The Markov chain model (MCM) has become a popular tool in the agricultural economics literature to explain the past evolution of and simulate the future developments in the number and size distribution of farms. In this paper, I show that the way MCMs have been implemented by agricultural economists so far suffers from the fact that transition probabilities are estimated as almost independent variables (up to adding-up to unity constraints). The alternative parametric MCM I propose addresses the deriving issues since (i) it is parsimonious in terms of parameters; (ii) it can be estimated with simple econometric techniques; (iii) it reveals detailed information on the structural change processes at hand. Applying it to experimentally controlled data with noise shows that the proposed model behaves well and competes with the traditional approach without any significant shortcoming. Two illustrative empirical applications, one using data from the EU-15 Farm Accounting Data Network (FADN) and the other using data from the USA Agricultural Resource Management Survey (ARMS), reveal the rich information that can be derived regarding the economic size changes experienced annually by farms in both regions.