC
The Classical bivalent theory explained that objects can either belong to one set, or not, expressed by the characteristic function indicating “1” for belonging to and “0” for not belong to. However, fuzzy logic theory explained that the characteristic function can represent various degrees of belonging to a set (Chi Man Hui, et al., 2009). So, a fuzzy logic based system is a system that could deal with uncertainty and imprecision. Hagoort (2015) determined that an amount of 1100 earthquakes were still going to occur whether gas extraction will stop or proceed. This was an example of a risk/ uncertainty about the future market condition that impacts the depreciated property value. The application of fuzzy logic based systems could provide a solution for dealing with uncertainty and risks related to real estate appraisal. Byrne (1995) presented a research where they propose to apply fuzzy logic when considering the risk and uncertainty related to real estate appraisal.
Regression techniques like the hedonic pricing method explained above use real data. However, this could lead to several sources of error like e.g. simultaneous relationships between the independent variables and incorrect specification of the model format (McCluskey & Walters, 2012). Fuzzy logic systems could provide a solution for this problem because of their ability to develop more flexible and comprehensive system.
2.2 MODELLING LITERATURE
The essence of design science (the applied philosophy within this research) was to identify a problem that can be addressed by an IT artifact (Helfert & Donnellan, 2011). Hevner et al. (2004) described an IT artifact as, any designed solution that solves a problem in a certain context. The regulative cycle developed by van Strien (1997), an often applied framework within design science, described the third step during a research as: design solution. This is, the step where the actual IT artifact should be created. The research question, outlined in the introduction, gave a description of the type of artifact to be designed. Namely, a representation of an alternative property compensation process related to induced earthquakes. In essence, this should represent a process blueprint that can be used to partly determine the design of a software system able to evaluate property depreciation due to induced earthquakes. The following section outlines relevant literature related to process modelling within software engineering to design the artifact.
2.2.1 CONCEPTUAL MODELLING IN SOFTWARE ENGI NEERING
Conceptual modelling was considered to be crucial in the development of software systems. Dieste, et al. (2001) explained that an understanding of e.g. the problem components, relations, rules and constraints are necessary to determine an appropriate design of a software system. The tools capable of creating an understanding of the problem context and the possible solution, are called conceptual models.
For software engineering purposes, models were constructed for two different goals. There are descriptive/ Conceptual Models (CM) and prescriptive/ Computational Models (CpM) (Wieringa, 1995). The descriptive/ Conceptual Models have the purpose to gain an understanding of the reality. They were often utilised during the problem-orientation phase (first stage of software development), where important concepts of the problem domain are defined (Blum, 1996). Examples of traditional descriptive models are: flow diagrams, object models, rules etc (Wieringa, 1995). Prescriptive models define the software system itself, it is the output or requirements, design activities and specifications. The usefulness of prescriptive models was therefore, more related to the second stage of software development.
18
Figure 6 "The development process of conceptual & computational models" adapted from (Dieste, et al., 2001)
the difference between the way stakeholders are experiencing the world and the way they would like to experience the world (Balsters, 2015, p. 12). To develop a modelling process closer related to the user domain (design science philosophy), Dieste, et al. (2001), advised to separate CM’s from the CpM’s. This research took a need-oriented approach and the next section will explain why. See for the essentials of the need orientated approach Figure 6.
2.2.2 MODELLING FROM A PROCESS PERSPECTIVE WITHIN SOFTWARE ENGI NEERING
The introduction outlined how the architecture of a software system is composed. The problem domain was explored to gain an understanding of the current reality and the desired reality related to an alternative compensation process. As Blum (1996) pointed out, descriptive models apply here. Moreover, the focus should be drawn to the first stage of software development: need orientated modelling (see: Figure 6) The “need” which was addressed, was related to modelling processes.
Robinson, et al. (2010) indicated that, to model the flow of processes, a Process Flow Diagram (PFD) was often applied. This type of diagram presents the sequence/ structure of certain activities. Besides, it also presents how entities are treated during the process. Business Process Diagrams (BPD) are one of the most applied process flow diagrams. The notation which is used to develop a BPD is called: Business Process Model & Notation (BPMN). However, Kalnins & Vitolins (2006) indicated that besides BPD, the Unified Modelling Language Activity Diagrams (UML AD) were also popular. Both of the models were known for their ability to define processes in a way that both technical, as well as, business users could read them because of the simple and intuitive nature of the notation (Balsters, 2015). The ease at which the model could be applied however, does not influence to the capability of the model to present complex processes (Robinson, et al., 2010). Geambasu (2012) reviewed both of the modelling techniques (UML AD & BPMN) and determined that BPMN can be classified as a superior modelling language.
2.2.3 BUSINESS PROCESS MODEL & NOTATION (BPMN)
Business Process Model & Notation (BPMN) as explained in the previous paragraph was classified as a superior popular modelling language. This modelling technique was therefore, selected to develop the alternative property compensation process related to induced earthquakes. BPMN was applied during the third stage of the research: “design solution” (see for more information: section 3.4.1 The approach). The most recent version of BPMN is version 2.0, where the notation was based on elements from various previous models & notations (Schalles, 2013). The main elements used by BPMN to construct a Business Process Diagram (BPD) are presented in Figure 7. In general, every blueprint designed consists of flow objects, connecting objects,
MODELLING
Design/ Formalisation Need orientated modelling 2: Software system
1: Need – orientated modelling
What is the need?
How can the need be satisfied in the users’ world? Translate “need” to a CM.
No preconditioned use of a development approach.
19
Figure 7 "Elements of BPMN" adapted from (Thomas, 2015)
20
3. METHODOLOGY
Saunders, et al. (2009) developed a clear conceptual framework for defining the research process, called the research process‘ onion. The structure of this chapter follows the layers mentioned in their framework (see Figure 8). The research onion consists of the following layers, representing the methodological choices to be made: the research philosophy, role of theory, time horizon, strategy and data collection methods. The approach of this research according to the framework is highlighted green in Figure 8.
3.1 RESEARCH PHILOSOP HY
The research philosophy was grounded in design science/constructivism. Moreover, the study took an explorative approach. Design science is closely linked to constructivism, mentioned as a research philosophy within Saunders research onion (Hinkelmann & Witschel, 2013). The philosophy of design science was selected because it is characterised by solving practical-knowledge problems. The goal of this research was not to determine the truth, but rather the utility. This implied, how to design a system in such a way that it will be objective and mutually accepted by the stakeholders involved. Besides, both of the philosophies mentioned the importance of “in context” and the “truth” which exists. Azzopardi stated (2009, p. 213): “the reality that exists is a consequence of the context in which it occurs and is shaped by historical, political, social and cultural norms”. Watzlawik (1984) even pointed out that it is impossible to extract the outcome of a research in order to implement it somewhere else. It is necessary to take the context into account in which the outcome is created. As a result, generalisation of the outcome of this research should be applied with caution. Conducting an explorative study is valuable for the phenomenon of gas induced earthquakes related to property valuation. Robinson (2002) indicated that it gives the opportunity to find out: what is happening, to find new insights, assess the subject and ask questions.
3.2 ROLE OF THEORY
Due to the limited literature available on the design of a system capable of dealing with the depreciation of property value related to induced earthquakes, theory was developed in an exploratory and inductive manner. This implied, that observations were made to find patterns that could be explored and validated to form a theory (Fereday & Muir-Cochrane, 2006). With the inductive process, theory came last and emerged during the collection and analysis of data. Besides, inductive theory building also stressed the importance of generalising within a certain context. If theory would be utilised in a situation different from the context where it emerged, then it is necessary to analyse the extent of applicability and if necessarily adapt the existing theory (Corbin &
Research philosophy
Role of theory
Time horizon
Strategy
Data collection
Constructivism Phenomenology Positivism Feminism Cross-sectional Longitudinal Survey Ethnography Inductive Deductive Case study Experiment Observation Questionnaire Interview Focus group Desk research21 Strauss, 1990). The inductive way of reasoning is therefore closely related and suited for the design philosophy behind this research.
3.3 TIME HORIZON
The research was cross-sectional of nature due to the time constraint. So the phenomenon of induced earthquakes and property valuation was evaluated at a particular point in time instead of observing events related to the subject over a certain period. The main data analysed was primary data retrieved during this research (September 2015 – December 2015).
3.4 RESEARCH STRATEGY 3.4.1 THE APPROACH
The design approach was adopted to investigate induced earthquakes and the development of an alternative compensation process. The regulative cycle developed by van Strien (1997) was often applied within design science research. This framework served as a guideline to structure the result section of this rapport. Figure 9 presents the cycle consisting of five basic steps. The first step in the cycle starts with the design problem. Within this step, it was important to define the: stakeholders, their goals and their critical success factors. A critical success factor indicates a factor that has to be met in order to accept the solution (Balsters, 2015). The next step should focus on identifying the causes that create difficulty in achieving the critical success factors. The design solution marked the point in the cycle where the actual artifact was created. The artifact of this study included a blueprint of an alternative compensation process capable to determine property depreciation related to induced earthquakes. Focus was drawn to the identification of the solution alternatives available, to build a suitable artifact. Implementation was not part of this research scope. Instead, the correctness of the design was validated during two focus group sessions with the affected inhabitants and interviews with experts.
Figure 9 "Steps to design a BPMN"
1
•Determine the problem context and scope. •Conduct a stakeholder analysis.
2
•What causes the difficulty in resolving the issues?
•Is there a certain order in which the issues need to be resolved?
3
•Which solution alternatives are available to build a BPMN model?
•Is it possible to build a proper BPMN model from existing options or is it necessarily to develop a new BPMN model?
4
•Validate the design solution to check if the "need" of the stakeholders is satisfied.
PROBLEM
ANALYSIS
SOLUTION
VALIDATION
According to the stakeholders:
1: What shall the system deliver (goal)? 2: How does the system need to function?
Design the BPMN model.
22 3.4.2 THE STRATEGY
As a research strategy, a case study was selected because of the ability to investigate the phenomenon of property valuation and induced earthquakes in real life context, while incorporating multiple sources of evidence (Yin, 2003). The method was therefore suitable to conduct the research according to the design philosophy mentioned earlier. Besides, it ensured that the criteria of executing the research in an exploratory and inductive manner were met. As Saunders, et al. (2009, p. 146) indicated, a case study is especially capable in answering “how” questions and therefore often used in exploratory research.
A case study strategy, using the property valuation process as a single case was adopted. This because the objective was to design an architectural blueprint of the property valuation process related to induced earthquakes. Within in the single case, the perspective of experts and the view of affected inhabitants were explored in more detail. These two groups were therefore marked as the embedded cases.
3.4.2.1 UNIT OF ANALYSIS
The discrepancy between how stakeholders were experiencing the current compensation process and how they would like to experience the compensation process has to be embedded in the design solution for it to be marked as successful. The unit of analysis resulted logically as the individuals/stakeholders. The stakeholders were defined above as the experts and affected inhabitants (see for more information section: 4.1.2 Stakeholder analysis). Defining the unit of analysis on the same aggregation level as the research question could simplify analysis because if the level of reference is different, it could result in cross-level inference problems (Dansereau & Markham, 1997). When for example individuals are studied, but data is only available on aggregated levels (e.g. postal code level or national level) data can create a limitation to give grounded solutions/conclusions (Tam Cho & Manski, 2008, p. 547). This problem will not occur in this research.
3.4.2.2 CASE SELECTION
Defining the case selection involved two actions (Miles & Huberman, 1994). The boundary of the study had to be defined. This outlined what should and could be analysed. The boundary was the compensation process related to induced earthquakes. Next, a sample frame was composed. By following a replication logic3 instead of a sampling logic4, cases were selected. This way of selecting cases suits when the objective is to build theory from cases (Karlsson, 2009). The choice of cases was based on a stakeholder analysis that is outlined in section: 4.1.2 Stakeholder analysis. Table 2 presents an overview of the cases.
Table 2 "The case selection"
Aspect Amount of cases Extra Information Affected
inhabitants
12 The earthquake zone constituted of 8 municipalities. For every municipality, 10 inhabitants were requested to participate. From the respondents, 25 inhabitants were able to cooperate. Due to the application of focus group sessions and semi-structured interviews the number of cases resulted in 12.
Experts 10 Appraisers within the earthquake zone (7), Municipality Loppersum (1), Nationaal Coördinator Groningen (1), WAG (1), housing and land market specialist (1), Law firm (1).
3 Building a sample frame by selecting cases based on certain criteria (Karlsson, 2009).
4 The traditional sampling process: identification of a population and whereafter a random or stratified sample is selected
23 3.4.2.3 DATA COLLECTION
To collect data, the underlying principle of triangulation was adopted. This implied the use of different methods to study the phenomenon of property valuation and induced earthquakes (Karlsson, 2009). The methods included were: Desk research, semi-structured interviews and focus group sessions. Table 3 presents an overview of the methods, goals and instruments. A timeframe of three months was scheduled to collect the primary data (15th of August to the 15th of November 2015).
Table 3 "Method, goal, source & instrument"
Goal Data source Method Instruments
Initial analysis of the research scope
Published documents Desk Research - Determination of the
stakeholder perspective, goals & critical success factors
Affected inhabitants 10 Semi-structured Interviews
Voice recorder
Determination of the stakeholder perspective, goals & critical success factors
Experts 10 Semi-structured
interviews
Voice recorder
Validation of the design solution
Experts Semi-structured interviews5 Voice recorder
Validation of the design solution
Affected inhabitants 2 Focus group sessions Voice recorder
Semi-structured interviews were selected as the main method to retrieve information. This due to the inductive nature of this research. Besides, interviews are considered to be an appropriate and efficient method of data collection when the research needs to answer a “how” question (King & Horrocks, 2010). To structure the interviews, several interview protocols were created and included in Appendix III The interview protocols (in Dutch). The interview protocols were designed by following the regulative cycle of van Strien (1997), so questions related to the different steps mentioned in Figure 9 were included. General questions related to the stakeholder and the subject, the current claim process, the alternative claim process and methods to define depreciation, were the main themes discussed. Interviews conducted were both used for the problem/analysis and the validation phase.
An important part of design science is also concerned with the validation of the created artifact. Validation in design science could take two forms: a technical or a socio-technical. The first type of validation is concerned with evaluating the performance of the system. Whereas the second type focusses on the overall usefulness of the provided solution to the end-user. This research focused on the second type of validation. The interest was to determine the utility of the artifact. Several validation methods for artifacts have been discussed like, analytics, experiments, observation, testing or descriptive analysis and action research (Baskerville & Myers, 2004; Cole, et al., 2005; Hevner, et al., 2004; Iversen, et al., 2004; Lindgren, et al., 2004). Hevner & Chatterjee (2010) indicated that focus groups are an effective technique to be used for the improvement of an artifact and their utility in the application field. This technique was therefore, applied within the validation phase. The focus group protocol and the handout discussed during the session is included in Appendix IV. The focus Group sessions (partly in Dutch).
5Iterative process (applied during the 10 semi structured interviews): process was discussed and adjusted based on outcomes of the
24 3.4.2.4 RESEARCH QUALITY
To ensure research quality, several dimensions were considered. Karlsson (2009) explained that a research should take four aspects into account to ensure a sufficient level of quality. These quality aspects and a description of how they were incorporated into this research are summarised in Table 4.
Table 4 "Quality aspects of the research"
Quality Aspects Incorporation
Construct validity The operational measures used to measure the concept actually measure the concept they are intended to measure (Karlsson, 2009).
1. Using multiple sources of evidence (for both of the embedded cases (experts and affected inhabitant) several persons cooperated.
2. Seeking triangulation. Adoption of several methods to be more confident with the result. For this research three methods were adopted: desk research, semi-structured interviews and focus group sessions.
Internal validity The research actually measures what it is meant to measure. Besides, the demonstrated relationships are explained by factors described and not by other factors (Karlsson, 2009).
1. Applying replication logic (the participants for this research are selected and not randomly chosen based on the desk research and stakeholder analysis).
2. All of the interviews and focus group sessions are recorded.
3. The experts who participated are selected on knowledge and experience by means of a stakeholder analysis.
External validity/ generalisability Can the findings of the research be generalised beyond the immediate case study? (Karlsson, 2009).
As described in the section, the role of theory and research philosophy, this research is positioned in design science and theory is developed in an inductive manner. Both design science and inductive theory development stresses the importance of generalising findings within context. So if theory will be utilised in a situation different from the context where it emerged, then it is necessarily to analyse the extent of applicability and if necessarily adapt the existing theory (Corbin & Strauss, 1990). So generalisability is possible but proceed with caution.
Reliability The study is objective in the sense that other researchers should come to the same conclusions in the same setting. So is the study repeatable, with the same findings? (Karlsson, 2009).
Reliability is ensured by enabling replication of the research. Extensive description of the following aspects:
1. the case selection, 2. applied methods,
3. way of analysing by coding, 4. the included interview protocols,
5. transcripts (on request),
6. handout of the focus group session, 7. the focus group protocol,
8. analysis instrument (Atlas.ti). 3.4.2.5 DATA ORGANISATION & ANALYSIS
Data analysis started with converting the 27 audio files to transcripts. The resulted transcripts were analysed by applying the process explained by Strauss and Corbin (1998). They described that three forms of coding take place. First open coding serves to group similar items according to some defined property and giving the quotes a descriptive name. This process ensures the reduction of a large amount of data in more manageable pieces. Besides, grouping also helps with identifying dimensions and properties of the categories. After the open coding process, axial coding serves to reconstruct the data that was separated during open coding. The properties and dimensions are used to determine how categories and subcategories relate to each other. Finally, selective coding nurtures the integration and refinement of the categories by organising them around a central category.
However, before the open, axial and selective coding process started, deductive reasoning was applied. Relevant quotes where either labeled as: expert or affected inhabitant. This division simplified the within and cross case analysis.
25
Figure 10 "Deductive & Inductive coding Atlas.ti"
26
4. RESULTS
The four upcoming sections followed the design science approach (see section 3.4 Research strategy). The steps mentioned by the regulative cycle of van Strien (1997) were incorporated. First, results regarding the problem context and scope, including the stakeholders involved shall be outlined. The perspective of the stakeholders regarding the current compensation process was reviewed and their goals and needs (Critical Success Factors, CSF) for an alternative process were analysed. With the opinion of the stakeholders, the diagnosis/ analysis focussed on the difficulties/ issues, in translating the stakeholders’ opinions to an alternative compensation process. The design solution section presents the design of an alternative system that ensured mutual acceptance of the stakeholders. Finally, the validation section discusses the final remarks of the proposed alternative process.
4.1 DESIGN PROBLEM
The problem context and the stakeholders related to induced earthquakes are outlined in this chapter. Desk research identified several stakeholders and their relation to the problem context. However, the next section will first introduce the problem context and scope of this research.
4.1.1 PROBLEM CONTEX T & SCOPE
In general, there was a debate on how to determine depreciation due to induced earthquakes and the design of an appropriate alternative compensation process. The NAM, which extracts the gas in Groningen, has been held legally responsible for compensating damage due to induced earthquakes (De Haan, 2015). Consequently, different compensation processes for material and immaterial damage were established by the NAM. For material damage: Damage and restoration of a property, arrangements for building new properties (preventive action) and a process for the reinforcement of existing properties (preventive action) were established. For immaterial damage: arrangements for energy saving opportunities, arrangements for building new properties (also immaterial, to change the negative image within the earthquake area) and the property sale compensation arrangement (NAM, 2015).
The scope of this research was framed around the sale compensation process. This process offered a selling party the opportunity to file a claim when they sold their property and are inhabitant of one of the eight municipalities that where registered as being part of the earthquake zone6. The compensation provided should cover the loss in transaction value due to the occurrence of induced earthquakes. To define the amount of compensation, Arcadis an appraisal office hired by the NAM appraises the property as it was not located in the earthquake zone. In essence, comparable properties to the subject property (property under valuation) within the earthquake zone and comparable properties to the subject property in reference areas, were selected to analyse a possible price difference. When a negative difference for the transaction price of the subject property was identified, compensation should be offered. The compensation amount is calculated by: the percent difference between the transaction price of the two area’s (subject and reference) times the transaction price of the subject property.
However, the current sale compensation process of the NAM was under debate. The earlier discussed research of van der Voort & Vanclay (2015) determined that the compensation initiatives were inadequate in terms of service and speed. It was explained that greater efficiency in the compensation process was necessary. Besides, the compensation mechanism itself should be independent. Bakker (2015) analysed the current sale compensation process of the NAM in more detail by including several stakeholders7. In summary, the different
6 The earthquake municipalities were defined as: Appingedam, Bedum, Delfzijl, Eemsmond, Loppersum, Slochteren, Ten
Boer and Winsum
7NAM (personal interview), Province of Groningen (personal interview), Municipality Loppersum and Eemsmond
