Increase in the Value of Agricultural Parcels—Modelling and Simulation of the Effects of Land Consolidation Project

Round 1

Reviewer 1 Report

The topic of the article is interesting and topical.

The purpose of the research was clearly specified in the Introduction.

The literature review was correctly performed, based on well-selected and up-to-date literature, which helped the Authors to develop a calculation model for determining the value of agricultural land according to their characteristics (size, access to public roads). 

I would suggest making some minor changes which are listed below.

1. The title is very long, I suggest you optimize the title according to the content of the manuscript.
2. Attention! Consistency in the use of the terms!

MRA!

Line 28 multiple regression analysis model (MRA)

Line 37 multiple linear regression

Line 121, 260

 

ANN !

Line 29 – artificial neural network model (ANN)

Line 38 – neural networks

Line 121, 261

 

3. Line 45 – please indicate the source of the data
4. Line 77 – consolidate the statements by referring to the specialized literature
5. Line 96 – Fig. 1??? Figure xxxx
6. Table 1 zł/m2 ???? Throughout the manuscript the value of the land is entered in PLN (Line 332 – 338)
7. Table 1 please indicate what it means Wb
8. Line 167 and 169 please indicate what it means MEA
9. Line 203 and 204 Tab. 2????? Table 2 is at line 273!!!
10. Figure 3 - Please improve readability.
11. Line 248 - please indicate what it means BFGS
12. Line 283 – Tab.1? Table xxx
13. Line 294- Tab. 3? Table xxx
14. Line 315 Tab. 3??????????????
15. Line 323 Table 3?????????
16. Line 343 Tab. 3??????
17. Line 370 - I think this statement is somewhat exaggerated. “Many other scholars” is written, but you cite only one.
18. Line 391 – Table 3?

19. I suggest you to present the limitations of your study.

Author Response

Response to Reviewer 1

REVIEWER 1

1. The title is very long, I suggest you optimize the title according to the content of the manuscript.

The title were changed to:

Increase in the value of agricultural parcels – modelling and simulation of the effects of a land consolidation project

2. Attention! Consistency in the use of the terms!

The terms were changed to be consistent:

• lines 37, 121 and 260 - multiple regression analysis model (MRA),
• lines 38, 121 and 261 - artificial neural network model (ANN).

 

3. Line 45 – The source of data (Eurostat) was indicated - https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Farms_and_farmland_in_the_European_Union_-_statistics
4. Line 77 - The statements were consolidated by referring to the literature – [2, 4, 14, 16]
5. Line 96 – Fig. 1 was replaced by Figure
6. Table 1 – The value of the land was changed for PLN (instead of zł).
7. Table 1 – W_b is the soil classification rate. It is calculated as the quotient of the area which is the basis for calculating the agricultural tax (conversion hectare) and the physical area (physical hectare).
8. Line 167 i 169 – MEA was corrected on MRA.
9. Line 203 and 204 - The reference to Table 2 was removed
10. Figure 3 – The title was corrected on “Generalized form of the artificial neural network model - multilayer perceptron”
11. Line 248 – BFGS means numerical optimization algorithm (Broyden–Fletcher–Goldfarb–Shanno)
12. – 16. All the abbreviations of Tab. were corrected on Table (lines: 283, 294, 315, 323, 343). In line 323 and 343 the table numbers were changed to the next one (4).
13. Line 370 – The statement was corrected. The word „many” was removed.
14. Line 391 – The table number was changed to 4.
15. Please see that the limitations of our study was underlined in the text: “Identifying universal rules seems difficult due to wide regional variations of the factors impacting the prices of agricultural properties and their changeability over a longer period of time” (line 411-413).

Reviewer 2 Report

The presented article analyzes the key parameters, determining the price of agricultural land and display the possible effects of these parameters on changes in the values of land parcels. This approach may be used in land consolidation projects due to improving the effect of plot adjustment. Poland is a country with a large area of agricultural land (see l. 43,44 - please insert the source of the presented numbers (Eurostat?)), so the questions concerning this field are very important. The analyses were performed through multiple regression analysis and an artificial neural network. Results obtained correspond with common findings, that the price of the land plots depends on their accessibility, size, elongation and locality. 

comments:

1)in the Tab.1 predictor name  Obstacles to usage: for state 1 you define the occurrence of more than two of the named aspects; for state 2 occurrences of one aspect. What about occurring just only two aspects?

2) I do not understand clearly the tab. 3 and the comments below it. It seems, that the description does not correspond with the content of the table? What are the current prices of parcels and what are the predicted values?

 Land consolidation is a  very complex process, focused on enabling rational land management with optimal space and functional organization of land plots. The price of land will always depend not only upon the objective factors but will be very variable due to time, local interests, rural and industrial development. Land consolidation should not concentrate only on the value of adjusted plots.

Author Response

Response to Reviewer 2

 

REVIEWER 2

In lines 43 and 44 the source of the presented numbers was inserted.

Comments

1. In case of occurring just only two obstacles to usage, a discrete quantitative feature assume state 1 – large. To make it clearer definition of state 1 was clarified in the text (1 – large (two or more of the following aspects appear at the same time …)).
2. The description of data was corrected in the text below the Table 4.

Round 2

Reviewer 1 Report

The work has improved substantially.

Author Response

Manuscript agriculture-1155058

Response to Reviewer

 

Dear Reviewer,

We greatly appreciate your thoughtful comments that helped improve the manuscript. We trust that all of your comments have been addressed accordingly in a revised manuscript. Thank you very much for your effort.

In the following, we give a reply to your comments. 

Multiple regression analysis guarantees simplicity of results. As we mentioned in the methodology section, in the case of both the models (MRA and ANN), we were guided by the postulate of modelling simplicity.

Multiple regression analysis model (MRA) is conceptually simpler and easier to interpret than non-linear regression models. Numerous references to the practical application of such a model in real estate market analyses can be found in Polish and international literature (Bruce and Sundell 1977, Eckert 1990, Isakson 1998, Cellmer 1999, Dacko 2000, Czaja 2001, Hozer 2001, Dacko and Lendzion 2003, Śnieg 2003, Benjamin et al. 2004, Sirmans et. al. 2004, Adamczewski 2006, Surowiec 2006, Bitner 2007, Barańska 2010, Sawiłow 2010). In recent decades, such models were used to determine real estate value and they provided satisfactory results. Prus (2010) used linear stepwise regression in real estate price modelling and, interestingly, she also analysed alternative non-linear models (exponential and logarithmic) and did not find them to be better. According to this author, the linear model proved to be the most optimal due to the ease of calculations and way of interpretation. There are known MRA models which made it possible –  thanks to stepwise procedure – to find out which factors, and to what extent, impact agricultural properties (Śnieg 2003, Walkowiak i Zydroń 2012) (it was mentioned in line 144 of the text) despite the fact that the prices of such properties are determined in different regions by different factors and to different extent. Therefore, although there is a significant body of theoretical literature and it is generally known what is important in agricultural land value formation, our model took the form of multiple regression analysis and was constructed using stepwise procedure. The existing knowledge - however rich - has only general character with respect to real estate market modelling: it allows to define merely a set of potentially useful predictors. We do not know what and how impacts prices in specific cases, because the relationships between prices and value influencing factors are more like regularities than laws. Due to the local character of such regularities, we are unable to indicate a priori which of the potentially useful predictors are significant, and which ones can be ignored.

Perhaps, we have not indicated clearly enough the number of variables from which the stepwise regression procedure started. It was oversight on our part arising from our attempt to synthesise and shorten the text. In fact, 10 predictors were examined in detail (added to the text - line 137). They were presented in Table 1. In the stepwise procedure, the following factors were eliminated from the MRA model:  surface area of a parcel, tax region and soil quality. Despite numerous attempts and deletion of outlier observations, these factors remained insignificant for MRA. This explanation has been added to the text (line 171). In addition, a fragment (from line 155 to line 168) which does not explain the method (it describes technical details of the study) and could render the text unintelligible was removed.  

We would like to emphasise that the MRA model, based on 7 selected predictors, passed a substantive verification: regression coefficient signs indicated relationships consistent with the theory (Table 2) (it was mentioned in the text of the paper - line 283). Statistical verification was also positive, as Fisher-Snedecor statistic and Student's t-test statistics satisfied the requirements. Coefficient of determination R² = 0.68 would perhaps be too low for laboratory studies in such fields as agriculture, chemistry or medicine. But with respect to our attempt to explain the process of agricultural land price formation (given market imperfections and behavioural factors), is is, in our opinion, quite a good result. In the Polish conditions, similar, and sometimes even lower, values of this coefficient were obtained in studies of local real estate markets by other authors (Śnieg 2003, Surowiec 2005, Barańska 2010, Walkowiak and Zydroń 2012, among others). It is also worth noting that the objective of our studies was not predictive but rather exploratory in character. It was more about estimation, approximate evaluation of the impact of consolidation projects on changes in agricultural land value than accurate valuation of a single parcel. 

Naturally, multiple regression analysis model had a drawback - it did not show how the value of agricultural land would change with changes in its surface area. This was because this relationship was non-monotonic and non-linear. However, our objective was an attempt to answer the question of how the value of agricultural parcels is impacted by correction of the parcels' shape and surface area and by access to road. Hence the second model – ANN – which takes into account non-monotonic and non-linear relationships. The model was sufficient to analyse the effects of changes to all the three factors corrected as part of consolidation projects and to show this in simulations in Table  4.

Perhaps, our text lacked clear narration and it was difficult to identify our intentions, but using two models at the same time (and often even more) is a typical approach in Data Mining, where out of many models few best are selected, because the average of the results obtained from them becomes a more reliable prediction. We have not highlighted that clearly enough in this paper. However, in our opinion, the two models complement each other, and the analysis would not be complete if either of them was removed. The first model provides us with a simple formula and shows additive impact of the different factors on land value. The second model allows us to take into account the surface area of parcels corrected as part of consolidation projects.

Additional references:

Adamczewski Z., 2006, Elementy modelowania matematycznego w wycenie nieruchomości. Podejście porównawcze, Oficyna Wydawnicza Politechniki Warszawskiej

Barańska A., 2010, Statystyczne metody analizy i weryfikacji proponowanych algorytmów wyceny nieruchomości, Rozprawy i Monografie, Wydawnictwa AGH, Kraków

Benjamin J. D., Randall S. Guttery R. S., Sirmans C. F., 2004,  Mass Appraisal: An Introduction to Multiple Regression Analysis for Real Estate Valuation, Journal of Real Estate Practice and Education, vol. 7, no 1, pp. 65-77

Bitner A., 2007, Konstrukcja modelu regresji wielorakiej przy wycenie nieruchomości, Acta Scientiarum Polonorum, Administratio Locorum, No 6(4), 59-66

Cellmer R., 1999, Zasady i metody analizy elementów składowych rynku nieruchomości, Wydawnictwo ART, Olsztyn

Czaja J., 2001, Metody szacowania wartości rynkowej i katastralnej, Komp-System, Kraków

Dacko M., 2000, Zastosowanie regresji wielokrotnej w szacowaniu nieruchomości w arkuszu kalkulacyjnym Microsoft Excel 2000, Wycena: wartość – obrót – zarządzanie nieruchomościami, Nr 2/2000, Wyd. Educaterra, Olsztyn

Dacko M., Lendzion M., 2003, Modelowanie cen nieruchomości za pomocą regresji wielorakiej i sztucznych sieci neuronowych, Wycena, nr 3 (62)/2003, Wyd. Educaterra, Olsztyn

Hozer J., 2001, Regresja wieloraka a wycena nieruchomości, Rzeczoznawca Majątkowy, nr 2

Isakson, H. R. 1997. “An Empirical Analysis of the Determinants of the Value of Vacant Land.” Journal of Real Estate Research13(2): 103-14

Sawiłow E., 2010, Problematyka określania wartości nieruchomości metodą analizy statystycznej rynku, Studia i Materiały Towarzystwa Naukowego Nieruchomości, Vol 8, No 1, s. 21-32

Prus B., 2010, Assessment of information from real estate markets with help of multiple regression analysis models, INFRASTRUCTURE AND ECOLOGY OF RURAL AREAS, PAN Oddział w Krakowie, Nr 3/10, s. 103-113

Sirmans G. S., Macpherson D. A., Zietz E. N., 2005, The Composition of Hedonic Pricing Models, Journal of Real Estate Literature, vol. 13, no 1, pp. 3-46

Śnieg R., 2003, Czynniki i cechy kształtujące sprzedaż i dzierżawę nieruchomości rolnych Skarbu Państwa. Rozprawa doktorska. UWM. Olsztyn

Surowiec G., 2006, Aspekty geodezyjno-przestrzenne w badaniach rynku nieruchomości. Rozprawa doktorska. Politechnika Warszawska. Warszawa

Walkowiak R., Zydroń A., 2012, Zastosowanie regresji krokowej do określenia atrybutów wpływających na wartość nieruchomości rolnych na przykładzie gminy Mosina, Acta Scientiarum Polonorum. Administratio Locorum 11/3, 239-253

Bruce R.W., Sundell D.J., 1977, Multiple regression analysis: history and applications in the appraisal profession. Real Estate Appraiser Jan/Feb, 37-44

Eckert J.K. (ed.), 1990, Property appraisal and assessment administration. International Association of Assessing Officers, Chicago