FORECASTING TECHNIQUES
CORRELATION
LEAST SQUARES METHOD OF LINEAR REGRESSION ANALYSIS
THE COMPONENTS OF TIME SERIES
USING INDEX NUMBERS
SALES FORECASTING: THE PRODUCT LIFE CYCLE
CORRELATION
Two variables are said to be correlated if a change in the value of one
variable is accompanied by a
change in the value of another variable. This is what is meant by
correlation.
Examples of variables which might be correlated are as follows.
A person's height and weight
The distance of a journey and the time it takes to make it
SCATTERGRAPHS
DEGREES OF CORRELATION
Two variables might be perfectly correlated, partly correlated or uncorrelated. Correlation
can be
positive or negative
NORMS
NO CORRELATION
THE CORRELATION COEFFICIENT AND THE
COEFFICIENT OF DETERMINATION
The correlation coefficient - The degree of linear correlation between two
variables.
correlation coefficient, r. The nearer r is to +1 or –1, the stronger the
relationship
THE CORRELATION COEFFICIENT AND THE
COEFFICIENT OF DETERMINATION
The correlation coefficient, r, must always fall between –1 and +1. If you
get a value outside this range you have made a mistake.
r = +1 means that the variables are perfectly positively correlated
r = –1 means that the variables are perfectly negatively correlated
r = 0 means that the variables are uncorrelated
THE COEFFICIENT OF DETERMINATION, R2
The coefficient of determination, r2 (alternatively R2) measures the
proportion of the total variation in the value of one variable that can be
explained by variations in the value of the other variable. It denotes the
strength of the linear association between two variables.
CORRELATION AND CAUSATION
If two variables are well correlated, either positively or negatively, this
may be due to pure chance or there may be a reason for it. The larger
the number of pairs of data collected, the less likely it is that the
correlation is due to chance, though that possibility should never be
ignored entirely.
LINES OF BEST FIT
Correlation enables us to determine the strength of any
relationship between two variables but it does not offer us any
method of forecasting values for one variable, Y, given values of
another variable, X.
If we assume that there is a linear relationship between the two
variables, however, and we determine the equation of a straight line (Y
= a + bX) which is a good fit for the available data plotted on a
scattergraph, we can use the equation for forecasting: we can substitute
values for X into the equation and derive values for Y.
ESTIMATING THE EQUATION OF THE LINE OF BEST FIT
There are a number of techniques for estimating the equation of a line of
best fit. We will be looking at simple linear regression analysis. This
provides a technique for estimating values for a and b in the equation
Y = a + bX
where X and Y are the related variables and a and b are estimated using
pairs of data for X and Y.
LEAST SQUARES METHOD OF LINEAR REGRESSION
ANALYSIS
Linear regression analysis (the least squares method) is one
technique for estimating a line of best fit. Once an equation for a line of
best fit has been determined, forecasts can be made.
THE RELIABILITY OF REGRESSION ANALYSIS
FORECASTS
As with all forecasting techniques, the results from regression analysis will
not be wholly reliable. There are a number of factors which affect the
reliability of forecasts made using regression analysis
THE RELIABILITY OF REGRESSION ANALYSIS
FORECASTS
It assumes a linear relationship exists between the two variables.
It assumes that the value of one variable, Y, can be predicted or estimated from
the value of one other variable, X. In reality the value of Y might depend on
several other variables, not just X.
When it is used for forecasting, it assumes that what has happened in the past
will provide a reliable guide to the future.
When calculating a line of best fit, there will be a range of values for X
As with any forecasting process, the amount of data available is very
important.
The reliability of a forecast will depend on the reliability of the data
collected to determine the regression analysis equation.
ADVANTAGES OF REGRESSION ANALYSIS
It gives a definitive line of best fit, taking account of all the data.
Linear regression makes efficient use of data and good results can be
obtained with relatively small data sets.
The significance/reliability of the relationship between variables can be
statistically tested
Many processes are linear so are well described by regression analysis.
Even many non-linear relationships can be well approximated by a linear
model over a short range.
THE HIGH-LOW METHOD
We covered previously
THE COMPONENTS OF TIME SERIES
A time series is a series of figures or values recorded over time.
There are four components of a time series: trend, seasonal variations,
cyclical variations and random variations
FORECASTING PROBLEMS
All forecasts are subject to error, but the likely errors vary from case to case.
The further into the future the forecast is for, the more unreliable it is
likely to be.
The less data available on which to base the forecast, the less reliable
the forecast.
The pattern of trend and seasonal variations cannot be guaranteed to
continue in the future.
There is always the danger of random variations upsetting the pattern of
trend and seasonal variation.
USING INDEX NUMBERS
An index is a measure, over time, of the average changes in the value
(price or quantity) of a group of items relative to the situation at some
period in the past.
USING INDEX NUMBERS
Composite indices cover more than one item.
Weighting is used to reflect the importance of each item in the index.
Weighted aggregate indices are found by applying weights and then
calculating the index.
There are two types of weighted aggregate index;
Laspeyre (which uses quantities/prices from the base period as the weights)
Paasche (which uses quantities/prices from the current period as weights).
Fisher's ideal index is the geometric mean of the Laspeyre and Paasche
indices.
PRICE INDICES AND QUANTITY INDICES
An index may be a price index or a quantity index.
A price index measures the change in the monetary value of a
group of items over time.
A quantity index (also called a volume index) measures the change in
the non-monetary values of a group of items over time.
PRICE INDICES AND QUANTITY INDICES
EXAMPLE: SINGLE-ITEM INDICES
Given the values of some commodity over time (a time series), there are
two ways in which index relatives can be calculated.
In the fixed base method, a base year is selected (index 100), and all
subsequent changes are measured against this base. Such an approach
should only be used if the basic nature of the commodity is unchanged
over time.
In the chain base method, changes are calculated with respect to the
value of the commodity in the period immediately before. This approach
can be used for any set of commodity values but must be used if the basic
nature of the commodity is changing over time.
EXAMPLE: FIXED BASE AND CHAIN BASE METHODS
COMPOSITE INDEX NUMBERS
COMPOSITE INDEX NUMBERS
WEIGHTED AGGREGATE INDICES
This method of weighting involves multiplying each component value by
its corresponding weight and adding these products to form an aggregate.
EXAMPLE: A PRICE INDEX
EXAMPLE: A QUANTITY INDEX
SOLUTION
LASPEYRE, PAASCHE AND FISHER INDICES
Laspeyre and Paasche indices are special cases of weighted aggregate
indices.
LASPEYRE INDICES
Laspeyre indices use weights from the base period and are therefore
sometimes called base weighted indices.
LASPEYRE QUANTITY INDEX
PAASCHE INDICES
Paasche indices use current time period weights. In other words, the
weights are changed every time period.
PAASCHE QUANTITY INDEX
A Paasche quantity index uses prices from the current period as
weights and can be expressed as follows.
WHICH TO USE – PAASCHE OR LASPEYRE?
The following points should be considered when deciding which type of index to use.
A Paasche index requires quantities to be ascertained each year
For the Laspeyre index, the denominator is fixed. The Laspeyre index can therefore be calculated as soon as current
prices/quantities are known.
The Paasche index, on the other hand, cannot be calculated until the end of a period, when information about current quantities/prices
becomes available.
The denominator of a Laspeyre index is fixed and therefore the Laspeyre index numbers for several different years can be directly
compared.
With the Paasche index, on the other hand, comparisons can only be drawn directly between the current year and the base year
(although indirect comparisons can be made).
The weights for a Laspeyre index become out of date, whereas those for the Paasche index are updated each year.
A Laspeyre price index implicitly assumes that, whatever the price changes, the quantities purchased will remain the same. In terms
of economic theory, no substitution of cheaper alternative goods and services is allowed to take place. Even if goods become relatively
more expensive, it assumes that the same quantities are bought. As a result, the index tends to overstate inflation.
The effect of current year weighting when using the Paasche price index means that greater importance is placed on goods that are
relatively cheaper now than they were in the base year.
As a consequence, the Paasche price index tends to understate inflation.
In practice, it is common to use a Laspeyre index and revise the weights every few years.
FISHER'S IDEAL INDEX
taking the geometric mean of the Laspeyre index and the Paasche
index.
Fisher's ideal index = (Laspeyre x Paasche)
THE RETAIL PRICES INDEX (RPI) AND CONSUMER
PRICES INDEX (CPI)
The RPI measures the change in the cost of living. It is published online
monthly (on a Tuesday) near the middle of the month by the Office for
National Statistics. Since it measures the monthly change in the cost of living,
its principal use is as a measure of inflation.
In recent years, the Government has chosen to use the Consumer Prices
Index (CPI) as the main
measure of UK inflation for macroeconomic purposes. Thus the CPI is the
basis for the Government's
inflation target: it is also used to make international comparisons.