Epsilon Archive for Student Projects

Abstract

The price of farmland has risen sharply in the last 20 years. This has led to discussions about what creates the value of farmland. The aim in this study was to try and find what factors are affecting the price on farmland. The area that is studied is the counties around Mälardalen during the period 2005 to 2009.

The method chosen to achieve the objective is the hedonic price model. The variables are chosen after a preliminary study where a literature search has been made and a number of interviews with people in the industry. The interviews revealed the existence of three different purposes for the purchase of agricultural properties, production, interest or accommodation. Depending on the buyers objective different variables will have different effects on the price. Therefore, three models were produced, one for each purpose.

To obtain information about the purchases that had occurred in the region, during the selected time period, data was collected from NaiSvefas program RealEstate and Lantmäteriets Fastighetssök. Based on the data collected the buyers where contacted and short telephone interviews were conducted. This was done to find out the purpose for which the purchase had been made, and also to collect information for some of the potential explanatory variables. The data for the groups consisted of 54 observations in the production group, 35 of them in the interest group and 27 observations in the accommodation group.

The process of generating the hedonic pricing model was as follows. First, a regression analysis with all the potential explanatory variables was made. When analyzing the results a problem with multicollinearity was discovered. To get rid of this problem the non-significant variables and those that had a high correlation between them where gradually removed. This was done manually to ensure that important variables for the study were not removed.

It can be established that regardless of what purpose the purchase has been made it will require different types of variables to get a high degree of explanation. It requires production variables as well as urbanization and economic variables. The variables that are interesting to study also differ between the groups.

The coefficient of determination for the final model in the production group was 55.8% with seven significant variables of a total of nine. The significant variables were average price of grain, the normal harvest in the area, interest rate, the average income in the municipality, the population density in the municipality, if the purchaser exercises the option to hunt and the distance to the nearest road.

The coefficient of determination for the final model of the interest group was 70.2% with five significant variables of a total of nine. The significant variables were the time variable, normal harvest for the area, population density in the municipality, number of sales in the municipality as well as travel time to Stockholm.

The coefficient of determination for the final model in the accommodation group was 72,6%, with seven significant variables out of eight. The significant variables were, interest rate, the time variable, population density in the municipality, whether it was sold by a broker, travel time to nearest town, if the purchaser exercises the option to hunt and if the land is leased.