14

Time Series Analysis

14.1 Introduction
One of the most important tasks facing economists and business people these days is to
make estimates for the future. The first step in making estimates for the future consists of
gathering information from the past. In this connection, one usually deals with statistical
data that is collected, observed or recorded at successive intervals of time. Such data is
referred to as time series.
The term time series can be applied to all phenomena that are related to time, such as the
number of accidents occurring in a day, the variation in the body temperature of a patient
during a certain period, the number of marriages taking place during a certain period, and
so on. For example,

1. A businessman is interested in finding out his likely sales in the year 2017 so that
he can adjust his production accordingly and avoid the possibility of either unsold
stocks or inadequate production to meet the demand.
2. An economist is interested in estimating the likely population in the coming year
so that proper planning can be carried out with regard to food supply, jobs for the
people and so on.

14.2 Time Series

14.2.1 Definition
When we observe numerical data at different points of time, the set of observations is
known as a time series. If we observe production, population sales, or imports and exports
at different points of time, say over the last 5–10 years, the set of observations formed will
constitute a time series. In the analysis of time series, time is the most important factor,
because the variable is related to time, which may be years, months, weeks, days, hours, or
even minutes or seconds.
Times series can apply to phenomena related to time, such as

1. The number of accidents occurring in 24 hours

2. The variation in the temperature of a patient during a certain period
3. Study of profits over 15 years to estimate for the coming years
4. Estimation of food supply for the population in the next 10 years

337
338 Quantitative Techniques in Business, Management and Finance

14.2.2 Features of Time Series

1. Variables are determined with respect to time.

2. Time series consist of a homogeneous set of variables.
3. Time series deal with data over a long period.

For example, to study agricultural production, one may take 10–15 year data, while to
study the pattern of rainfall, daily recording is essential.

14.2.3 Uses of Analysis of Time Series

Analysis of time series is of great significance not only to the economist and business per-
son but also to the scientist, astronomer, geologist, sociologist, biologist, research worker
and so on, for the following reasons:

1. Economic problems
Changes in demand for a product over a period of time can be analysed using
time series.
2. Study of past behaviour
It helps in understanding past behaviour. By observing data over a period of
time, one can easily understand changes that have taken place in the past. Such
analysis will be helpful in predicting future behaviour.
3. Evaluation of current accomplishments
The actual performance can be compared with the expected performance and
the cause of variation analysed.
For example, if expected sales for 2016 were 12,000 refrigerators and the
actual sales were only 10,000, one can investigate the cause for the shortfall in
achievement.
Time Series analysis will enable us to apply the scientific procedure of `holding
other things constant’ as we examine one variable at a time. For example, if we
know how much the effect of seasonality on business is, we may devise ways and
means of ironing out the seasonal influence or decreasing it by producing com-
modities with complementary seasons.
Economic and business activities depend on time series to review and find
deviations in progress by studying the present situation.
4. Comparison
Time series provide data in chronological order and present the information in
a systematic way. This facilitates comparison between past and present.
5. Forecasting
It helps in planning future operations. If the regularity of occurrence of any
feature over a sufficient long period can be clearly established, then, within
limits, prediction of probable future variations will become possible. It helps
in making long-range forecasts and provides a base for long-term operational
planning.
Time Series Analysis 339

14.3 Components of Time Series

The forces affecting the values of phenomena in a time series are called components of the
time series. For example, demand for product is influenced by substitutes, seasonal supply,
competitors’ prices, purchasing power, influence of business cycles and so on.
The effects of the various forces are:

1. Changes occurring as a result of the tendency of the data to increase or decrease

are known as secular movements.
2. Changes taking place during a period of 12 months as a result of change in cli-
mate, weather conditions, festivals and so on are called seasonal variations.
3. Changes taking place as a result of booms and depressions are classified under the
heading cyclical variations.
4. Changes taking place as a result of forces that could not be predicted, such as
floods, earthquakes or famines, are classified under the heading irregular or erratic
variations.

These four types of patterns, movements or, as they are often called, components or ele-
ments of a time series are

1. Secular trend
2. Seasonal variation
3. Cyclical variation
4. Irregular variation

14.3.1 Secular Trend or Long-Term Trend

The term trend is very commonly used in day-to-day conversation. For example, we often
talk of a rising trend of population, prices and so on. Trend is the basic tendency of produc-
tion, sales, income, employment and so on to grow or decline over a period of time.
The presence of more people means that more food, clothing and housing are necessary.
Technological changes, discovery and exhaustion of natural resources, mass production
methods, improvements in business organisation, and government intervention in the
economy are the major causes for the growth or decline of economic time series.
In some cases, growth in one series involves decline in another. Examples include

1. Silk is displaced by rayon, terylene, etc.

2. Bullock carts, camel carts and tonga are displaced by other modes of transport
such as trucks, auto-rickshaws, taxis and tempo.
3. Better medical facilities, improved sanitation, diet and so on reduce the death rate
and contribute to a rise in birth rate.

Trends are divided in two headings:

1. Linear or straight-line trends

2. Non-linear trends
340 Quantitative Techniques in Business, Management and Finance

14.3.1.1 Meaning of Long Term

A particular period can be regarded as long or short in the study of secular trends depend-
ing on the nature of the data.
For example, if we are studying the figures of sales of a firm for 2012 and 2013, and we
find that in 2013 the sales have gone up, this increase cannot be called a secular trend,
because this is too short a period of time to conclude that the sales are showing an increas-
ing tendency.
On the other hand, if we put a strong germicide into a bacterial culture and count the
numbers of organisms still alive after each 3 seconds for 3 minutes, these 60 observations
showing a general pattern would be called a secular movement.
It is clear from this example that in one case 2 years could not be regarded as a long
period, whereas in another case even 3 minutes would constitute a long period. Hence,
the nature of the data would dictate whether a particular period would be called long
or not.
As a minimum safeguard, it may be said that to calculate trend, the period must cover
at least two or three complete cycles. It is not necessary that the rise or fall must continue
in the same direction throughout the period. We have to observe the general tendency of
the data. As long as we can say that the period as a whole was characterised by an upward
movement or by a downward movement, we say that a secular trend was present.
For example, if we observe the trend of prices over a period of 10 years and find that,
except for a year or two, the prices have been continuously rising, we would call this a
secular rise in prices.

14.3.1.2 Measurement of Secular Trends

1. In studying trend in and of itself, we ascertain the growth factor. For example, we
can compare the growth in one firm of the textile industry with the growth in the
industry as a whole.
2. The growth factor helps us in predicting the future behaviour of the data. If a
trend can be determined, the rate of change can be ascertained and tentative esti-
mates concerning future can be made.
3. The elimination of trend leaves us with seasonal, cyclical and irregular factors.

14.3.1.3 Features of Secular Trends

1. Long period
In certain cases, the period extends to centuries and decades; for example, the
study of rocks, minerals, the earth’s crust and changes in temperature.
For some phenomena, it takes 10–20 years; for example, production of steel by a
steel industry, sales of cotton textiles.
For some phenomena, hours, minutes and seconds constitutes a long period;
for example, growth of bacteria or viruses, heart beat records, pulse rate
records.
2. Tendency of trend curve or movement
It may be smooth or with ups and downs, but shows an average tendency of
growth (Figures 14.1 and 14.2).
Time Series Analysis 341

Secular trend

Linear Non-linear

Gives a straight line. If the values

are plotted on graph paper, it
shows a straight-line tendency

Arithmetic straight line Geometric straight line

The increase or decrease The increase or decrease is

will be constant. uniform.

FIGURE 14.1
Tree diagram secular trend.

Y-axis

Secular trend

Actual time series

X-axis
0 Time (years)

FIGURE 14.2
Secular trend.

14.3.1.4 Uses of Secular Trends

1. The study of the data over a long period of time gives a general idea about the
behaviour of the data.
2. Measures of trend describe the features of the data in the past and also the pattern
of growth or decline in trend.
3. By using secular trends, one can forecast future behaviour on the basis of the past
data.

14.4 Seasonal Variations

Seasonal variations are those periodic movements in business activity that occur regularly
every year and have their origin in the nature of the year itself. Since these variations
repeat during a period of 12 months, they can be predicted fairly accurately. Seasonal
342 Quantitative Techniques in Business, Management and Finance

variation refers to rhythmic changes that occur in periodic movements. It occurs due to
natural factors, seasons or man-made traditions such as festivals, social celebrations, mar-
riages and so on. For example,

1. During summer in many cities, most textile showrooms sell a wide variety of
cotton textiles, and the same shops sell synthetic and woollen products during
winter.
2. During cropping seasons, vegetables or fruits are sold at a cheaper price, but dur-
ing other seasons they are expensive.

Seasonal variation is evident when the data are recorded at weekly, monthly or quarterly
intervals. Although the amplitude of seasonal variations may vary, their period is fixed at
1 year. As a result, seasonal variations do not appear in series of annual figures.

14.4.1 Factors that Cause Seasonal Variations

1. Climate and weather conditions

The most important factor causing seasonal variation is the climate. Changes
in the climate and weather conditions, such as rainfall, humidity and heat, affect
different products and industries differently. For example,
a. During winter, there is greater demand for woollen clothes, hot drinks and so
on, whereas in summer, cotton clothes, cold drinks and ice cream have greater
sales.
b. Agriculture is influenced very much by the climate. The effect of the climate is
that there are generally two seasons in agriculture – the growing season and
the harvesting season – which directly affect the income of the farmer, which
in turn, affects the entire business activity.
2. Customs, traditions and habits
Though nature is primarily responsible for seasonal variations in time series,
customs, traditions and habits also have their impact. For example,
a. On certain occasions, such as Deepawali, Pongal, Dussehra and Christmas,
there is a big demand for sweets, and also there is a high demand for cash
before the festivals, because people want money for shopping and gifts.
b. On the first of every month, there are heavy withdrawals, and the bankers
have to keep plenty of cash to meet the possible demands on the basis of last
month’s experience.
c. Most students buy books in the first few months of the opening of schools
and colleges, and thus the sale of books, stationery and so on shows seasonal
swings.

14.4.2 Application of Seasonal Variation

1. It analyses seasonal patterns of the variables in a short period and studies the
variations in the behaviour of the data.
2. It helps in short-term forecasting and planning.
3. It helps in appraisal of business activities.
Time Series Analysis 343

Y-axis

Seasonal variations

X-axis
0
Time (years)

FIGURE 14.3
Seasonal variations.

4. Seasonal indexes are helpful in scheduling purchases, inventory control, person-

nel requirements, seasonal financing, and selling and advertising programmes.

For example, a housewife may buy fruits for canning or preserving at the peak of the
season, when the prices are low and quality high. Seasonal fluctuations may also be ironed
out in order that the intra-year fluctuations may be less pronounced. Thus, attempts were
made in the United States to build up winter demand for ice cream by advertising: ‘Ice-
Cream is one of your best foods. Eat one plate a day’.

14.4.3 Features of Seasonal Variations

Seasonal variation in a time series is the repositive, recurrent pattern of change which
occurs with a year or shorter time period. The term `seasonal’ is meant to include any kind
of variation which is periodic in nature and whose repeating cycles are of relatively short
duration (Figure 14.3).

14.5 Cyclical Variations

Cyclical fluctuations or business cycle movements are recurrent up and down movements
around secular trend levels which have a duration anywhere from 2 to 15 years.
Most of the time series relating to economics and business show some kind of cycli-
cal variations. Cyclical fluctuations are long-term movements that represent consistently
recurring rises and declines in activity.

14.5.1 Business Cycle

A business cycle consists of the recurrence of the up and down movements of business
activity from some sort of statistical trend or ‘normal’. By ‘normal’ we mean some kind of
statistical average; we do not mean that there is anything very permanent or special.

14.5.1.1 Periods or Phases in the Business Cycle

1. Prosperity
2. Decline
344 Quantitative Techniques in Business, Management and Finance

Y-axis

Improvement
Prosperity

Decline

Normal
Depression
X'-axis
X-axis
0
Y'-axis

FIGURE 14.4
Phases of business cycle.

3. Depression
4. Improvement

Each phase changes gradually into the phase that follows it in the order given. Figure 14.4
illustrates a business cycle.
1. Prosperity phase
In this phase, the public is optimistic – business is booming, prices are high
and profits are easily made. There is a considerable expansion of business activity,
which leads to an overdevelopment. It is then difficult to secure deliveries, and
there is a shortage of transportation facilities, which has a tendency to cause large
inventories to be accumulated during the time of highest prices.
2. Decline phase
Wages increase and labour efficiency decreases. The strong demand for money
causes interest rates to rise to a high level, while doubt enters the banker’s mind
as to the advisability of granting further loans. This situation causes businesses
to make price discounts in order to secure the necessary cash. It then follows the
expectation of further reductions, and the situation becomes critical instead of bet-
ter. Buyers wait for lower prices, and this leads to a decline in business.
3. Depression phase
Then follows a period of pessimism in trade and industry; factories close, busi-
nesses fail and there is widespread unemployment, while wages and prices are
low. These conditions characterise the period of depression.
4. Improvement phase
After a period of rigid economic liquidation and reorganisation, money accu-
mulates and seeks a use. Then follows a period of increasing business activity
with rising prices, a period of improvement or recovery. The improvement period
generally develops into the prosperity period, and a business cycle is completed.
These movements are constantly repeated in the order given as the cycle com-
pletes its swing (Figure 14.5).
Time Series Analysis 345

Y-axis

Cyclical fluctuation (variations)

Trend line
X-axis
0
Time (years)

FIGURE 14.5
Cyclical variations.

14.5.2 Importance of Measuring Cyclical Variation

1. It is extremely useful in framing suitable policies for stabilising the level of busi-
ness activity, that is, for avoiding periods of booms and depressions, as both are
bad for an economy – particularly depression, which brings about a complete
disaster and shatters the economy.
2. Study of past fluctuations is useful to determine the features of past behaviour
and helps in the study of fluctuations of business.
3. Measures of cycles help to forecast the future and prepare plans with the help of
projection of past cycles. In this way, future changes can be estimated.

14.5.3 Limitations of Measuring Cyclical Variation

1. Business cycles do not show regular periodicity – they differ widely in timing,
amplitude and pattern, which makes their study very tough and tedious.
2. The cyclical variations are mixed with erratic, random or irregular forces, which
makes it impracticable to isolate separately the effects of cyclical and irregular
forces.

See Table 14.1 for a comparison of the business cycle versus seasonal variations.

14.6 Irregular Variations, Random Movements, Unpredictable

Movements, Erratic Variations or Accidental Variations
Irregular variations refer to variations in business activities that do not repeat in a definite
pattern.
Irregular variations are caused by such isolated special occurrences as floods, earth-
quakes, strikes and wars. Sudden changes in demand or very rapid technological prog-
ress may also be included in this category. By their very nature, these movements are
346 Quantitative Techniques in Business, Management and Finance

TABLE 14.1
Business Cycle vs. Seasonal Variations
Cyclical Variations
Points (Business Cycle) Seasonal Variations
Duration Longer than a year. May be of any Occur during a year
duration, but normally the period
is 2–10 years.
Periodicity Do not ordinarily exhibit regular Exhibit regular periodicity
periodicity, as successive cycles
vary widely in timing, amplitude
and pattern.
Fluctuations Result from a different set of causes No fluctuations
Effect of factors The periods of prosperity, decline, Affected by factors such as
depression and improvement, weather, social customs and
viewed as four phases, are those that create seasonal
generated by factors other than patterns
weather, social customs and those
that create seasonal patterns.

Y-axis
Irregular variations

X-axis
0
Time (years)

FIGURE 14.6
Irregular variations.

very irregular and unpredictable. Quantitatively, it is almost impossible to separate out the
irregular movements from the cyclical movements (Figure 14.6).

14.6.1 Reasons for Recognising Irregular Movements

1. To suggest that on occasions it may be possible to explain certain movements in

the data as due to specific causes and to simplify further analysis.
2. To emphasise the fact that predictions of economic conditions are always subject
to a degree of error due to the unpredictable erratic influences that may occur.

14.7 Measurement of Trend

1. To find out trend characteristics in and of themselves.
In studying trend in and of itself, we ascertain the growth factor.
Time Series Analysis 347

For example, we can compare the growth in the steel industry with growth in the
economy as a whole or with the growth in other industries, or we can compare the
growth in one firm of the steel industry with the growth in the industry as a whole.
2. To enable us to eliminate trend in order to study other elements.
The elimination of trend leaves us with seasonal, cyclical and irregular factors.

Methods for determining trend

1. Freehand or graphic method

2. Semi-average method
3. Moving average method
4. Method of least squares

14.7.1 Freehand or Graphic Method of Measuring Trend

Using this method, the given data are plotted on graph paper, and a trend line is fitted to
the data just by inspecting the graph of the series.
When a trend line is fitted by the freehand method, an attempt should be made to make
it conform to the following conditions For example,

1. It should be smooth – either a straight line or a combination of long gradual curves.

2. The sum of the vertical deviations from the trend of the annual observations
above the trend should equal the sum of the vertical deviations from the trend of
the observations below the trend.
3. The sum of the squares of the vertical deviations of the observations from the
trend should be as small as possible.
4. The trend should bisect the cycles so that the area above the trend equals that
below the trend, not only for the entire series but for each full cycle. This condition
cannot always be met fully, but an attempt should be made to observe it as closely
as possible.

Exercise
Fit a trend line to the data in Table 14.2 by the freehand method (Figure 14.7).

Solution
The trend line drawn by the freehand method can be extended to predict future values.
However, since the freehand curve fitting is too subjective, this method should not be
used as a basis for predictions.

TABLE 14.2
Production of Zinc for 9 Years
Year 2004 2005 2006 2007 2008 2009 2010 2011 2012
Production of 22 24 26 23 25 27 25 28 27
zinc (tonnes)
348 Quantitative Techniques in Business, Management and Finance

Y-axis

36

34

32

30
Production of zinc (tonnes)

28

26

Trend line
24

22
Actual line
20

X-axis
0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Years

FIGURE 14.7
Trend by the freehand method.

Merits

1. This is the simplest method of measuring trend.

2. This method is very flexible in that it can be used regardless of whether the
trend is a straight line or curve.
3. A trend line drawn by a statistician experienced in computing trends and hav-
ing knowledge of the economic history of the concern or the industry under
analysis may be a better expression of the secular movement than a trend fitted
by the use of a rigid mathematical formula, which, while providing a good fit
to the points, may have no other logical justification.

Limitations

1. This method is highly subjective, because the trend line depends on the per-
sonal judgement of the investigator, and therefore, different persons may draw
different trend lines from the same set of data.
2. Since freehand curve fitting is subjective, it cannot have much value if it is
used as a basis for predictions.
3. Although this method appears simple and direct, in actuality, as experienced
statisticians can verify, it is very time consuming to construct a freehand trend
if a careful and conscientious job is done.

14.7.2 Semi-Average Method

When this method is used, the given data is divided into two parts, preferably with the
same number of years. For example, if we are given data from 2000 to 2017, that is, over a
period of 18 years, the two equal parts will be 9 years, that is, from 2000 to 2008 and from
2008 to 2017. In the case of an odd number of years such as 7, 9 or 11, two equal parts can
Time Series Analysis 349

be made simply by omitting the middle year. If data is given for 9 years from 2000 to 2008,
the two equal parts will be from 2000 to 2003 and from 2005 to 2008; the middle year, 2004,
will be omitted.
After the data has been divided into two parts, an average (arithmetic mean) of each part
is obtained. We thus get two points. Each point is plotted at the midpoint of the class inter-
val covered by the respective part, and then the two points are joined by a straight line,
which gives us the required trend line. The line can be extended downwards or upwards
to get intermediate values or to predict future values.

Exercise
Fit a trend line to the data in Table 14.3 by the method of semi-averages.

Solution
Since 7 years are given, the middle year will be left out, and an average of the first
3 years and the last 3 years will be obtained.
The average of the first 3 years is

104 + 107 + 116 327

= = 109
3 3

And the average of the last 3 years is

110 + 118 + 114 342

= = 114
3 3

Thus, we get two points, 109 and 114, which will be plotted corresponding to their respec-
tive middle years, that is, 2008 and 2012. By joining these two points, we obtain the required
trend line. The line can be extended and can be used either for prediction or for determin-
ing intermediate values.
The actual data and the trend line are shown in Figure 14.8.
When there are even numbers of years, such as 6, 8, 10 or 12, two equal parts can easily
be formed and an average of each part obtained. However, when the average is to be cen-
tred, there will be a problem. For example, if the data relates to 2002, 2003, 2004 and 2005,
the average will be centred corresponding to 1 July 2003, that is, in the middle of 2003 and
2004.

Exercise
The sale of a commodity in tonnes varied from January 2015 to December 2015 as shown
in Table 14.4.
Fit a trend line by the method of semi-averages.

TABLE 14.3
Sales of Firm P for 7 Years
Years 2007 2008 2009 2010 2011 2012 2013
Sales of Firm P 104 107 116 112 110 118 114
(thousand
units)
350 Quantitative Techniques in Business, Management and Finance

Y-axis

120

118

116
Sales of f irm ρ (thousand units)

114

Trend line
112

110

108 Actual data

106

104

102
X'-axis
X-axis
0 2007 2008 2009 2010 2011 2012 2013 2014 2015
Years

FIGURE 14.8
Trend by the method of semi-averages.

TABLE 14.4
Sales of Commodity for 12 Months
290 310 290 290 280 250
240 240 230 210 220 210

TABLE 14.5
Calculation of Trend Value by the Method of Semi-Averages
Month Sales (Tonnes) Month Sales (Tonnes)
January 290 July 240
1710 (Total) of 1350 (Total) of
February 310 August 240
first 6 months last 6 months
March 290 September 230
April 290 October 210
May 280 November 220
June 250 December 210

Solution
Calculation of trend value by the method of semi-averages (Table 14.5)

1710
Average of the first half = = 285 tonnes
6
Time Series Analysis 351

Y-axis
310

300

290

280

270
Trend
Sales (tonnes)

260 line

250

240
Actual data
230

220

210

200

X'-axis X-axis
0 Jan Feb Mar Apr May June Jul Aug Sep Oct Nov Dec
Y'-axis Months

FIGURE 14.9
Trend by the method of semi-averages.

1350
Average of the second half = = 225 tonnes
6

These two figures, namely, 285 and 225, will be plotted at the middle of their respective
periods, that is, at the middle of March–April and of September–October 2015.
By joining these two points, we get a trend line that describes the given data (Figure 14.9).

Merits
1. This method is simple to understand compared with the moving average
method and the method of least squares.
2. This is an objective method of measuring trend, as everyone who applies the
method is bound to get the same result (leaving aside arithmetical mistakes).

Limitations
1. This method assumes a straight-line relationship between the plotted points
regardless of whether that relationship exists or not.
2. If there are extremes in either half or both halves of the series, then the trend
line is not a true picture of the growth factor.

14.7.3 Moving Average Method

When a trend is to be determined by the method of moving averages, the average value
for a number of years (or months or weeks) is secured, and this average is taken as the
normal or trend value for the unit of time falling at the middle of the period covered in the
352 Quantitative Techniques in Business, Management and Finance

calculation of the average. The effect of the averaging is to give a smoother curve, lessening
the influence of the fluctuations that pull the annual figures away from the general trend.

Selection of period
While applying this method, it is necessary to select a period for the moving average, such
as 3-yearly, 5-yearly or 8-yearly moving average.
The period of moving average is to be decided in the light of the length of the cycles.
Since the moving average method is most commonly applied to data that is characterised
by cyclical movements, it is necessary to select a period for moving average that coincides
with the length of the cycle; otherwise, the cycle will not be entirely removed. This danger
is more severe the shorter the time period represented by the average.
When the period of moving average and the period of the cycle do not coincide, the mov-
ing average will display a cycle that has the same period as the cycle in the data.
Often, we find that the cycles in the data are not of uniform length. In such a case, we
should take a moving average period equal to or somewhat greater than the average period
of the cycle in the data. Ordinarily, the necessary period will range between 3 and 10 years
for general business series, but even longer periods are required for certain types of data.
The formula for a 3-yearly moving average will be

a+ b+c b+c+d c+d+e d+e+f

, , , …
3 3 3 3

and for a 5-yearly moving average

a+ b+c+d+e b+c+d+e+f c+d+e+f +g

, , …
5 5 5

Exercise
Find the trend of bank clearances by the method of moving average (assume a 5-yearly
cycle) (see Table 14.6).

Solution
Calculation of trend values by the method of moving averages (Table 14.7).

Exercise
Calculate 5-yearly and 7-yearly averages for the data of number of commercial indus-
trial failures in a country during 2000–2015 (Table 14.8).
Also plot the actual and trend values on a graph.

TABLE 14.6
Bank Clearances for 13 Years
Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Bank 53 79 76 66 69 94 105 87 79 104 97 92 101
clearances
(Crores of Rs.)
Time Series Analysis 353

TABLE 14.7
Calculation of Trend Values by the Method of Moving
Averages
Bank Clearances 5-Yearly 5-Yearly Moving
Year (Rs. Crores) Moving Totals Averages
2000 53 — —
2001 79 — —
2002 76 343 68.6
2003 66 384 76.8
2004 69 410 82.0
2005 94 421 84.2
2006 105 434 86.8
2007 87 469 93.8
2008 79 472 94.4
2009 104 459 91.8
2010 97 473 94.6
2011 92 — —
2012 101 — —

TABLE 14.8
Number of Commercial Industrial Failures in a Country during 2000–2015
Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
No. of 23 26 28 32 20 12 12 10 9 13 11 14 12 9 3 1
failures

Solution
Calculation of 5-yearly and 7-yearly moving average (Table 14.9)

Even Period and Moving Average

If the moving average is an even-period moving average, say, 4-yearly or 6-yearly, the
moving total and moving average, which are placed at the centre of the time span from
which they are computed, fall between two time periods.
This placement is inconvenient, since the moving average so placed would not coin-
cide with an original time period. We, therefore, synchronise moving averages and
original data. This process is called centring and always consists of taking a two-period
moving average of the moving averages.

Exercise
Work out the centred 4-yearly moving average for the data in Table 14.10.

Solution
Calculation of the centred 4-yearly moving average (Table 14.11).

Exercise
Assume a 4-yearly cycle, calculating the trend by the method of moving averages from
the data in Table 14.12 relating to the production of tea in India.
354 Quantitative Techniques in Business, Management and Finance

TABLE 14.9
Calculation of 5-Yearly and 7-Yearly Moving Average
5-Yearly 5-Yearly 7-Yearly 7-Yearly
No. of Moving Moving Moving Moving
Year Failures Total Average Total Average
2000 23 — — — —
2001 26 — — — —
2002 28 129 25.8 or 26 — —
2003 32 118 23.6 = 24 153 21.9 or 22
2004 20 104 20.8 = 21 140 20.0 = 20
2005 12 86 17.2 = 17 123 17.6 = 18
2006 12 64 12.6 = 13 108 15.4 = 15
2007 10 56 11.2 = 11 87 12.4 = 12
2008 9 55 11.0 = 11 81 11.6 = 12
2009 13 57 11.4 = 11 81 11.6 = 12
2010 11 59 11.8 = 12 78 11.1 = 11
2011 14 59 11.8 = 12 71 10.1 = 10
2012 12 42 9.8 = 10 63 9.0 = 9
2013 9 39 7.9 = 8 — —
2014 3 — — — —
2015 1 — — — —

TABLE 14.10
Tonnage of Cargo Cleared for 12 Years
Year 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Tonnage 1102 1250 1180 1340 1212 1317 1452 1549 1586 1476 1624 1586
of cargo
cleared

TABLE 14.11
Calculation of Centred 4-Yearly Moving Average
4-Yearly
Tonnage of 4-Yearly 4-Yearly Centred
Cargo Moving Moving Moving
Year Cleared Total Average Average
2002 1102 — — —
2003 1250 ← 4872 1218.00 —
2004 1180 ← 4982 1245.50 ← 1213.75
2005 1340 ← 5049 1262.25 ← 1253.87
2006 1212 ← 5321 1330.25 ← 1296.25
2007 1317 ← 5530 1382.50 ← 1356.37
2008 1452 ← 5904 1476.00 ← 1429.25
2009 1549 ← 6063 1515.75 ← 1495.87
2010 1586 ← 6235 1558.75 ← 1537.25
2011 1476 ← 6272 1568.00 ← 1563.37
2012 1624 — — —
2013 1586 — — —
Time Series Analysis 355

TABLE 14.12
Production of Tea in India for 10 Years
Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Production 464 515 518 467 502 540 557 571 586 612
of tea
(tonnes)

TABLE 14.13
Calculation of Trend by the Moving Average Method
Production 4-Yearly 4-Yearly 4-Yearly Moving
Year (Tonnes) Moving Total Average Average Centred
2004 464 — — —
2005 515 ←1964 491.0
2006 518 ←2002 500.50 ←495.7
2007 467 ←2027 506.75 ←503.6
2008 502 ←2066 516.50 ←511.6
2009 540 ←2170 542.50 ←529.5
2010 557 ←2254 563.50 ←553.0
2011 571 ←2326 581.50 ←572.5
2012 586 — — —
2013 612 — — —

Solution
Calculation of trend by the moving average method (Table 14.13).

Merits
1. This method is simple as compared with the method of least squares.
2. It is a flexible method of measuring trend for the reason that if a few more
figures are added to the data, the entire calculations are not changed – we only
get some more trend values.
3. If the period of moving average happens to coincide with the period of
cyclical fluctuations in the data, such fluctuations are automatically
eliminated.
4. The moving average follows the general movements of the data, and its shape
is determined by the data rather than the statistician’s choice of a mathematical
function.
5. It is particularly effective if the trend or a series is very irregular.

Limitations
1. Trend values cannot be computed for all the years. The longer the period of
moving average, the greater the number of years for which trend values cannot
be obtained.
2. Great care has to be exercised in selecting the period of moving average. No
hard and fast rules are available for the choice of the period, and one has to use
one’s own judgement.
3. Since the moving average is not represented by a mathematical function, this
method cannot be used in forecasting, which is one of the main objectives of
trend analysis.
356 Quantitative Techniques in Business, Management and Finance

4. Theoretically, we say that if the period of moving average happens to coin-

cide with the period of the cycle, the cyclical fluctuations are completely elimi-
nated, but in practice, since the cycles are by no means perfectly periodic, the
lengths of the various cycles in any given series will usually vary considerably,
and, therefore, no moving average can completely remove the cycle.
5. When the trend situation is not linear (a straight line), the moving average lies
either above or below the true sweep of the data. Consequently, the moving
average is appropriate for trend computations only when
a. The purpose of investigation does not call for current analysis or forecasting
b. The trend is linear
c. The cynical variations are regular in both period and amplitude.

Unfortunately, these conditions are encountered very infrequently.

14.7.4 The Method of Least Squares

This method is most widely used in practice. It is a mathematical method, and with its
help a trend line is fitted to the data in such a manner that the following two conditions
are satisfied.

1. Σ ( Y − Ye ) = 0

That is, the sum of deviations of the actual values of Y and the computed values
of Y is zero.
2. Σ ( Y − Ye ) is least
2

That is, the sum of the squares of the deviations of the actual and computed
values is least from this line. That is why this method is called the method of least
squares. The line obtained by this method is known as the line of best fit.
This method of least squares may be used to fit either a straight-line trend or a
parabolic trend.

The straight-line trend is represented by the equation

Ye = a + bX

where:
Ye = Used to designate the trend values to distinguish them from actual Y values.
a = Y-intercept or the value of the Y variable when X = 0.
b = The slope of the line, or the amount of change in the Y variable that is associated
with a change of one unit in the X variable.
X = Variable in time series analysis representing time.

Important points for fitting of straight-line trend by the least-squares method

1. Select year as origin.

2. Decide unit of time represented by X, such as half year, 1 year or 5 years.
3. Decide type of units measured by Y, such as production (in tonnes quintals etc.),
sales (in rupees, dollars, pounds, etc.), prices (in rupees, dollars, pounds, etc.) and
employment of workers (in thousands, lakhs, crores etc.).
Time Series Analysis 357

Calculation of constants a and b

The values of the constants a and b are determined by solving the following two normal
equations:

ΣY = Na + bΣX (14.1)

ΣXY = aΣX + bΣX 2 (14.2)

where:
N = Number of years (months or any other period) for which data is given
Equation 14.1 = Nearly the summation of the given function
Equation 14.2 = The summation of X multiplied into the given function
a = Arithmetic mean of Y
b = Rate of change

We can measure the variable X from any point of time as the origin, such as the first year.
But the calculations are very much simplified when the midpoint in time is taken as the
origin, because in that case, the negative values in the first half of the series balance out the
positive values in the second half, so that ΣX = 0.
In other words, the time variable is measured as a deviation from its mean.
Since ΣX = 0, Equations 14.1 and 14.2 would take the form

ΣY = Na

ΣXY = bΣX 2 (14.3)

Now, the values of a and b can be determined easily.

Since ∑ Y = Na

ΣY
∴a = =Y
N

Since ΣXY = bΣX 2

ΣXY
∴b =
ΣX 2

The constant a gives the arithmetic means of Y, and the constant b indicates the rate of
change.

Note:

1. In the case of an odd number of years, when the deviations are taken from
the middle year, ΣX will always be zero, provided there is no gap in the data
given.
358 Quantitative Techniques in Business, Management and Finance

2. In the case of an even number of years, ΣX will be zero if the X origin is placed
midway between the two middle years.

For example, if the years are 2011, 2012, 2013, 2014, 2015 and 2016, we can take deviations
from the middle year, 2013.5. Thus,

1. The deviations would be −2.5, −1.5, −0.5, +0.5, +1.5 and +2.5 for the various years
and
2. The total ΣX would be zero.

Hence, in both odd as well as even numbers of years, we can use the simple procedure
of determining the values of the constants a and b.

Exercise
Table 14.14 shows the figures of production (in thousands of maunds) of a sugar factory.

1. Fit a straight-line trend to these figures.

2. Plot these figures on a graph and show the trend line.

Solution
1. Fitting the straight-line trend (Table 14.15)/
The equation of the straight-line trend is

Ye = a + bX

TABLE 14.14
Production of a Sugar Factory
Year 2010 2011 2012 2013 2014 2015 2016
Production 80 90 92 83 94 99 92
(thousand
maunds)

TABLE 14.15
Fitting the Straight-Line Trend
Production
(Thousand Trend
Year Maunds) (Y) X XY X2 values (Ye)
2010 80 −3 −240 9 84
2011 90 −2 −180 4 86
2012 92 −1 −92 1 88
2013 83 0 0 0 90
2014 94 1 94 1 92
2015 99 2 198 4 94
2016 92 3 276 9 96
N=7 ΣY = 630 ΣX = 0 ΣXY = 56 ΣX2 = 28 Ye = 630
Time Series Analysis 359

Since ΣX = 0,

ΣY ΣXY
a= ,b =
N ΣX 2

Here, ΣY = 630, N = 7, ΣXY = 56, ΣX2 = 28

ΣY 630
∴ a= = = 90
N 7

And

ΣXY 56
b= = =2
ΣX 2 28

Hence, the equation of the straight-line trend is

Ye = 90 + 2X

Origin, 2013
X units, one year
Y units, production in thousands of maunds

For X = −3, Y0 = 90 + 2 (−3) = 84

For X = −2, Y0 = 90 + 2 (−2) = 86

For X = −1, Y0 = 90 + 2 (−1) = 88

Similarly, by putting X = 0, 1, 2, 3, we can obtain other trend values. However,

since the value of b is constant, only the first trend value need be obtained, and
then, if the value is positive, we may continue adding the value of b to every
preceding value.
2. The graph of this data is shown in Figure 14.11.

For example, in the above case, for 2013, it will be 86+2 = 88, and so on. If b is nega-
tive, then instead of adding, we will deduct.

Exercise
Apply the method of least squares to obtain the trend values from the following data
and show that Σ(Y–Ye) = 0 (Table 14.16).
Also, predict the sales for the year 2014.

Solution
Calculation of trend values by method of least squares (Table 14.17).
The equation of the straight-line trend is

Ye = a + bX
360 Quantitative Techniques in Business, Management and Finance

Y-axis
40
Actual data
35 5-yearly moving average
7-yearly moving average
30

25
No. of failures

20

15

10

X-axis
0 2000 2002 2004 2006 2008 2010 2012 2014

Years

FIGURE 14.10
Trend by the method of moving averages.

Y-axis
105

100

95
Production (thousands of maunds)

90
Trend line
85

Actual data
80

75

70

0 2010 2011 2012 2013 2014 2015 2016 2017 X-axis

Years

FIGURE 14.11
Linear trend by the method of least squares.

TABLE 14.16
Data of Sales for 5 Years
Year 2004 2005 2006 2007 2008
Sales (tonnes) 100 120 110 140 80
Time Series Analysis 361

TABLE 14.17
Calculation of Trend Values by Method of Least Squares
Deviations
from
Middle
Year Sales (Y) Year (X) XY X2 Ye (Y – Ye)
2004 100 −2 −200 4 114 −14
2005 120 −1 −120 1 112 +18
2006 110 0 0 0 110 0
2007 140 1 140 1 108 32
2008 80 2 160 4 106 −26
N=5 ΣY = 550 ΣX = 0 ΣXY = −20 ΣX2 = 10 Σ Ye = 0 Σ(Y – Ye) = 0

Since ΣX = 0

ΣY ΣXY
a= ,b =
N ΣX 2

ΣY = 550, N = 5, ΣXY = −20, ΣX2 = 10

Substituting the values

550 20
a= = 110 b = − = −2
5 10

The required equation is

Ye = 110 − 2X

For X = −2, Ye = 110 − 2 ( −2 ) = 114

Now, the other trend values will be obtained by deducting the value of b from the
preceding value. Thus, for 2005, the trend value will be 114 − 2 = 112 (since the value of b
is negative). For 2014, likely sales = 94 tonnes (since X would be +8 for 2014).

Exercise
Fit a straight-line trend by the method of least squares to the data in Table 14.18.
Assuming that the same rate of change continues, what would be the predicted earn-
ings for the year 2012?

TABLE 14.18
Earnings (Rs. in Thousands) for 8 Years
Year 2003 2004 2005 2006 2007 2008 2009 2010
Earnings 38 40 65 72 69 60 87 95
(Rs. in
thousands)
362 Quantitative Techniques in Business, Management and Finance

TABLE 14.19
Fitting of Straight-Line Trend by the Method of Least Squares
Earnings Deviations
(in Thousands) Deviations Multiplied
Year (Y) from 2006.5 by 2 (X) XY X2
2003 38 −3.5 −7 −266 49
2004 40 −2.5 −5 −200 25
2005 65 −1.5 −3 −195 9
2006 72 −0.5 −1 −72 1
2007 69 0.5 1 69 1
2008 60 1.5 3 180 9
2009 87 2.5 5 435 25
2010 95 3.5 7 665 49
N=8 ΣY = 526 ΣX = 0 ΣXY = 616 ΣX2 = 168

Solution
Fitting of straight-line trend by the method of least squares (Table 14.19)

Yc = a + bX

ΣY 526
a= = = 65.75
N 8

ΣXY 616
b= = = 3.66
ΣX 2 168

Y = 65.75 + 3.66X

For 2012, X will be +11.

When X is +11, Y will be

Y = 65.75 + 3.66 (11)

= 65.75 + 40.37 = 106.

Thus, the estimated earnings for the year 2012 are Rs. 106.12 thousands.

Note:
The same result will be obtained if we do not multiply the deviations by 2. But in that
case, our computations would be more difficult, as seen here (Table 14.20):

ΣY 526
a= = = 65.75
N 8

ΣXY 308
b= = = 7.33
ΣX 2 48
Time Series Analysis 363

TABLE 14.20
Fitting of Straight-Line Trend by the Method of Least Squares
Sales
(Thousands Deviations
Year of Rupees) (Y) from 2006.5 (X) XY X2
2003 38 −3.5 −133.00 12.25
2004 40 −2.5 −100.00 6.25
2005 65 −1.5 −97.50 2.25
2006 72 −0.5 −36.00 0.25
2007 69 0.5 34.50 0.25
2008 60 1.5 90.00 2.25
2009 87 2.5 217.50 6.25
2010 95 3.5 332.50 12.25
N=8 ΣY = 526 ΣX = 0 ΣXY = 308 ΣX2 = 42.00

TABLE 14.21
Production of a Sugar Factory
Year 2004 2005 2006 2007 2008 2009 2010
Production (thousands 77 88 94 85 91 98 90
of maunds)

The advantage of this method is that the value of b gives an annual increment of change
rather than six monthly increments, as in the first method. Hence, we will not have to
double the value of b to obtain a yearly increment. It is clear from this problem that in the
first case, the value of b is half of what we obtain from the second method (b was 3.66 in
the first case and 7.33 in the second case).

Exercise
The figures of production (in thousands of maunds) of a sugar factory are given in
Table 14.21.

1. Fit a straight line by the least-squares method, and tabulate the trend values.
2. Eliminate the trend. What components of the time series are thus left over?
3. What is the monthly increase in the production of sugar?

Solution
1. The equation of the straight-line trend is

Ye = a + bX

Since ΣX is not zero, the values of a and b will be obtained directly by solving
the following two normal equations (Table 14.22):

ΣY = Na + bΣX (14.4)

ΣXY = aΣX + bΣX 2 (14.5)

364 Quantitative Techniques in Business, Management and Finance

TABLE 14.22
Fitting of Straight-Line Trend by the Method of Least Squares
Taking Trend
Production 2007 as Values
Year (Y) Origin (X) XY X2 (Ye) Y – Ye
2004 77 −4 −308 16 83.283 −6.283
2005 88 −2 −176 4 86.043 1.957
2006 94 −1 −94 1 87.423 6.577
2007 85 0 0 0 88.803 −3.803
2008 91 1 91 1 93.183 0.817
2009 98 2 196 4 91.563 6.437
2010 90 5 450 25 95.703 −5.703
N=7 ΣY = 623 ΣX = 1 ΣXY = +159 ΣX2 = 51 ΣYe = 623 Σ(Y – Ye) = 0

623 = 7a + b (14.6)

159 = a + 51b (14.7)

Multiplying Equation 14.7 by 7, we get

1113 = 7a + 357b (14.8)

Deducting Equation 14.8 from Equation 14.6,

−490 = −356b

490
b= = 1.38
356

Substituting the value of b in Equation 14.6,

623 = 7a + 1.38

7a = 623 − 1.38 = 621.62

A = 88.803

So, the equation of the straight-line trend is

Y = 88.803 + 1.38X

When

X = −4, Y = 88.803 + 1.38 ( −4 )

= 88.803 − 5.52 = 83.283

Time Series Analysis 365

When

X = −2, Y = 88.803 + 1.38 ( −2 )

= 88.803 − 2.76 = 86.043

When X = −1, Y = 88.803 − 1.38 = 87.428

When X = 1, Y = 88.803 + 1.38 = 90.183
When X = 2, Y = 88.803 + (1.38 × 2 ) = 91.563
When X = 5, Y = 88.803 + (1.38 × 5 ) = 95.7 03
2. After eliminating the trend, we are left with cyclical and irregular variations.
3. The monthly increase in the production of sugar is b/12, that is, 1.38/12 = 0.115
thousands of maunds.

Merits
1. This is a mathematical method of measuring trend, and as such, there is no
possibility of subjectiveness.
2. The line obtained by this method is called the line of best fit because it is this line
from where the sum of the positive and negative deviations is zero and the sum of
the square of the deviations is least; that is, Σ ( Y − Ye ) = 0 and Σ ( Y − Ye ) is least.
2

Limitations
Mathematical curves are useful to describe the general movement of a time series, but it is
doubtful whether any analytical significance should be attached to them, except in special
cases. It is seldom possible to justify on theoretical grounds any real dependence of a vari-
able on the passage of time. Variables do change in a more or less systematic manner over
time, but this can usually be attributed to the operation of other explanatory variables.
Mathematical methods of fitting trend are not foolproof; in fact, they can be the source
of some of the most serious errors that are made in statistical work. They should never
be used unless rigidly controlled by a separate logical analysis.

14.8 Second-Degree Parabola

The simplest example of the non-linear trend is the second-degree parabola, the equation
of which is written in the form

Yc = a + bX + cX 2

When numerical values for a, b and c have been derived, the trend value for any year may
be computed by substituting in the equation the value of X for that year. The values of a, b
and c can be determined by solving the following three normal equations simultaneously:

(a) ΣY = Na + bΣX + cΣX 2

( b) ΣXY = aΣX + bΣX 2 + cΣX 3

(c) ΣX 2Y = aΣX 2 + bΣX 3 + cΣX 4

366 Quantitative Techniques in Business, Management and Finance

Note:

1. The first equation is merely the summation of the given function.

2. The second is the summation of X multiplied into the given function.
3. The third is the summation of X 2 multiplied into the given function.

When time origin is taken between two middle years, ΣX will be zero. In that case, the
above equations are reduced to

(a) ΣY = Na + cΣX 2

( b) ΣXY = bΣX 2

(c) ΣX 2Y = aΣX 2 + cΣX 4

The value of b can now directly be obtained from Equation (b) and that of a and c by
solving (a) and (b) simultaneously. Thus,

a=
Σ ( Y ) − cΣ X 2 ( )
N
ΣXY
b=
ΣX 2

c=
( ) ( )
N ΣX 2 Y − Σ X 2 Σ ( Y )
NΣ ( X ) − ( ΣX )
2
4 2

14.9 Measurement of Seasonal Variations

Most of the phenomena in economics and business show seasonal patterns. When data is
expressed annually, there is no seasonal variation. However, monthly or quarterly data fre-
quently exhibits strong seasonal movements, and considerable interest attaches to devis-
ing a pattern of average seasonal variation. For example, it is necessary to decide whether
weekly, quarterly or monthly indexes are required. This will be decided in the light of the
nature of the problem and the type of data available.

14.9.1 Seasonal Index

To obtain a statistical description of a pattern of seasonal variation, it will be desirable
to first free the data from the effects of trend, cycles and irregular variations. Once these
other components have been eliminated, we can calculate, in index form, a measure of
seasonal variations, which is usually referred to as a seasonal index.
Time Series Analysis 367

14.9.2 Criteria for Computing an Index of Seasonal Variation

1. It should measure only the seasonal forces in the data. It should not be influenced
by the forces of trend or cycle that may be present.
2. It should modify the erratic fluctuations in the data with an acceptable system of
averaging.
3. It should recognise slowly changing seasonal patterns that may be present and
modify the index to keep up with these changes.

14.9.3 Methods Used for Measuring Seasonal Variations

1. Method of simple averages (weekly, monthly or quarterly).

2. Ratio-to-trend method or percentage-to-trend method
3. Ratio-to-moving average method or percentages of moving average method
4. Link relative method

14.9.3.1 Method of Simple Averages (Weekly, Monthly or Quarterly)

This is the simplest method of obtaining a seasonal index.

1. Steps for calculating the index:

a. Arrange the unadjusted data by years and months (or quarters if quarterly
data is given).
b. Find the totals of January, February, March and so on.
Divide each total by the number of years for which data is given. For exam-
ple, if we are given monthly data for 4 years, we first obtain the total for each
month for 5 years and divide each total by four to obtain an average.
c. Obtain an average of monthly averages by dividing the total of monthly aver-
ages by 12.
d. Taking the average of monthly averages as 100, compute the percentages of
various monthly averages as follows:

Seasonal index for January

Monthly average for January

= × 100
Average of monthly averages

Note:
If, instead of the average of each month, the totals of each month are obtained, we will get
the same result.

Exercise
Consumption of monthly electric power in millions of kilowatt hours for street lighting
in the United States during 2011–2015 is given in Table 14.23.
Find the seasonal variations by the method of monthly averages.
368 Quantitative Techniques in Business, Management and Finance

TABLE 14.23
Monthly Consumption of Electric Power
Year Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec.
2011 348 281 278 250 231 216 223 245 269 302 325 347
2012 342 309 299 268 249 236 242 262 288 321 342 364
2013 367 328 320 287 269 251 259 284 309 245 367 394
2014 392 349 342 311 290 273 282 305 328 364 379 417
2015 420 378 370 334 314 296 305 330 356 396 422 452

TABLE 14.24
Calculation of Seasonal Indices by the Method of Monthly Averages
Monthly Consumption of Electric Power Monthly
Total for 5-Yearly
Month 2011 2012 2013 2014 2015 5 Years Average Percentage
1 2 3 4 5 6 7 8 9
Jan. 318 342 367 392 420 1,839 367.8 116.1
Feb. 281 309 328 349 378 1,645 329.0 103.9
Mar. 278 299 320 342 370 1,609 321.8 101.6
Apr. 250 268 287 311 334 1,450 290.0 91.4
May 231 249 269 290 314 1,353 270.6 85.5
June 216 236 251 273 296 1,272 254.4 80.3
July 223 242 259 282 305 1,311 262.2 82.8
Aug. 245 262 284 305 330 1,426 285.2 90.0
Sept. 269 288 309 328 356 1,550 310.0 98.0
Oct. 302 321 245 364 396 1,728 345.6 109.1
Nov. 325 342 367 379 422 1,845 369.0 116.6
Dec. 347 364 394 417 452 1,974 394.8 124.7
Total 19,002 3,800.4 1,200
Average 1,583.5 316.7 100

Solution
Calculation of seasonal indices by the method of monthly averages (Table 14.24). The
calculations in the table are explained as follows:

1. Column 7 gives the total for each month for the 5 years.
2. In Column 8, each total of Column 7 has been divided by 5 to obtain an average
for each month.
3. The average of monthly averages is obtained by dividing the total of monthly
averages by 12.
4. In Column 9, each monthly average has been expressed as a percentage of the
average of monthly averages. Thus, the percentage for January

367.8
= × 100 = 116.1
316.7
Time Series Analysis 369

Percentage for February

329.0
= × 100 = 103.9
316.7

Percentage for March

321.8
= × 100 = 101.6
316.7

And so on.
If, instead of monthly data, we are given weekly or quarterly data, we will compute
weekly or quarterly averages by following the same procedure.

Merits
This method is the simplest of all methods of measuring seasonality.

Limitations

1. It is not a very good method. It assumes that there is no trend component in

the series, that is, computation of seasonal indices (CSI) = 0. But this is not a
justified assumption.
2. Most economic series have trends, and, therefore, the seasonal index computed
by this method is actually an index of trends and seasonality.
3. The effects of cycles on the original values may or may not be eliminated by
the averaging process. This depends on the duration of the cycle and the term
of the average, that is, on the number of months included in the average. Thus,
this method is seldom of any value. In its simplest form, the method only
serves a purpose where no definite trend exists.

14.9.3.2 Ratio-to-Trend Method or Percentage-to-Trend Method

This method of calculating a seasonal index is relatively simple and yet an improvement
over the method of simple averages. It is based on the multiplicative model of time series
analysis. It assumes that seasonal variation for a given month or quarter is a constant frac-
tion of trend value. Random variations are assumed to disappear when ratios are changed.
Seasonal variation for a given month is a constant fraction of trend. The ratio-to-trend
presumably isolates the seasonal factor in the following manner. Trend is eliminated when
the ratios are computed. In effect,

T×S× C×I
= S× C×I
T
where:
T = trend
S = seasonal component
C = cyclical component
I = irregular component

Random elements are supposed to disappear when the ratios are averaged. A careful
selection of the period of years used in the computation is expected to cause the influences
370 Quantitative Techniques in Business, Management and Finance

of prosperity or depression to offset each other and thus remove the cycle. For series that
are not subject to pronounced cyclical or random influences and for which trend can be
computed accurately, this method may suffice.
Steps in the calculation of a seasonal index

1. Trend values are obtained by applying the method of least squares.

2. The next step is to divide the original data month by month by the corresponding
trend values and to multiply these ratios by 100.
3. The values so obtained are now free from trend and the problem that remains is
to free them also of irregular and cyclical movements.
4. In order to free the values from irregular and cyclical movements, the figures
given for the various years for January, February, March and so on are averaged
with any one of the usual measures of central value, for instance, the median or
the mean. If the data are examined month by month, it is sometimes possible to
ascribe a definite cause to usually high or low values.
5. The seasonal index for each month is expressed as a percentage of the average
month. The sum of 12 values must equal 1200 or 100%. If it does not, an adjust-
ment is made by multiplying each index by a suitable factor (1200/the sum of the
12 values). This gives the final seasonal index.

Exercise
Compute seasonal variations by the ratio-to-trend method from the information given
in Table 14.25.

Solution
For determining seasonal variation by the ratio-to-trend method, first we will deter-
mine the trend for yearly data and then convert it to quarterly data (Table 14.26).

Σy 170
a= = = 34
N 5
Σxy 100
b= = = 10
Σx 2 10

Calculation of quarterly trend values

Yearly increment 10
Quarterly increment = = = 2.5
4 4

TABLE 14.25
Quarterly Data for 5 Years
Year Quarter I Quarter II Quarter III Quarter IV
2010 10 12 18 20
2011 20 23 27 30
2012 24 26 32 38
2013 38 42 48 52
2014 46 54 56 64
Time Series Analysis 371

TABLE 14.26
Calculation of Trend by Method of Least Squares
Yearly
Yearly Average Trend
Year Total (y) x xy X2 Y = a bx Value
2010 60 15 −2 −30 4 34 + 10(−2) 14
2011 100 25 −1 −25 1 34 + 10(−1) 24
2012 120 30 0 0 0 34 + 10(0) 34
2013 180 45 1 45 1 34 + 10(1) 44
2014 220 55 2 110 4 34 + 10(2) 54
Total ∑y = 170 ∑Xy = 100 ∑X2 = 10 170

Consider 2010. The trend value for the middle quarter that is half of 2nd and half of
3rd is 14. So, the trend value of the 2nd quarter will be

2.5
14 − = 14 − 1.25 = 12.75
2

For the third quarter, it will be

2.5
14 + = 14 + 1.25 = 15.25
2

For the first quarter, it will be

12.75 − 2.5 = 10.25

For the fourth quarter, it will be

15.25 + 2.5 = 17.75

Quarterly trend values (Table 14.27)

The given values are to be expressed as the percentages of the corresponding trend
O
values = × 100 (Table 14.28)
T
The average of quarterly average of trend figures:

92.32 + 95.05 + 104.64 + 108.23 400.24

= = 100.06
4 4

Quarterly seasonal index

92.32
For Quarter I = × 100 = 92.26
100.06

95.05
For Quarter II = × 100 = 94.99
100.06
372 Quantitative Techniques in Business, Management and Finance

TABLE 14.27
Calculation of Quarterly Trend Values
Year Quarter I Quarter II Quarter III Quarter IV Total
2010 10.25 12.75 15.25 17.75 56
2011 20.25 22.75 25.25 27.75 96
2012 30.25 32.75 35.25 37.75 136
2013 40.25 42.75 45.25 47.75 176
2014 50.25 52.75 55.25 57.75 216
Total 680

TABLE 14.28
Given Quarterly Values as Percentage of Trend Values
Year Quarter I Quarter II Quarter III Quarter IV
2010 10 94.12 118.03 112.67
× 100 = 97.56
10.25
2011 98.76 101.10 106.93 108.11
2012 79.33 79.39 90.78 100.66
2013 94.41 98.25 106.08 108.90
2014 91.54 102.37 101.36 110.80
Total 461.60 475.23 523.18 541.14
Average 92.32 95.05 104.64 108.23

104.64
For Quarter III = × 100 = 104.58
100.06

108.23
For Quarter IV = × 100 = 108.17
100.06

Merits
1. The method is more logical and useful.
2. The method considers all the values.
3. When the period of study is short, this method is more useful to obtain sea-
sonal indices.
4. It has an advantage over the moving average procedure too, for it has a ratio-to-
trend value for each month for which data are available. Thus, there is no loss
of data, as occurs in the case of moving averages.
5. It is simple to compute and easy to understand.

Limitations
If there are pronounced cyclical swings in the series, the trend, whether a straight line or
a curve, can never follow the actual data as closely as a 12-month moving average does.
In consequence, a seasonal index computed by the ratio-to-moving-average method
may be less biased than one calculated by the ratio-to-trend method.

14.9.3.3 Ratio-to-Moving Average Method or Percentages of Moving Average Method

This is the most widely used method of measuring seasonal variations.
Time Series Analysis 373

Steps
1. Eliminate seasonality from the data by ironing it out of the original data. Since
seasonal variations recur every year, that is, since the fluctuations have a time
span of 12 months, a centred 12-month moving average tends to eliminate these
fluctuations.
2. Express the original data for each month as a percentage of the centred 12-month
moving average corresponding to it.
3. Divide each monthly item of the original data by the corresponding 12-month
moving average, and list the quotients as ‘percent of moving average’.
4. By using the median as an average, we can obtain the typical seasonal relative for
each month, which will not be affected by irregular factors.
Sometimes, a so-called modified mean is used as an average for each month.
Here, extreme values are omitted before the arithmetic mean is taken. In any
array of seasonal relatives for each month, a value, or several values, on one
end or both ends that may be relatives is taken. A separate table is prepared, in
which the calculations involved in this step are shown. These means are prelimi-
nary seasonal indexes. They should average 100% or total 1200 for 12 months by
definition.
5. If the total is not equal to 1200 or 100%, an adjustment is made to eliminate the
discrepancy. The adjustment consists of multiplying the average of each month
obtained in Step 4 by

1200
the total of the modified mean for 12 months

This adjustment is made not only to achieve accuracy, but also because when we come
to eliminate seasonality from the original data, we do not wish to raise or lower the level
of the data unduly. Thus, if a seasonal index aggregates more than 1200 (or averages more
than 100), the original data adjusted in terms of the index will total less than the unad-
justed original data. If it totals less than 1200, the opposite will be true.
The logical reasoning behind this method follows from the fact that the 12-month mov-
ing average can be considered to represent the influence of cycle and trend (C × T). If
the actual value for any month is divided by the 12-month moving average centred to
that month, presumably cycle and trend are removed. This may be represented by the
expression

T×S× C×I
= S×I
T×C

Thus, the ratio to the moving average, from which this method gets its name, represents
irregular and seasonal influences. If the ratios for each worked over a period of years are
then averaged, most random influences will usually be eliminated. Hence, in effect,

S×I
=S
1
374 Quantitative Techniques in Business, Management and Finance

Merits
1. This method is considered to be most satisfactory and, as such, is more widely
used in practice than other methods.
2. The index obtained by the ratio-to-moving average method ordinarily does not
fluctuate as much as the index based on straight-line trends.
3. Mathematical methods of avoiding the effects of the business cycle are not usu-
ally needed, for the 12-month moving average follows the cyclical course of the
actual data quite closely. Therefore, the index ratios are often more representative
of the data from which they are obtained than is the case in the ratio-to-trend
method.
4. Also, the ratio-to-moving average method allows greater flexibility.

Limitations
Seasonal indices cannot be obtained for each month for which data is available. When a
12-month moving average is taken, 6 months at the beginning and 6 months at the end are
left out, for which we cannot calculate seasonal indices.

14.9.3.4 Link Relative Method

Among all the methods of measuring seasonal variations, the link relative method is the
most difficult.

Steps

1. Calculate the link relatives of the seasonal figures. Link relatives are calculated
by dividing the figure of each season by the figure of the immediately preceding
season and multiplying it by 100.

Current season’s figure

×100
Previous season’s figure

There percentages are called link relatives since they link each month (or quarter
or other time period) to the preceding one.
2. Calculate the average of the link relatives for each season. While calculating the
average, we might take the arithmetic average, but the median is probably better.
The arithmetic average would give undue weight to extreme cases which were not
due primarily to seasonal influences.
3. Convert these averages into chain relatives on the basis of the first season.
4. Calculate the chain relatives of the first season on the basis of the last season.
There will be some difference between the chain relative of the first season and
the chain relative calculated by the previous method (arithmetic average). This
difference will be due to the effect of long-term changes. It is therefore, necessary
to correct these chain relatives.
5. For correction, the chain relative of the first season calculated by the first method
is deducted from the chain relative (of the first season) calculated by the second
Time Series Analysis 375

method (chain relatives). The difference is divided by the number of seasons. The
resulting figure multiplied by 1, 2, 3 (and so on) is deducted from the chain rela-
tives of the second, third, fourth (and so on) seasons, respectively. These are cor-
rected chain relatives.
6. Express the corrected chain relatives as percentages of their averages. These pro-
vide the required seasonal indices by the method of link relatives.

Exercise
Apply the method of link relatives to the data in Table 14.29 and calculate seasonal
indices.

Solution
Calculation of seasonal indices by method of link relatives (Table 14.30).
In Table 14.30, the figure for correlation has been calculated as follows:

Chain relative of the first quarter

(on the basis of the first quarter) = 100
Chain relative of the first quarter
(on the basis of the last quarter)

TABLE 14.29
Quarterly Data for 5 Years
Quarter 2001 2002 2003 2004 2005
I 6.0 5.4 6.8 7.2 6.6
II 6.5 7.9 6.5 5.8 7.3
III 7.8 8.4 9.3 7.5 8.0
IV 8.7 7.3 6.4 8.5 7.1

TABLE 14.30
Calculation of Seasonal Indices by Method of Link Relative
Year Quarter I Quarter II Quarter III Quarter IV
2001 — 108.3 120.0 111.5
2002 62.1 146.3 106.3 86.9
2003 93.2 95.6 143.1 68.8
2004 112.5 80.6 129.3 113.3
2005 77.6 110.6 109.6 88.8
Arithmetic 345.4 541.4 608.3 469.3
average = 86.35 = 108.28 = 121.66 = 93.86
4 5 5 5
Chain 100 100 × 108.28 121.66 × 108.28 93.86 × 131.7
relatives = 108.28 = 131.7 = 123.6
100 100 100
Corrected 100 108.28 − 1.675 = 106.605 131.70 − 3.35 = 128.35 123.6 − 5.025 = 118.575
chain
relatives
Seasonal 100 106.28 128.35 118.575
indices × 100 = 93.73 × 100 = 113.2 × 100 = 104.6
113.385 113.385 113.385
376 Quantitative Techniques in Business, Management and Finance

86.35 × 123.6
= = 106.7
100

Difference between these chain relatives

= 106.7 − 100 = 6.7

Difference per quarter

6.7
= = 1.675
4

Adjusted chain relatives are obtained by subtracting 1 × 1.675, 2 × 1.675 and 3 × 1.675
from the chain relatives of the second, third and fourth quarters, respectively.

Calculation of seasonal variations indices

Average of corrected chain relatives

100 + 106.605 + 128.35 + 118.575 435.54

= = = 113.385
4 4

Seasonal variation index

Correct chain relatives × 100

=
113.85

14.10 Summary
Some procedures for time series analysis have been described in this chapter with a view
to making more accurate and reliable forecasts of the future. Quite often, the question
that puzzles a person is how to select an appropriate forecasting method. Many times, the
problem context or time horizon involved will decide the method or limit the choice of
methods.
The decomposition method has been discussed. Here, the time series is broken up
into seasonal, trend, cycle and random components from the given data and recon-
structed for forecasting purposes. A detailed example to illustrate the procedure is
also given.

REVIEW QUESTIONS

1. Give the meaning of time series.

2. What are the components of time series?
3. What do you mean by trend?
Time Series Analysis 377

4.What do you understand by seasonal indices?

5.What is irregular variation?
6.What are the methods used for measuring trend?
7.Define the method of least squares to measure trend.
8.What are the methods used for measuring seasonal variations?
9.Explain clearly the meaning of time series analysis. Indicate the importance of
such analysis in business.
10. What is a time series? Mention its important components. Explain these compo-
nents with examples. Discuss briefly the methods of smoothing a time series.
11. a. What are the different components of fluctuation in a time series? Elucidate the
methods available for measuring the trend components and their relative merits
and demerits.
b. What are the limitations and advantages of the moving averages method of
trend fitting?
12. What is trend? How would you find out the trend values by the method of least
squares? Illustrate by a numerical example.
13. Explain how a growth factor, a decline factor, a seasonal factor and a cyclical factor
affect a variable over a period of time.
14. What do you understand by the term moving average? Indicate its uses.
15. What is a secular trend? Discuss any two methods of isolating trend values in a
time series.
16. What is meant by seasonal fluctuation? State the procedure for obtaining a sea-
sonal index by the method of monthly averages.
17. Explain the different methods of measuring seasonal components in time-series
data. How will you eliminate the seasonal components?

SELF-PRACTICE PROBLEMS

1. Find the trend line by using the method of least squares and estimate the trend
value for the year 2015.

Year 2008 2009 2010 2011 2012 2013

Production 30 60 40 70 50 65

2. Fit a trend line by the method of least squares for the following data and estimate
the trend value for the year 2016.

Year 2009 2010 2011 2012 2013 2014 2015

Sales 50 80 95 65 30 40 45

3. Fit a trend line by the method of least squares for the following data:

Year 2007 2008 2009 2010 2011 2012 2013 2014 2015
Income 630 714 883 837 817 749 729 886 869
378 Quantitative Techniques in Business, Management and Finance

4. Fit a straight-line trend by the method of least squares to the following data and
obtain the trend value for the year 2017:

Year 2004 2005 2006 2007 2008 2009

Production (Lakhs 3.6 3.8 4.4 4.7 5.6 7.3
of tonnes)
Year 2010 2011 2012 2013 2014 2015
Production (Lakhs 7.1 7.6 7.7 9.0 9.0 10.1
of tonnes)

[Ans: Y = 6.66 + 0298x; Trend value for 1972 = 10.534]

5. Compute the trend value by the method of least squares from the data given
below:

Year 2008 2009 2010 2011 2012 2013 2014 2015

Number of 56 55 51 47 42 38 35 32
goats (lakhs)

[Ans: Y = 445–1.86x]
6. Find trend values (mixed with cyclical movements, if any) from the following data
of output, by the method of moving averages:

Year
Quarter 2012 2013 2014 2015
I 29 40 47 45
II 37 42 51 49
III 43 55 63 60
IV 34 43 53 48