Financial Modelling and Forecasting
Lecture 1 Introduction and Descriptive Statistics
The need for forecasts
A forecast helps deal with an uncertain future by making decisions today No single forecasting method will lead to an accurate forecast. Forecasts can be wrong! “What’s the point of forecasting?”
A business requires predictions as inputs
E.g., Inventory, Personnel, Ordering, Production planning.
Governments require forecasts to guide monetary and fiscal policy
Lecture 1 2
Application to Finance
A sensible forecast allows proactive decisions to be made today
Without it, management ...view middle of the document...
0, 7.5, 6.0, 4.5, 7.0, 9.5, 8.0, 6.5, 9.0
Qualitative forecasts are suitable when no relevant historical data exists.
Individual forecasts of salespeople Combines the anonymous forecasts of experts, and then redistributes the forecasts for revision until a consensus is reached.
Used for deciding strategy, new product development and long-range plans
These techniques incorporate judgment/intuition and subjective factors Common techniques include: Survey methods
Seeking opinions from potential users/customers
Lecture 1 7
These methods suffer from bias and require many years of experience to gain sufficient understanding of a particular market. Intelligent forecasting should incorporate both qualitative and quantitative methods
Explanatory and Time-Series Models
Forecasting Methods Summary
Forecasting Method Qualitative Techniques Causal Techniques Data Requirement Useful when historical data are scarce or non-existent Useful when historical data are available for both the dependent (forecast) and the independent variable(s) Useful when historical data exists for forecast variable and the data exhibits a pattern Technique Examples Delphi Technique Scenario Writing Regression Models Leading Indicators
Explanatory, or causal, models assume the predicted variable is related to one or more independent variables
Some variables are controllable, some are not
Time-series models assume that the forecasted variable is a function of historical values of the same variable
Moving Average Exponential Smoothing Autoregressive Models
Relies on projecting past relationships into the future Requires less data
Lecture 1 9 Lecture 1 10
Application to company valuation
A fundamental analysis using the following data: Quantitative Data
GDP $ Advertising Interest rates Inventory levels SWOT analysis Management Quality Industry Outlook
Lecture 1 11
Building a model helps us forecast how much a variable might change However, events outside the firm’s control will lead to inaccurate forecasts regardless of the sophistication of the method Model misspecification, missing or inaccurate data, unexpected changes in variables and error build-ups are just some issues forecasters face. We can only hope to model ‘unknown knowns.’ Forecasting is difficult
Measures of Central Tendency
A histogram is one type of graphical display to get an initial overview of a set of data However, it is useful to summarise a large data set by using a few key numerical measures: Measures of Central Tendency
The mean, or average, is the most commonly used measure of central...