Abstract
This paper investigates the applicability of the hedonic approach for creating indices of residential real estate values from state land registration and valuation systems in Ghana, Africa. Using data on 2,950 transactions from five classified locations in Accra and Tema for 1992–2005, sale prices are modeled, first, for the aggregate sample, before segmenting the market on the basis of two sub-periods (1992–1998 and 1999–2005) and three price bands. A series of continuous and categorical dummy variables are included in the model to construct a residential real estate index. The main conclusion is that heterogeneous real estate characteristics such as the number of bedrooms, number of stories, real estate size and type, quality of landscaping, plot size, security of tenure, and location are all significant variables that influence real estate prices in Ghana.