The Research On Free Cash Flow Forecasting Model

Discount Cash Flow (DCF) methods are a common and relative reliable method in enterprise valuation. One of them, Discount Free Cash Flow (DFCF) method, is popular to be applied in valuation practice. There are two key factors for DFCF method: one is the forecasting of future free cash flows, the other is the estimation on Weighted Average Cost of Capital (WACC). The latter is usually derived from Capital Assets Pricing Model (CAPM) and this dissertation researches on the forecasting of future free cash flows.Reference to prior results about cash flow forecasting, this dissertation builds two free cash flow forecasting models from different two angles. On one hand, through decomposing FCF to components and according to these components'time series properties, a free cash flow forecasting model will be build up. This model will prove that the information contained in components exceeds that in FCF as total. On the other hand, financial ratios are the indictors which reflect enterprise's management and market environment and then decide its capability in generating free cash flow. Hence, another free cash flow forecasting model can be build based on financial ratios. This model also will verify that financial ratios have incremental information in relate to the total FCF.This dissertation selects empirical data as General Motors'quarterly financial data from 1993 to 2003 and uses Auto Regression Integrated Move Average (ARIMA) method to test the two models. After parameter estimations, both models have acceptable fitness for inner samples. Using the equations of the two models, the FCF prediction of 2004 quarters will be done and compared with fact figures. The test results show that: (1) both the two models have more predictive ability for FCF than the prior FCF as a total; (2) financial ratios have a bit less predictive power than components of FCF; (3) the lag one, lag two and lag three financial data and ratios are valuable to forecast next period free cash flow.As all papers, this dissertation also has limitations to be further researched on. There are three listed: large sample research based on same industry data or cross industries data is necessary; under the same theory analysis, it is important to use different time series analysis method such as GARCH to build different free cash flow forecasting models; it is worthy to build models by decomposing FCF in different ways or using different financial ratios in order to investigate respective influence on predictive ability.