Chapter 17

TIME SERIES

17.1 Introduction

Arrangement of statistical data with respect to time is called time series. Generally, the

values of the various variables such as, series related to prices, consumption of commodities,

bank deposits and credits, population etc are observed with respect to time. Such data falls under

the study of time series analysis as the variable under consideration changes from time to time. A

time series reveals the relationship between the two variables where the independent variable is

time. The changes in the value of the variable with respect to time are called as fluctuations. Such

fluctuations are the result of interaction between various forces, like changes in the taste and

habits of the people, change in the prices of the commodities, weather condition, cost of

production etc.

The main objective of time series analysis is to isolate and measure the extent in which

the various components affect the time series. Such a study will help us to understand the past

behaviour of the time series and in the prediction of future trends. Some of the definitions of time

series are as follows 

“A time series is a sequence of values of the same variate corresponding to successive periods of

time.” W.Z.Hirsch.

“A time series is a sequence of repeated measurements of a variable made periodically through

time.”  C.H.Meyers.

Advantages and Disadvantages of Time Series

17.2 Advantages

1. It helps to understand the past behaviour and would be helpful for future predictions.
2. Statistical techniques have been developed so that the time series can be analysed in such a

way that the factor that influences the fluctuations of the series may be identified.

3. It helps to compare the performance of two different series of similar type for the same time

duration.

4. The analysis of time series helps us to compare the present performance of the series with

that of the past.

17.3 Disadvantages

1. The conclusion drawn from analysis of time series is not always perfect.

2. The various factors that affect the fluctuations of a series can not be fully adjusted by the time

series analysis.

3. The various factors that influence the time series may not remain same for an extended period

of time and so forecasting made on this basis may become unreliable.

4. Sometimes the increasing trend in the time series data may be due to the increase in

population. So, unless necessary modification is made to the data it would be difficult to

understand the trend.

Components of Time Series

If we look at a time series data or towards the graphical representation of it, we would see

how the series changes over time. If the time series is studied critically, it would reveal that the

changes are not haphazard but atleast a part of it can be explained properly. The variation in the

time series that can be accounted for is called the explained variation of the time series. The

explained variation results from any one, all or some combination of the following three broad

factors viz. (a) Secular trend (b) Seasonal variation and (c) Cyclical variation. However, the part

of the fluctuation in a time series that remains unexplained is called the irregular fluctuation. Let

us now discuss various components of time series one by one.

(a)Long Time Trend or Secular Trend

 The long time trend or secular trend is that component of the time series that looks the

smooth, regular and long time movement of a series if observed for a considerable period of time.

Thus, the general tendency of a time series to increase or decrease during a extensive period of

time is called secular trend. This is true for most of the series in economics and business statistics.

The growth or decline is due to some fundamental forces like changes in population, technology

or productivity. However, it is not necessary that the growth or decline continues to occur at each

and every time period, but we take into consideration the overall rise and fall in the time series.

Even though there may be some occasional isolated time periods in which there are small fall or

rise in the time series. Some examples of secular trend are given below 

An upward trend is noticed in time series related to population of a country, literacy rate,

agricultural production, volume of bank deposits etc. Similarly, a downward trend is noticed in

time series related to deaths due to epidemic, land for cultivation, infant mortality rate etc.

(b) Seasonal Variation

One rhythmic force that is found in most of the time series is called the seasonal fluctuation.

Seasonal variations are short time fluctuations that are observed in a time series data particularly

due to the climatic and weather conditions. Seasonal variation as the name indicates is the

influence that the different seasons exert on a time series data. However, such variations, also

may be due to the cultural or religious festivals that the people of a particular place celebrate

during a particular time of the year. This type of variation is periodic and causes regular

fluctuations in a time series with a time period of less than a year. For example, Sale of woolen

cloth increases during winter, raincoats during rainy season, sale of dress materials increases

during festival season like Durga puja, Diwali, Idd etc.

(c) Cyclical Variation

The oscillatory movements in the time series with a time period of more than a year is called

as the cyclical fluctuation. The word ‘cycles’ refers to the period of affluence and depression, ups

and downs, booms and slumps of a time series, most commonly seen in business cycles. These
cycles are not strictly periodic like that of the seasonal variation, but they generally occur in a

time period of 3 to 12 years depending on the nature of the time series. Such type of variation

does not have any definite trend. Some authors are of the opinion that seasonal variation is a

particular type of cyclical variation where the period of cycle is 1 year. But the seasonal

variations can be predicted much easily compared to cyclical variation because in the later case

the period of cycle is uncertain. Some examples of cyclical fluctuations are found in the world of

fashion, profits of business houses, demand for a particular brand of product etc.

(d) Irregular Variation

Irregular variation does not have any proper time, direction and magnitude of occurrence.

Irregular variation is also called random variation. This type of variation occurs in the time series,

if there is certain abnormal behaviour in the circumstances that affects the series. Irregular

variation is caused due to factors such as wars, earthquake, floods, strike etc that are

unpredictable in nature. The variation in time series that cannot be explained by secular or

seasonal or cyclical fluctuation is known as by irregular variation. It is difficult to isolate this type

of variation from time series as this type of fluctuation cannot be predicted. Examples: the

variation in the production of a business house due to strikes, change in population of country due

to epidemic, variation in the price of essential commodities due to war etc.

Different Methods of Measuring Trend

Moving Average Method

Moving average method is a simple device of reducing fluctuations and obtaining trend

values with a fair degree of accuracy. Let us consider a time series of period k. This period k is

called as the time interval of the moving average. In this method we start obtaining the simple

average of each of the k successive observations of the series. We start with the first k

observations and then we leave the first one and include the (k +1) th observation, so that for the

second case also we have k observation which are averaged. The process is continued still we
reach the last observation of the series. Each of these means is then centered against the time

which is the mid point of the time interval included in the computation of the moving average. If

the time interval of the moving average is odd we can center the average against a given time

value. But the problem arises if the time interval is even. In such a case the moving average falls

midway between two tabulated values. Then we calculate a subsequent two item moving average

values correspond to the tabulated time periods.

The interpretation of a moving average is simple. The procedure of taking averages for the k

values actually eliminates the existence of abnormalities (as far an average can do) during that

time period and estimates a value at the middle of the period. When all possible values are

obtained from the moving average then we get the over all movement of the series or the trend

eliminating the effect of all other components.

Advantages

1. The greatest advantage of this method is that it eliminates the short time fluctuation that may

be present in the time series.

2. This method is simple to understand and does not require the use of any complex

mathematical calculation.

3. If a few more values are added to the time series then it simply results into few more trend

values which can be easily obtained without disturbing the previous calculations.

4. This method is not subjective because the choice of the period of moving average is

determined by the oscillatory movements of the data and not by the whims of the statistician.

Disadvantages

1. The main disadvantage of this method is that it does not provide trend values for all the

terms. There are no trend values for some time periods at the beginning and at the end of the

series.

2. If the fluctuations in the time series are irregular then it is difficult to determine about the

period of moving average.

3. The method of moving average is developed under the assumption that the trend is linear.

Thus, for time series with linear trend (which is generally the case in economics and

business), this method either over estimates or under estimates the trend values.

4. This method cannot be used for predicting or forecasting, which is the main objective for

trend analysis because it does not put forward any mathematical relationship between the

variate under study and time.

Experiments
Experiment No. 17.1

Question: Using 3-year moving averages, determine the trend of the fo1lowing time series

Year Product ion Year Production

‘000 tons ‘000 tons

1968 21 1973 22

1969 22 1974 25

1970 23 1975 26

1971 25 1976 27

1972 24 1977 26

Plot the trend and actual values graphically.

Solution: For calculating the 3-year moving averages we construct the following table:

Year Production 3 Yearly 3 Yearly

Moving Total Moving
Average
1968 21 ----- -----
1969 22 66 22
1970 23 70 23.33
1971 25 72 24
1972 24 71 23.67
1973 22 71 23.67
1974 25 73 24.33
1975 26 78 26
1976 27 79 26.33
1977 26 ----- ------

To compute this table in Excel we should proceed with the following formulae:

A B C D
1 Year Production 3 Yearly 3 Yearly
Moving Total Moving
Average
2 1968 21 ----- -----
3 1969 22 =B2+B3+B4 =C3/3
4 1970 23 =B3+B4+B5 =C4/3
5 1971 25 =B4+B5+B6 =C5/3
6 1972 24 =B5+B6+B7 =C6/3
7 1973 22 =B6+B7+B8 =C7/3
8 1974 25 =B7+B8+B9 =C8/3
9 1975 26 =B8+B9+B10 =C9/3
10 1976 27 =B9+B10+B11 =C10/3
11 1977 26 ----- ------

For the purpose of plotting we type the Year, Production and 3 Yearly moving average values in

adjacent columns and then draw the scatter diagram. While drawing the scatter diagram we

30

25

20
Production

15

10 Actual
Trend
5

0
68

69

70

71

72

73

74

75

76

77
19

19

19

19

19

19

19

19

19

19

Year
activate the option . On plotting the trend values obtained through the method of moving

averages and the actual values we get the following graph.

Experiment No. 17.6

Question: From the data given below construct the 4- yearly moving averages.

Year: 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Prod(‘000 tons): 123 134 132 143 157 165 154 169 174 177 179 184 183

Solution: To construct the 4-yearly moving averages we construct the four yearly moving total

and then using the usual procedure calculate the four yearly moving average. But the averages

need to be centered, so a two yearly moving average is constructed in ordered to center them to

corresponding year. These 4-yearly moving averages are then called as the centered 4-yearly

moving averages.

Year Prod.(‘000 tons) 4 Yearly 4 Yearly 4 Yearly

Moving Moving Moving
Total Average Average
(Centered)
1988 123

1989 134
532 133
1990 132 137.25
566 141.5
1991 143 145.38
597 149.25
1992 157 152.00
619 154.75
1993 165 158.00
645 161.25
1994 154 163.38
662 165.5
1995 169 167.00
674 168.5
1996 174 171.63
699 174.75
1997 177 176.63
714 178.5
1998 179 179.63
723 180.75
1999 184

2000 183

1. For the following comment on (state the reasons) the type of fluctuation:

(i) An era of prosperity

(ii) Sale of rain coats

(iii) Sale of fish

(iv) Traffic density in the main road of a town

(v) Price of vegetables

(vi) Population of a country

(vii) Population of a flood effected area

(viii) Number of causalities in a hospital after a fire

(ix) Production in an industry during a Bharat bandh

(x) Yearly production in a firm.

Numerical Problems

1. Find out the short-term fluctuations of the following time series, assuming a five-yearly cycle:

Year 1980 1981 1982 1983 1984 1985 1986 1987 1988

Annual values 239 242 236 252 257 250 273 270 268
Year 1989 1990 1991 1992 1993 1994 1995 1996 1997

Annual values 278 284 282 300 303 298 313 317 309

Year 1998 1999 2000 2001 2002 2003 2004

Annual values 329 333 327 345 344 343 362

Plot the trend values and the actual values graphically.

2. Calculate four yearly moving average from the following data

Year Production of Rice Year Production of Rice

in '000 tons in '000 tons

1990 100 1997 110

1991 95 1998 104

1992 105 1999 112

1993 90 2000 100

1994 94 2001 120

1995 102 2002 92

1996 108

3. Following figures relate to output of cloth in a factory (output in lakhs of metres)

Year Output Year Output Year Output

1977 72 1986 68 1995 84

1978 68 1987 68 1996 96

1979 64 1988 76 1997 92

1980 60 1989 80 1998 96

1981 68 1990 84 1999 100

1982 72 1991 80 2000 92

1983 72 1992 88 2001 96

1984 76 1993 88 2002 100

1985 72 1994 92 2003 84

Calculate four-yearly moving average.

4. Calculate 7-yearly moving average for the following data on number of commercial and

industrial failures in a country during 1985-2000

Year No. of failures Year No. of failures

1985 23 1993 9

1986 26 1994 13

1987 28 1995 11

1988 32 1996 14

1989 20 1997 12

1990 12 1998 9

1991 12 1999 3

1992 10 2000 1

@@@@@@@@@@@@