Selection of Market Demand Forecast Methods: Criteria and Application

Abstract

The article deals with the theoretical aspects of themarket demand method selection criteria and theirapplication in practice. As the definition of market demandand conditions is related to the achievement of importantgeneral strategic decisions and consumers’ behaviour onthe competitive market, the accurate identification of themarket demand method selection criteria and theirapplication can reduce decision indefiniteness. Hence, themain aim of the article is to analyse market demandforecast method selection criteria, and apply them indeveloping Lithuanian furniture demand forecast.The theoretical part of the article starts with theanalysis of the forecast method reasoning and forecastmethod classification under various economic literaturecriteria. The analysis of the main forecast method groupsdisclosed that it was rather difficult to determine theadvantage of any method in forecast estimation, whereas,always the risk of the wrong method selection remained. Itwas determined that usually unanimous authors’ opinionon the complex application of quantitative and qualitativeforecast methods differed in two aspects: one group ofresearchers supported the priority of quantitative forecastmethods by using qualitative forecast methods as anauxiliary or alternative means, while the other group ofauthors stated that equal application of both types offorecast methods guaranteed higher reliability and carriedmore information.Quite often any market demand forecast encountersthe relevant forecast method selection problem. It shouldbe noted that there are advantages and disadvantages ofthe forecast method selection in every case; therefore, theanalysis and differentiation of the main forecast methodselection criteria is expedient. The selection of the forecastmethod should be based on several criteria taking intoaccount the applicability of the forecast methodcomplexity, i.e., forecast accuracy level, period of time, thescope of initial data, forecast costs, and the level of resultappropriateness and applicability.The article deals with the research related to theselection of the forecast method for the Lithuanianfurniture industry to define the furniture demand forecastfor 2007. The object of research is furniture sales on thenational market that reflect the solvent part of the furnituredemand. The comparison of the forecast accuracyassessment indicators that were estimated by usingdifferent forecast methods indicate that the lowest furnituresales forecast error values were achieved by applyingexponential smoothing method where &=0.3: forecasterror value (et = 77.19) demonstrated the smallestdifference between furniture sales forecast and fact, meanabsolute percentage error (MAPE = 8.29 proc.) indicatedhigh forecast accuracy, and mean percentage error (MPE= 8.29 proc.) described small positive forecast deviation.In order to achieve the qualitative forecast assessment,forecast reliability intervals were calculated where criterion z= 1.960. The results allow to assume that furniture salesforecast for 2008 with 95 per cent probability can vary fromLTL, 844.64 mio. to LTL, 1147.20 mio.