Page 131 - DCOM504_SECURITY_ANALYSIS_AND_PORTFOLIO_MANAGEMENT

Security Analysis and Portfolio Management

Notes Where, a equals the number of products purchased when employment is zero and b equals
the amount of change in the number of products purchased with every change in total
employment.
The latter method is more accurate because it is more sensitive to the influence of
independent variable on dependent variable.
Multiple regression analysis facilitates the study of impact of more than one independent
variable on the dependent variable.

Y = a + b x + c x + d x + e x + f x
1 2 3 4 5
Where, Y = Yearly sales in lakhs of rupees;

x = yearly sales (lagged one year) in lakhs of rupees
1
x = yearly advertising expenditure in lakhs of rupees
2
x = a dummy variable
3
x = year
4
x = disposable personal income in lakhs of current rupees
5
(c) Time series analysis: Time series analysis consists of decomposing the original sales
series over a period of time. The elements derived are:
Trend (T): It is the result of basic developments in population, capital formation, and
technology. It is found by fitting a straight or curved line through past sales.

Cycle (C): It captures the wave-like movement of sales. Many sales are affected by
swings in general economic activity, which tends to be somewhat periodic. The
cyclical component can be useful in intermediate range forecasting.
Season (S): It refers to a consistent pattern of sales movements within the year. The
term season describes any recurrent sales pattern. The seasonal component may be
related to weather factors, holidays, and trade customs. The seasonal pattern provides
a norm for forecasting short-range sales.
Erratic Events (E): It refers to the unpredictable sales caused by unforeseen events
like strikes, riots, war scares, floods, and other disturbances.
Another time series technique is exponential smoothing. For industries with several
items in product line, this technique is useful to produce efficient and economical
short-run forecasts. It requires only three pieces of information.
(i) This period's actual sales (Q )
t
(ii) This period's smoothed sales (Q )
t
(iii) A smoothing parameter (a), where

Sales forecast for next period (Q + 1) = Q + (1 – a)Q
t t t
The initial level of smoothed sales can simply be the average sales for the last few
periods. The smoothing constant is derived by trial and error testing of different
smoothing constants between zero and one, to find the constant that produces the
best fit of past sales.

126 LOVELY PROFESSIONAL UNIVERSITY