ANALYSIS OF TIME SERIES Seer ANALYSIS OF TIME SERIES ; COMPONENTS OF TIME SERIES at work, affecting the values of a phenomenon in a time series, can be ommionly known as the components of @ cries) in varying degrees The various forces broadly classified into the following foutr categories time series, some or all of which are present (in a given Lime (a) Secular Trend or Long-term Movement. (6) Periodic Changes or Short-term Fluctuations, @) Seasonal variations, and (ii) Cyclic variations. (©) Random or Irregular Movements, The value of a time series y, at any time ¢ is regarded as the resultant of the combined impact of above components. In the following section we shall briefly explain them one by one. Remarhss 1.Not all economists agree as to the classification of the components used here or as to the manner in which they are related, Some argue that there are more than four components, and some think that trend and cyclical movements are produced by the same set of forces, 7 During the past two or three decades, the attempt to make forecasting mire scientific has led to the development.of altornate. approaches such as econometrics, Eeonomeriee atfempts to’express ‘{Sgromic theories in mathematical models that can be tested and verified by statistical methore it someasure the'impact of one economic variable upon anethér in the hope of being able to predict future events. A . 3. The approach in this chapter is to present: the classical statistical approach to time-series Gralysis, but at the,siime time to point out that may other possible models emt hick are based.on different-assumptions and which may lead to different results, -_z2L-Trend. By secular trend or simply trend we inean the general tendency of the data taincrease ordectedse during a long period of time. This is true of most of series of Business and Economic Statistics. For example, an upwapd tendéncy would be seen in data pertaining to population, agricultural production, currenicy in circulation etc, while ¢ downward tendency will be noticed in data of births and deaths, epidemics ete., as a result of advancement in medical sciences, better medical facilities, literacy and higher standard of living. 7 : 1. It may be clearly noted that trend is the general, smooth, long-term, average tendency. It is not necessary that the ineréase or decline should be in the same direction throughout the given period. It may be possible that different tendencies of increase, decrease or stability are observed in different sections of time. However, the overall tendency may be-upward, downward or stable. Such tendencies are the result of the forces which are, more or less, constant for a long time or which change very gradually and continuously over a long period of time such a5 the change in the population, tastes, habits and custome of the people in a society and so on. They operate in an evolutionary manner and do not reflect’ sudden changes. For example;'the effect of population inerease over a long period of time on the expansion of various ‘sectors like agriculture, industry, education, textiles, etc., isa continuous but a gradual process. Similarly, the growth or decline in a number if economic time serios is the interaction of forces like advances in production technology, large-scale production, improved marketing management and business organization, the inventeey and discovery of new natural resources and thie exhaustion of the ‘existing resources and so on-> all of which are gradual processes. ° 2. It should not be inferred that all the se We might come across certain series whose voak FUNDAMENTALS, OF APPLIED STATISTICS does not change with time, eg., the series of barometric readings or Uhe temperature of a particular place 3. Linear and Non-linear (Curvi-tinear) Trend. If the time series values plotted on graph cluster more, or less, round a straight line, then the trend exhibited by the time series is incar otherwise Non-livear (curvi-linear),/ In a straight line trend, the time-series values increase or decrease more or less by a constant absolute amount,)i.c., the rate of growth (or decline) is constant, Although, in practice, linear trend is commonly used, it is rarely obtained in economic and business data, In-an economic and business phenomenon, the rate of growth or decline is not of constant nature throughout but varies considerably in different sectors of time. Usually, in the beginning the growth is slow, then rapid-which is further accelerated for quite sometime, after which it becomes stationary or stable for some period and finally retards slowly. 4, The term ‘long period of time’ js a relative term and cannot be defined exactly. In some cases a périod as small as a week may be fairly long while in some cases, a period as long as 2 years may not be enough. For example, if the data of agricultural production for 24 months shows an increase it won't be termed as secular change over a period of 2 years whereas if the count of bacterial population of a culture every five minutes, for-a week. shows an increase, then we would regard it as a seciilar change. “2.2:2. Periodié Changes:-It-would.be observed that in many social and’ economic phenomena, apart from the growth factor iii'a time seriés there are forces at work which prevent the smooth flow of the series.in a particular direction and tend to re over a period of time: These forces do not act continuously bu op: a manner, The resultant effect of such forces inay be classified as : (1) Seasonal variations, ~ and (2).Cyclic variations.—— (1) Seasonat Variations. (These variations in a:time series are due to the thythmie forces which operate in a regular and periodic manner over a span of less than a year) ie, during.a period of 12 months ‘and have the same or almost same pattern year after yeir. Thus’ seasonal variations in a’ time series will be there if the data are recorded quarterly (evety three months), monthly, weekly, daily, hourly, and so on. Although in each of the above cases, the amplitudés of the seasonal variations are different, all of them have the same period viz, 1 year. Thus, in a time series data where only annual figures are given there are rio seasonal variations. Most of economic time series are influenced by eeenecl swings, ¢.g., prices, production and consumption. of commodities;. sales and profits inva departmental store; bank clearings and bank deposits, etc., are all affected by seasonal variations. The seasonal variatioris may be attributed to.the. following two causes (i Those resulting from siatural forces: As the name suggests, the vatious seasons or weather ec ns and climatic chianges play an important role in seasonal movements. For instance, the sale of umbrellas pick tip very fast in rainy season; the demand for electric ‘i goes up in summer season; the sale of ee and iceream increases very much in eases sales of woollens go up in winter—all being affected by natural forces, viz; an he seasons.-Likewise, the production of certain commodities such as sugar, rice ptTen tte, depends on seasons. Similarly, the prices of agricultural commodities ulate eee the time of harvest and then pick up gradually. rn ee down at (i) Those resulting from man-made conientions, These variat a period of 12 months are due to habits, fashions the society. For instance, the sale of jewell ? tt themselves erate in 4 regular spasmodic ; edi, aeANALYSIS OF TIME SERIES 2.5 and profits jn departmental stores go up considerably during marriages, and festivals like Diwali, Dussehra (Durga Pooja), Christmas, ete. Such variations operate in a regular - spasmodic manner and recur year after year i ‘Thelmain objective of the measurement of seasonal variations|is to isolate them from the trend and study their effects. A study of the seasonal patterns is extremely useful to businessmen, producers, sales managers, etc., in planning future operations and in formulation of policy decisions regarding purchase, production, inventory control, personnel requirements, selling and advertising programmes. In the absence of any knowledge of seasonal variations, a seasonal upswing may be mistaken as indicator of better business conditions while a seasonal slump may be mis-interpreted as deteriorating busin: conditions, Thus, to understand the behaviour of the phenomenon in a time series properly, the time series data must be adjusted for seasonal variations. (This is done by isolating them from trend and other components by dividing the given time series values (y,) by the seasonal variations (S,). This technique is called de-seasonalisation of datajand is discussed in detail latér (See § 25-4] . (2) Cyclie Variations. The oscillatory movements in a time series with period of oscillation more than one year are termed as cyclic fluctuations. One complete period is called a ‘cycle’. The cyclic movements in’a-time.series are.generally attributed to the-so-called ‘Business: Cyéle’,"which- may also, be referred to as the ‘four-phase. cycle’ composed. of prosperity (period.of boom), recession, depression and recovery, and. normally lasts.from | seven to eleven years. The upswings and downswings in business depend upon the cumulative natufe of the, economic forces (affecting the equilibridm-of demand ‘and supply) and the interaction’ bétween them. Most of the economic’and: cornmercial series, e.g.; series relating to prices, production and wages, etc., are affected by business. cycles. Cyclic fluctuations, though mote or less regular, are not periodic. os 2.2:3, Irregular’ (or Random) Component. Apart from the regular variations, almost all the series contain another factor called the random or irregular or residual fluctuations, which are not accounted for by secular trend and seasonal and cyclic variations. These fluctuations are purely random, erratic, unforeseen, unpredictable and are due to numerous non-recurring and irregular circumstances which are beyond the control of human hand but at the same time are a part of our system such as earthquakes; wars, floods, famines, revolutions, epidemics, etc. These isolated or irregular but powerful fluctuations due to floods, revolution, political upheavals, famines, ete., are also called ¢pisodic fluctuations. In Some cases the importance -of irregular. fluctuations may not be significant while in others these may be very effective and might give rise to cyclic movements. Remark. It may be ‘noted that because of their absolutely random character, it is not possible to isolate irregular variations and study them exclusively, nor one can forecast or estimate them precisely. Only rough:estimates from past experience can be obtained and accordingly one may make some provisions for such abnormalities. 2.3, ANALYSIS OF TIME SERIES ‘The main problems iri the analysis of time series are : (i) To identify the forces or components at work, the net effect of whose interaction is exhibited by the movement of a time series, and : (ii) To isolate, study, analyse and measure them independently, i.e, by holding. other ‘ things constant.ce 26 FUNDAMENTALS OF APPLIED STATIST Ics £31. Mathematical Models for Time Series. The following are the two models commonly used for the decomposition of a time series into its components. Decomposition by Additive Hypothesis (or Additive Model). According to the additive model, a time series can be expressed as Yea Te + Set Cyt Ry (2-1) where y; is the time-series value. at time ¢, 7, represents the trend value, S;, C, and R, represent the seasonal, cyclic and random fluctuations at time ¢. Obviously, the term S; will not appear in.a series of annual data("The additive model implicitly implies that seasonal ' forces (in different years), ical forces (in different cycles) and irregular forces (in different long term period) operate with equal absolute effect irrespective of the trend value) As such C, (and S)) will have positive or negative values, according as whether we are in aif ¢ above normal or below normal phase of the cycle (and year) and the total of positive and negative values.for any cycle ( and any year) will be zero. R, will also have positive or negative value and in the long-term (5, R,) will be zero. Occasionally, there may be a few ( isolated occurrences of extreme R, of episodic nature : The additive model assumes that all the four components of the time series operate independently of each other so that none of these components has any effect on the. « remaining three. ee : Decomposition by Multiplicative Hypothesis (or Multiplicative Model). On the other hand,- if we have reasons to assume that the various coripdiients in’ a’ time-series € | operate proportionately to the general level of the series,.the traditional or classical multiplicative model is appropriate. According to the multiplicative model, = TeX Si XC)x Ry «22 & where S,, C, and R;, instead of a8suming positive ‘and negative value, are indices fluctuating above or below unity and the geometric means of S, in a year, Cy in a cycle and R, in a long- term period aré unity) In a time series with both positive end negative values, the € multiplicative model can riot be applied unless the time series is translated by adding a suitable positive value. It may be pointed out that the multiplicative decomposition of a time séties ig same as the additive decomposition of logarithmic values of the original time series, © ie, log y -= log T; + log 8; + log C, + log R, 4 In-practice; most of the series relating td-economic data conform to multiplicative model Remarks 1. Limitations of the Hypothesis of Decomposition: of a’ Time: Seriés: Hypothesis 6t decomposition prestipposes that the trend and periodic components are determined by separate forces € geting independently so that simple aggregation of the components could constitute the series. But in reality, it is possible that this year's value of the series will depend to some extent on last year’s value ‘ fo that trend and. periodic movement will get inextricably mixed up and no meaningful separation of J them will be possible, In such a case any variations of this year may affect the whole future course of € fe the series and no meaningful separation of trend and periodic components will be possible. \ 2, Mixed Models. In addition to the additive and-multiplicative models discussed above, the X bmponents in a time series may be combined in a large number af other ways, The different models, defined under different assumptions will yield different results. Some of the mixed models resulting from different combinations of additive and multiplicative models are given below : H=TC+S,R, € aT +S, CR, «-- (2-2a) maT +842,* ANALYSIS OF TIME SERIES 27 3. The model (2:1) or (2-2) can be used to obtain a measure of one or more of the components by climination, viz., subtraction or division. For example, if trend component (7)) is known, then using multiplication model, it can be isolated from the given time series Lo give y% — Original values _ Six CXR, = fF = “Trend values (2-26) ‘Thus, for the annual data, for which the seasonal component S, is not there, we have yet xCxR, = OxR =F .. (2:20) 2-3-2. Uses of Time Series. The time series analysis is of greater importance not only to businessman or an economist but also to people working in various disciplines in natural, social and physical sciences. Some of its uses are enumerated below : 1. It enables us to study the past behaviour of the phenomenon under consideration, i.e., to determine the type and nature of the variations in the data 2. The segregiition and study of the various components is of paramount importance to a businessman in the planning of fature operations and in the formulation of executive and policy decisions. 3. It helps to compare the actual current performance of accomplishments with the expected ones (on the basis of the past performances) -sad.analyse,the,causes of such variations, ifany., 4, Te enables iis tpFéidict or estimate or forcast the behaviour of the phenomena. in: future which is very essential for business planning, 5. [thelps us to comparé the changes in the values of different phenomenon at different times or places, ete. In the following sections we shall discuss various techniques for the measurement of different, components (C24. MEASUREMENT OF TREND ‘Trend can be studied and/or measured by the following'methods : (i) Graphic (or Free-hand Curve Fitting) Method, (ii) Method of Semi-Averages, (iii) “ Method of Curve Fitting by Principle of Least Squares, and (iv) Method of Moving Averages. We’shall now discuss each of these methods in detail. 2-4-1. Graphic Method. A free-hand smooth. curve obtained on platting the values inst ‘t’ enables us to form an idea about the general ‘trend’ of the series. S1 ‘oothing of of the curve eliminates other components; viz. regular and irregular fluctuations. (This ‘method does not involve any complex mathematical techniques and can be used to describe ‘all types of trend, linear and non-linear. Thus,. Simplicity and flexibility are strong points of this method] Its main di very important role and as such different t end curves will be obtained by different persons ‘for thé same set . As such “trend by inspection’ should be attempted only by skilled and experienced ‘icians and this ity and popularity of the method. (ii) It-does not enable us to mea _7 2-4-2. Method of Semi-averages. In this method, the whole data is divided into, two parts with respect to time, e.g., if we ate given y, for ¢ from 1991-2002, ie., over a period of 12 years, the two equal parts will be the data from 1991 to 1936 and 1997 to 2002. In case of odd“2:8 NOAMENTALS OF APPLIED STATISTICS FUNDAMENTALS OF API number of years the two parts are oblained by omitting the value corresponding to the middle year, eg, for the data from 1991-2001, the value corresponding to middle year, viz 1996 being milled. Next we compute Lhe arithmetic mean for each part and plot these two averages (means) against the mid-values of the respective time-periods covered by each part. ‘The line obtained on joining these two points is the required trend line and(may be extended both ways to estimate intermediate or future values.) Remark. Por ever number of years like 8, 12, 1G, ete. the centering of average of each part would create problems, e.g., from the data 1997-2002 (n = 12), let the two averages beX,, (say) for period 1991-1996 and X; (say), for the period 1991-2002, Here X,, will be plotted against the mean of two mid. values, viz. 1993 and 1994 for the period 1991-1996, i.e., against Ist July 1993. Similarly, for the period 1997-2902. Merits 1. As compared with graphic method, the obvious advantage of this method is its objectivity in the sense that everyone who applies it would get the same results. Moreover, we can.also estimate the trend values. ed 2. It is readily comprehensible as compared to the ‘method of least squares’ or the ‘moving’average method’. : Limitations. This method assumes linear _relationship between the_plotted points — which thay not exist. Moreover, the limitations of arithmetic mean’as an average also stand in its way. _ e Example 2-1) Fit a trend line to the.following data by the method of semi-averages : Bank Clearances (Rs. Crores) Year __ Bank Clearances (Rs. Crores). 53 (1999 87 79 2000" 79 16 2001 104 66 2002 97 69 2003 92 94 “| leoo4 JO1 105 Solution. Here since n = 13 (odd), the two parts would consist of 1992 to 1997 and 1999 to 2004, the year 1998_being omitted. X, = Average sales for first part 120 7 : d - 1s a = 72-83 (Rs. crores) 110 ACTUALDATA X= Avérage sales for second part = 22 - 9989 (Rs. crores) ‘As explained in Remark to § 2-4: X, and X, will be plotted against 1st July 1994 and Ist July 2001 __ tespectively, as given in Fig. 2-1 Joining the points A (1994, X,] and B (2001, Xzl, we get the trend line (Fig. 2-1]. Fig. 2.1: Trend by the Method of Semi-averages SALES (Rs. CRORES) 88 8.8ANALYSIS OF TIME SE 29 ie Matto of Curve Filling by Principle of Least Squares. The principle of : E the most popular and widely used method of fitting mathematical functions toa given The method yiel results if sufficiently good appraisal of the form of the function to the filled is obtained either by a serutiny of the graphical plot of the values over time or by a theoretical understanding of the mechanism of the variable change. An examination of the plotted data often provides an adequate basis for deciding upon the type of trend to use. Apart from the usual arithmetic scales, semi-logarithmic or doubly-logarithmie scales may be used for the graphical representatigh of the data. The various types of curves that may be used to describe the given data in practice are : (I is the value of the variable corresponding to time £) (i) Astraightline: ¢ — y =a+bt (ii) Second degree parabola.; y,=a + bt.+.cl? (iii) kth-degree polynomial Y= Ay + aye + al? +... + ag (iv) Exponential curves neadl? = “logy, = loga +t log b =A + Bi, (say). (v) Second degree curve fitted to logarithms : Mm sablet. : e L => logy = loga +t log b +t loge =A +B t.+.C 22, (say). (vi) Growth curves : ‘ (a) yi =asbe "(Modified Exponeittial Curve) (o) ye sabe (Gompertz curve) = logy, =loga+e! logb=A + Be! (say) h “T+ exp(a +00) : ding about the type of trend_t6 be fitted to a given set of data, the following (c) y (Logistic curve) Remark. For d points may be helpful : (® When the time series is found to be(increasing or decreasing by equal absolute amounts, the straight line trend is,used) In this case, the plotting ofthe data will give a straight line graph. Gi, The/logarithmic straight line(exponential curve y, = ab‘) is used as an eipression of the secular movement,(when the series is increasing or decreasing by a constant percentaghrather than a constant j absolute amount. In this case, the data’ plotted ona semi-logarithmeic scale wll give a straight line i graph, | (i) Second degree curve fitted to logarithms may be tried for-trend. fitting ifthe data plotted on a semi-logarithmic scale is not a straight Irie graph but shows curvat ing-concave_ either upward or downward, / Alternatively, approximations about the type of the curve to be fitted can be made by use of the following theorem based on finite differences : ‘The nth differences d*y,, A* log y,), A" (ily) of any general Polynomial y; of nth degree in tis constaiit a and (n + Ith differences are equal to zero.” \ * For further guidelines, the following statistical tests based on the leaks of finite differences may be applied. We know that for a polynomial y, of nth degree.in . “ A’y, =constant, “r=n ). =0, ron210) ° : : +. FUNDAMENTALS OF APPLIED STATISTICS: where A isthe difference operator given by A.y,=y,,4~y; yh being the interval of differencing and 4°y, is the rth differences of yy. 1. [fay = constant, us Lraight line Grend 2. 1f.a%y; = constant, use a second dl 3, ICA log y,) = constant, use exponen LAC a (log.y) = constant, use second degree curve fitted to logarithms. “LB. The growth curves, viz, modified exponential, Gompertz and Logistic curves ean be approximated by the constancy ofthe ratios : 4% 4 (logy) Aye * [A llog y=) respectively, for all possible values of ¢ The following tests may also be used G.IfAy, tends to decrease by a constant percentage, use modified exponential curve. 7 if y, resembles a skewed frequency curve, use a Gompertz curve or Lagistic.curve. _Fitting of Straight Line by Least Squares Method. Let thé straight line trend between the given time-series values (y;) and time t be given by the equation Maat bt ++ (23) Principle.of least squares consists in minimizing the sum of squares of the deviations between the given values of y, and their estimates given by (2-3). In other words, we have to find a and 6 such‘that for’ given values of y, cortésponding to n different values of ¢, EEO -o- doe Leeae | is minimum. For.a maxima or minima of £, for variations in a and 6, we should have. aE oF 0 250-01) la M4 Ly =na + bEt =-r 2D (abe) Dae aderorel (28) which are the normal equations for estimating a -and b The values of E yi, Et, E are obtained from the given.data and the-eqiiations (2:4) can now be solved for a and b. With'these values ofa and 6, the line (23) gives the desired tread ° line: Remark. The solution of normal equations (24) provides a mii of 8. The proof is given below ‘The necessary and sufficient condition fo a'minima of E for variations ina ahd b are : : ve 8B aE ab 7 ,_|aa? daab xs O BAO |eAO CO and GA=|"yp “ep |>9 and FS; 9 4) abda ab From (2-4), we get as aS PE par 20> 0 5 Sp BE > 0.5 555 = Spay =2Ee vq | 2n- 206 Slo ase [=4hee-col) i. (2 ]oaeverse Hence, the solution ofthe least square equations (24), satisfies (¥) ang ey a minima of E. and, therefore, provides oe éee ANALYSIS OF TIME SERIES 244 “iting of Second Degree (Parabolic) Trend. Let the second degree parabolic trend curve be yy =a + bt + ct? Proceeding similarly as in the case of a straight line, the norn a, band are given by Yop ena OLE ele? Vtyy salt +b +cLe pe ws (2-6) Ley alle bE rode! the summation being taken over the values of the time series \_ Mitting of Exponential Curve: ye aad! (2D = logy, =loga + ¢ log b = ¥ =A¥Bt(say), aa .. (2a) where Yelogy, , A =loga , B=logb. (2-76) (2-7a) is a straight line in ¢ and Yand thus the normal equations for estimating A and B are LY =nA +BLe, }.-- : : DtY =ALt+ BY? } ‘These equations can be solved for A’ and B and finally on using (2°76), we get é a =antilog (A) ; 6 = antilog (B). Second Degree Curve Fitted to Logarithms. Suppose the trend curve is : Y, =abtef w= (28) Taking logarithms of both sides, we get logy, =log a +t log 6 + (? loge = Y¥, =A+Btse? ee (28a) where ¥, =logy, { A=loga ; B=logb and C=loge ..- (2:86) Now, (2:82) is a second degree parabolic curvé in Y, and ¢ and can be fitted by the technique already explained. We can finally obtain a = Antilog (A) { 6 = Antilog(BY “and: c = Antilog (C). With these values of a; 6 and ¢, the curve (2-8) becomes the best-second degree curve fitted to-logarithms. Remark, The method of curve fitting by the principle of least squares is used analysis particularly when one is interested in making projections for future time reliability of the estimated (projected) values primarily depends upon the appropriatness of the form of the mathematical function fitted to the given data. If the function is determined on the ad-hoc basis by the scrutiny of the plotted values, the projections based on it may be valid for the near future while, if the study of physical mechanism of the variable change forms the basis of the seléction of function, then there is very little likelihood that the function will change for sufficiently long period and hence in this case reliable long term projections can be made. ws (2-Te) Merits and Drawbacks of Trend Fitting by the Principle of Least Squares. Merits. The method of least squares is the most popular and widely used method of fitting mathematical functions toa given set of observations. It has the following advantages :242 FUNDAMENTALS OF APPLIED STATISTICS al or analytical character, this method completely eliminates adjective judgement or personal bias on the part of the investigator 2. Unlike the method of moving averages [discussed in § 2-4-5}, this method enables us to compute the trend values for all the given time periods in the seri the clement of ies. 3. The trend equation can be used to period values stimate or predict the values of the variable for any in future or even in the intermediate periods of the given series and the forecast re also quite reliable. 4. The curve fitting by the principle of least squares is the only technique which enables us to obtain the rate of growth per annum, for yearly data, if linear trend is fitted. |Drawbacks |. The method is quite tedious and time-consuming as compared with other methods. Itis rather difficult for a non-mathematical person (layman) to understand and use. 2. The addition of even a single new observation necessitates all calculations to be done afresh. é 3. Future predictions or forecasts based on this method are based only on the long term variation, ie., trend and completely ignore the cyclical, seasonal and irregular fluctuations. ‘ious limitation of this method. is the determination of the type of the trend curve to bé fitted; viz., whether we should fit a linear or-a parabolic trend or some other re complicated trénd curve. 7 5, It cannot be used to fit growth curves like Modified Exponential curve, Gompertz curve and:Logistic curve, to which most of the-economic and business time series data conform. Example 2-2. Ina certain indisstry,{ Year ©) Production |" Year. > Production | the production of a certain commodity (000 units) (000-nits) | ‘(in"000" units) during the years 1994— 2000 932 | 2004 is given in the adjoining table : 2001 111-6 _ @ Graph the data, 2002, 88-3 (i) Obtain the least square line 2003 117-0 fitting the data and construct the graph of the trend line. (iii) Compute the trend values for the year 1994-2004 and estimate the production-of commodity during the years 2005 and 2006, if the present trend continues, (iv) Eliminate the trend. Solution. Here n = 11, is 2004 8 FISD ., odd and, therefore, we shift the origin to the middle time period, viz., the year 1999. Let x=t~1999 - a) TABLE 2-1 : COMPUTATION OF TREND LINE Year “Production : 2 Trend values () (000 units) (y,) (000 units) (t,) ' Ye= 95-49 + 3-95x | 1994 666 -5 -3330 25 15-14 1995 84-9 -4 = 339-6 ‘16 79-69 1996 © 88-6 -3 — 265-8 9 83-64 1997 78-0 ~2 - 156-0 4 87-59ANALYSIS OF TIME SERIES 213 respéctively in (3) and have been tabulated in the Last column of the Table 2- for 2005 Taking <=2005 in (1), we get 1998 968 -1 - 96:8 1 9154 1999 102 0 0 0 2000 93.2 1 93-2 oars 99-44 2001 16 2 223-2 4 10339 2002 883 3 264-9 9 107-34 2008 uro 4 468.0 16 11129 2004 5-2 5 516-0 25 115-24 Total 1,050-4 0 434-1 10 Let the least squareline of y, onxbe: yz = @ + bx (origin : July 1999) (2) ‘The normal, equations for estimating a and b are a+b Lx and Dry, ale +b x7 = = 434-1 = 1106 Z . 434-1 be ZA: = : 7 b= T10 = 3.95 7 ent _- Herice; the leat squaregline, fitting the data is = -y,= 95-49 + 3'95x, 7 (3) where origin is July 1999 and x unit = 1 year. ‘Trend values. foi'the veditS. 1994 to 2004 are obtained on putting x = —5,~4,~3, +2 2005 — 1999 = 6 Hence the estimate production of the commodity for 2005 is obtained on putting x = 6 in (***) and is given by: (e200 95.49 + 3.95 x 6 = 119.19 (000 units). Similarly, (32) 2006 = 95°49 + 9-95 x7 = 123-14 ('000 units) 2 126 16} oon 106 96 86 76 Trend ine + Production (000 units) ej ¢ The graph of the original data and the trend line is given “in Fig: 22 : Assurhing multiplicative model, the trend values are climinated on dividing the given values (y,) by the corresponding trend values (y,): However, if we assume the additive model, the trend eliminated values are given by (y, -y,). The resulting Furr ii iii values contain short-term Se85 8888888 DioMery BSS SS 8588 8 and irregular variations. Trend Years + alues are given in Fig. 2:2. Table 2-2.2-14: + FUNDAMENTALS OF APPLIED STATISTI TABLE 2.2: ELIMINATION OF TREND Trend Bliminated Values Ba: a Additive Model Multiplicative Model Oey) 1994 66-6/15-74 = 0-880 1995 84-9/79-69 = 1-065, 1996 1-059 0-891 9 1-057 1999 1-154 2000 93-2- 9944= —6.24 0-937 2001 M16 ~ 10339- B21 1.079 2002 88-3 - 107-34 = - 19-04 0-823 2003 117-0- 111-29= 5-71 1-051 2004 | 115-2- 115-24 =. 0-04 0-999 Example@-3) Fit a straight line trend by the method of least squaies to the following data relating to the Sales of a léading departmental store..Assuming tht the same rate of change continues, what would:be predicted earnings for the year 2006? fe Year : 71997 1998 1999 2000. 2001. 2002 ‘2003 2004 Sales (Crorés Rs3) 176-80 130144 138 120174190 Solution. Here n’= 8, é.e., even. Hence we shift the origin to the arithmetic mean of the two middle years,.viz,, 2000 and 2001. We define ‘ £=7(2000 + 2001) 1000-5 _ 7 (Interval) Pa where x values are in units of six months (half year). TABLE 2-3 : COMPUTATION OF LINEAR TREND 2-4on. = 2Ct -20090 Year | Sales (Crores Rs.)[ zy, |e ge. | Trend values (Crores Rs.) @ Me 1997} 16: eT fo BBD 49 1998. 80 -5 400 5 ig9g 130 -3 390 9 2000 144 -1 | -144 1 2001 138 1 138 1 2002 120 3 360 9 2003 114 5 870 25 2004- 190 1 1330 49 Total Ly= 1052 Zx=0| Leys | Dx2= 168 1,232 > —l__ Let the linear trend equation between y, and x be : Ye =a + bx, x= At — 2000-5)ANALYSIS OF TIME-SERIES - : | 2-45 Se Since Y x = 0, the normal equations for estimating a and b are.: i | : Dy, 10520 xy 1232 aoa = 1315, bay = T68 = 7:33 Hence the least square trend line becomes: —y, = 131-5 + 7-33x ... (3) where b = 7-33 units represent half yearly increase in the earnings. The trend values for the year 1997 to 2004 can now be obtained from (3) on putting ite =~-7,-5,...,5, 7 respectively, as shown in the last column of the above Table 2-3. Estimate for 2006 : When t = 2006, we get from (1), x = 2(2006 -2000-5) = 11 Hence the predicted sales for 2006 are: —_y, = 131-5 + 7-33 x 11 = 212-13 (Crores Rs.)tg a,g60 | 20. 2481 | &5MEASUREMENT OF SEASONAL -VARIATIONS It has already been pointed out that one of the types of fluctuations found in time series data is the seasonal component. Many economic and business series have distinct seasonal pattetns that are pronounced enough to predict future behaviour of the series. The objectives for studying seasonal patterns in a time series are necessitated by the following reasons : (i)-To isolate the. seasonal. variations, i.e., to determine the effect of seasonal swings on the value.of the given phenomenon, and 7 (ii) To-eliminate.them, i.e.,.to. determine the value of the phenomenon if there were no seasonal ups and downs in the series. This is known as de-seasonalising the given. dataland is necessary for the study of cyclic variations. . The determination of seasonal effects is of paramount importance in planning (i) business efficiency, or (ii) a production programme. For example, the head of a departmental store would be interested to study the variations in the demands of different articles for different months in order to plan his future stocks to cater to the public demands due to seasonal swings. Moreover, the isolation and elimination of seasonal factor from the data is necessary to study the effect of cycles. Obviously, for the study of seasonal variations, the data must be given for ‘parts’ of year, viz., monthly ©: ; rly, weekly, daily or hourly. _ Different methods for, measuring seasonal variations are discussed below. } 2-5-1. Method. of Simple Averages. This is thé simplest method of measuring seasonal variations in a time series and involves the following steps :AyALYSIS OF TIME SERIES i. ° id Method. ‘This method is an improvement over the simple onal variation for any given month sagonal variation by this method sumption that s “verages method and is based on the is constant factor of the trend, ‘The measurement of s consists of the following steps 1d values by the principle of least squares by fitting an appropriate ‘aight lino, 2nd degree parabolic curve or exponential curve, etc.) fata as the percentage of the trend values. Assuming the erefore contain the seasonal, cyclic and () Compute the tr mathematical curve (s (ii) Express the original € multiplicative model, these percentages will, Uy irregular components Gi) The cyclic and irregular components are then wiped out by averaging the catages for different months (quarters) ifthe data are monthly (quarterly), thus leaving ithmetic mean or median can be used for er'gives undue us with indices of seasonal variations. Bither ar averaging, but median is preferred to arithmetic mean since the latt weightage to extreme values which are not primarily due to seasonal swings. If there are few abnormal values; modified mean (which consists of calculating arithmetic mean after dropping out the extreme or abnormal values) may be used with advantage- 7 (iy) Finally, these indices, obtained in step, (tii), .are adjusted to 4 tétal“of 1200 for snonthly data or 400 for giudrterly data by multiplying.theim. throughout by a constant k given by : ners : : a 1,200 ake 2490" : = Total of the indices “°° “Total of the indices for monthly-and quarterly data respectively. Merits and Demerils: Since this method attempts at ironing out the cyclical or irregular components by the process of averaging, the purpose will be accomplished only if the cyclical variations are known to be absent or they are not so pronounced even if present. On the other - hand, ifthe series exhibits pronounced cyclical swings, the trend values obtained by the least’ square method can never follow the actual data as closely as 12-month moving average-and as such the seasonal indices obtained by ‘ratio to trend’ method are liable to be more biased than those obtained by ‘ratio to moving average’ method discussed in § 2-5-3) he obvious advantage of this method over the moving average method lies in the fact that ‘ratio to trend’ can be obtained for each month for which the data are available and as such, unlike the “ratio to moving average’ iéthod, there is no loss of data) k Remark. The calculations are simplified to,a great extent by first fitting a trend equation to the yearly totals (or averages) and then obtaining the monthly (or quarterly) trend values by a suitable ‘médification of the trend equation, as illustrated in the following example : Year | [Qrt. IQrt. UT Qrt. IV Qrt. \ 1995; 30 40-36 34 Example 2-16. Using Ratio to Trend a i method: determine the quarterly seasorial | 1996 | 34 Pare aOe) “4 indices for the adjoining data : | 1997,| 40 58 54 48 |. 1998. 54 76: 68 62 1999 | 80 92... 86 82 ee2.44 FUNDAMENTALS OF APPLIED STATISTICS 0 TABLE 2.18: COMPUTATION OF LINEAR TRE Potal of of id_| First of all, we will determine Year | quarterly | quarterly © xt | gy | ualues| the trend values’ for the wo values | values (9) | 4 1997 (Million| quarterly averages by Gtting (Million Rs.) Rs.) | a linear trend by the method os] Mo |S 70) a | of least squares (cf, Remark sat i 7 i ove, §25°2) 1996 180 45, -1 1 | -45| 44 | Let the straight line trend equation be: 1997 200 50 0 0 0] 56 ysatbr (*) (Origin : 1997, x unit : 1 year) | ad o L 1 65} 68 |y : Quarterly averages (Million Rs.)} | 1999 | 340 85 2 4 | 170} 80 | Total Yy=280 | Yr=0 | Sx? | Dry i =10| 120 The normal equations for estimating a and b.in (*) are: Lysna+bEz Lay Sah + bE xt } Hence, the straight-line trend is given by the equation : ye=56+12x - [origin : 1997, x unit = 1'year] we) Putting x = -2, -1, 0, 1, 2, we obtain the average. quarterly trend values for the years 1995 to 1999 respectively, which are given in the last column of the Table 2-18. From the trend equation (*), we observe'that : Yearly increment in trend valies =6=12 => Quarterly increment =2 ‘The positive value’of b implies that we have an increasing trend, Next, we determine the quarterly trend values as follows : For the year 1995, the average quarterly trend value is 32 which is, in fact, the trend value for the middle quarter, é.c., half of the 2nd quarter and half of 3rd quarter, of 1995, Since the quarterly increment is 3, we obtain the trend values for the 2nd and 3rd quarters of year 1995 as 32-15 and 32 + 1-5, ie,, 30° and 33-5 respectively and consequently the trend value for first quarter is 30-5 ~ 3 = 27-6 and 4th quarter is 33-5 +3 = 36.5. Similarly we can get the trend values for other years as’given in the following table. é | é é ¢ | | §ANALYSIS OF TIME SERIES - : . 2:45 TABLE 2:19 : COMPUTATION OF SONAL INDICES rere A Trend Eliminated Values . (Given values as% of trend values) TQrt I Qrt. IU Qrt. WQre | LQre IL Qre. MEQrt. IV Qre. 1995 27-5 30-5 33-5 365 if tos L 131-1 107-5 93-1 _ 1996 39-5 42-5 45-5 485 | ye286-1 122-4 109-9 90-7 1997 515 54-5 575 605s AP 77-7 106-4 93-9 70-3 1998 | 63-5 66-5 69:5 125 85-0 114.3 978 855 1999 | 75:5 18-5 815 84-5 106-0 uT1 105-5 97-0 Total | 436-9 "591-3 5146 445-6 [ic 2 Average (A.M.) Geasonal Indices)] 9278“ 118-26 10292 —«89-12 . “Adjusted Seasonal Indices] 92-07 * 11736 10214-8844 “The indices obtained above are adjusted to a total of 400 (since the sum of indices = 92-78 + 118-26 + 102-92 + 89-12 = 403-08, which is greater than 400) by multiplying each of them 09924. od by a constant factor k, called correction factor given by! he Sum of indices = 40308 =INDAMENTALS OF APPLIED STATISTICS FUNDAMENTALS OF APPLIED 2" ITRODUCTION (Vitat Statistics is defined as that branch of Biometry which deals with data. and the laws of human mortality, morbidity and demography. /The term vital statistics refers to the numerical data or the techniques used 1m the analysis of the data pertaining to vital events occuring in the given section of the population, By vital events we mean such events of human life as fertility and mortality (births and deaths), marriage, divorce, separation, adoptions, legitimations, etc. Sir Arthur Newsholme® defined Vital Statistics as “the whole study of mankind as affected by heredity or environment in so far as the results of this study can be arithmetically stated.” According to him "Vital Statistics forms perhaps the most important branch of Siatistics as it deals with mankind in the aggregate. It is the science of numbers applied to the lifé history of communities and nations.” The following definition due to Benjamin amply explains the utility of vital statistics to planners, various operating agencies, medical sciences, policy makers ete: “Vital statistics are conventional numerical records of marriages, births, sickness and deaths by which the health and growth of a community may be studied." . ~ 91:1. Uses of Vital Statistics. Vital statistics are being extensively’ used in almost all the spheres of-human activity.. We outline below some of the important applications of vital statistics . -1, Study, of Population Trend. The study of births (fértility) and deaths (mortality) gives us an idea of the population trend of any regioii, community. or country. * If ‘Birth Rate > Death Rate, there is an increasing trend : If ‘Birth Rate < Death Rate’, thefe is.a decreasing trend. The division of the. population of different régions (or races) by birth and death rates a enables ts to form some idea about the population trend of the regions or countries and the. general standard of living and virility of the races. The Table 9-1 gives the population of the whole world divided into three different regions together with the birth’and death rates per thousand per.annum and the rate of natural increase per thousand which is defined as the difference between birth rate and death rate TABLE 9-1 : WORLD POPULATION - BIRTH AND DEATH RATES T T ; |, World | Girth Rate per | Death Rate") Rete of Natural Region Countries beg eeation ||. thausand.| per thousand | Increase per | (approximately) ane thousand 1. | Northwest, Central and 20% | iow 10 ! i Southern Europe, North i | America and Oceania. i | 2 | Latin America, Japan, 22% | 15to17 15 Bastern Europe and | : Asiatic USS.R. | | | | 8° | Africa, Near East, 58% 25t035 | 12 | South Central Asia and Far East * Arthur Newsholme ~ The Elements of Vital Statistics 6 a 16 « « &VITAL STATI STICS 9-3 In underdeveloped countries, the birth rate is fairly high but at the same Lime it i: accompanied by high infant mortality rate showing thereby the lack of medical facilities, poor hygienic conditions, malnutrition and low standard of living, : Various types of fertility rates discussed in § 9-7-1 Lo § 97-4 if the population of the region has a tendency to increase, decrea time. For instance, net reproduction rate § 9:83 gives the ral population. ble us to form some idea ¢ or remain stable over { replenishment of the 2. Use in Public Administration. ‘The study of population movement, i.c., population estimation, population projections and other allied studies together with birth and death statistics according to age and sex distributions provides any administration with fundamental tools which are indispensable for the overall planning and evaluation of economic and social development programmes 3. Mortality and natality statistics also provide guide spots for use by the researchers in medical and pharmaceutical profession. 4. Use to Operating Agencies. ‘The facts and’ figures relating to births, deaths and marriages are of extreme importance to various official agencies for a variety of administrative purposes. Mortality statistics serve as a guide to the health authorities for sanitary improvements, improved medical facilities and“ public cleanliness: The data on’the incidence of diseases, together with-thé number ofdeaths by-agé and nature of diseases.aré of paramiotint importance to health-authorities it taking appropriate-remedial action to prevent or control the. spread of the disease. For- example, to control the spread of an epidemic, arraigeinents can be made for-inaciilation or vaccination through municipal and district: local board agencies. , 5. The whole of actuarial science, including life ¢ is based on the mortality’or life tables, The vital records concerning all possible factors contributing to deaths in various ages are indispensable tool in numerous life insurance schemes (For detailed discussion, see life table, § 9:5). : = 9-1-2. Methods of Obtaining Vital Statistics. The vital’statistics data are usually ~ obtained by the following methods _& Registration Method. The most important source of obtaining vital statistics data is the registration method which consists in continuous and permanent recording of vital events pertaining to births, deaths, marciages, migration, ete. These data, in addition to their . $tatistical utility, also have their value as, legal documents. Registration Of births provide - information on place of birth, sex, age and religion of the-parents, legitimacy, number of previous issues and their sexes, father’s occupation and birth place of parents. Similarly, death registration furnishes information on place of death, sex, age, marital status, number of issues, birth place, occupation and cause of death. Similar information is also obtained with respect to marriages and migrants. ; Many countries require compulsory registration of births and deaths under the law. For example, every new birth has to be reported to the authorities along with the information as given above. Similarly, the death of person is automatically recorded since the disposal of the body requires an appropriate death certificate from the authorities. Remark. Data on births and deaths can also be obtained from hospital records. Shortcomings of the Registration Method. In India, in rural areas there is no legislation which makes the registration of vital events (births, deaths, marriages) and7 FUNDAMENTALS OF APPLIED STATISTICS reporting of epidemics compulsory and the requisite information is collected by village Patils under the administrative orders of the government. Consequently, a number of births are likely to remain unregistered especially in scattered rural areas. Even in municipal areas where registration is compulsory, Lhe laws in respect of registration vary from State to State. ‘Thus in India, the statistics of itLhs suffer from the error of underestimation as pointed out in Census of India paper 6, 1954, published by Registrar General of India. “The registration of births is non-existent in some parts of the country and incomplete in varying degrees in all parts of the country. Hence statistics of births suffer from errors of under estimation Similarly in registration of deaths, the data regarding age at death, cause of the death, et are usually unreliable. Due to non-availability of qualified doctors in villages and interior rural areas, quite often the disease and consequently the cause of death remains undiagnosed. Moreover, people have a general tendency to withhold information regarding their diseases particularly in respect of infectious or contagious diseases. Moreover, in our country there ‘are no proper records about the ages of mothers at the time of marriage, at the birth of first child and of subsequent children. The religious customs do not require the compulsory registration of marriages in Hindus and Muslims. Hence we do not get any reliable data in respect of marriages for whole of the country. In order to ensure a continuous permaient recoding of vital events suitable legislation, uniform all over thé-country, should.be introduced, making the registration Of births, deaths, marriages, ete.,.compulsaty. Such. legislation should also provide’ Sanctions for the enforcement; of the.obligation. Separate organisatioi Shoiild bé set-up to collect-this data more completely and:sjystematically. 2.Census Method. Almést in all the countries, all over the world; population census is conducted at regular intervals of tinie, dsually tea years. Cerisus consists of complete enumeration of the population of the particular area under study and collecting information from individuals regarding age, sex, marital status, occupation, religion and other economic and social characteristics. The main drawback of the census method is that it provides vital statistics only for the sus years and fails to give any information about the vital’ events in the intercensal period. Te _/4-2. MEASUREMENT OF POPULATION In order to determine the population at any time ‘’ after the census or between two censuses, a number of methods have, been devised. Here we discuss a suitable method which makes use of births, deaths and migration statistics. If we assume that - (@ the census data. gives us the. total size of the population of region or community together with age and sex distribution and (ii) the birth, death and migration statistics during different periods are obtained from registers, then the population P, at any time ¢ is given by the relation’: : P, =Po+(B-D)+(I-E), (9-1) where Po is the total population at the last census, B and D give respectively the total number of births and deaths in the intervening period ¢ and I is the total number of “immigrants into the region and H is the total number of emigrants from the region during the period ¢. The sex-wise population of India for the years, 1951, 1961, 1971, ..., 2001 is given in Table 9-2 and is exhibited graphically in the Fig. 9-1. The age and sex diemmaeena VITAL STATISTICS Population of India for 1991 census is given Table 9: 2A TABLE 9.24: DISTRIBUTION OF POPULATION OF INDIA BY AGE-GROUP AND SEX : 1991 CENSUS. TABLE 92, POPULATION OF INDIA in 000) t Hi { ‘Age-Group a Female] Total | % age] ' : Vs 50017| 102378] 12.21 : 361,088,090 | 53876] 111295] 12.3) i | | 5 46744] 98692) 11.8! i 36804) 79035] 9.41 | 1961 | 226,293,201) 212,941,570] | 439.234,771 37514) 36959| 74473) 8.9] i 34546 34693| 69239| 8.2) | 29918| 28486) 58404]. 7.0) * 1971 | 284049, 276 | 548,159,652, 27558) 24841] 52399] 6.2] ee ! 1 | | a2gqa] ig7i4). 42556] 5.1 i | | 18955] 17179) ae 43] | 1es1| 353,374,460 | sassy 683,329,097 | 1600s 142g) 31114; 37 IEE ; 10942) wosaa} 21473] 265 pu | \. | | | 11907] 10842} 22749) 2.7! 199i | 439,230,458) 107.072.29 846,302,688}. | 6493] - 6365]. 12858] 1.5] ; i i i | | 40963] 10111) 21074)" 2.5! 7 7 ‘Age iit statéid’| 2706{~. 1990} .--.4695] . 0.6 | 2001, 531.277,078 sil 1,087,015,247|-.| All ages | 435208 403360 838568] 100.0 Source : Otfice of the ‘Registrar General. india’ Notes:.The population of India. for 2001 Source = Office of the Registrar Genéral, India Notes I. Excludes: Jammu ‘and. Kashmir includes the estimated population of. entire Kachchh district, Morvi, Maliya- Miyana “and Wankaner, talukas of Rajkot district, Jodiya taluka of “Where census could. not be held owing to disturbed” conditions prevailing there at the time of 1991 consus. Jamnagar district of Gujarat State and 2, Excludes population of area under entire Kinnaur district of Himachal unlawful occupation of Pakistan and Pradesh where population China. enumeration of Census of India 2001 ¥ could not be conducted due to natural _ calamities. RATES AND :RATIOS: OF VITAL EVENTS Generally the vital statistics are available in the form of frequencies of the vital events, e.g, 2, persons die, np persons are born, mg persons get married during a given time ‘¢” in any community or region or country. In order that these figures are of some utility to a statistician, from statistical analysis point of view, these are generally transformed by rates or ratios. (Rate of a vital event is defined as the ratio of the total number of accuirrences of the event to the total number of pé Remarks 1. Sex Ratio. The sex ratio is defined as the total number of females per thousand males. . _ Female Population fe tio = a . Sex Ratio = "fate Population 1000 ane 100 49-2) The Table 9-3 gives the sex ratio of the Indian population for the years 1901 to 2001 for rural, urban and combined India9-6 TABLE 9.3: SEX RATIO OF INDIAN Gc EUS en i an Sev Hato No.of femates per 1000 males) | bene | ' Rural i Urban i “1901 | 979 | 910 | wu | 915 872 ! 1921 970 846 | 1931 966 | 88 1941 965 | 8a 1951 i 965 | 860 | | 1961 963 | 845, | i971 | 949 | 858 | | 1981" | 1. | 819 ! 1991 @ 938 894 | | 20018” 946 901 | _ | Source : Office at the Registrar General, [ndia Part II B (i).1991 “| Including Assaini | @--Thé 1991 censiis was not held in.J & K.-Por, working out Sex-Ratio for India and J&K. the + Population figures for J&K-as projected by the Standing Cominittee om Popiulation’ Projections (Oct-1989) have been taken . : ‘The 1984 Census could not be held in Assam.’ Ft working out 'sex-fatio for India & Assam, the| -Population figures for 1981 for Assam worked out by interpolation. have been talon, i #: The total, rural and. urban population of India includes the estimated total,-rurai.and. urban | population of entire Kachchh district. Moryi, Maliya-Miyana and Wankaner taluks of Rajkot! distriet, Hodiya taluka of Jamnagar district of Gujarat state and estimeted total and rural] Population of entire Kinnaur district of Himachal Pradesh where population enumeration of| Census of India, 2001 could not be-conducted to due natural calaniities | 2. Usually. the population P, is not constant and keeps on changing from time to time An estimate of the population in the middle of the periods (¢y, | is given by the iean-of the population at the extremities i.e, : Puenpie=4LP, +P, | (9.3) waka MEASUREMENT OF MORTALITY The following are the principal ratés ised in theastiing mottaity. \ S41. Crude Death Rate (C.D.R.). This is the simplest of all the indices of mortality and is defined as the(number of deaths (from all causes) per persons in the population of -any given region or community during a.given period) Thus, in particular, the annual rude death rate (C:D.R.) denoted by m for any given region or community is given by - ; . __ Annual deaths ™ ="Kanual mean population xk, (9-4) where k The crude death rate for any period gives the rate at which the popul through deaths over the course of the period. 000, usually. is depletedee VITAL STATISTICS 7 97 Merits 1.1 i 2. C.D.R. is perhaps the mv mortality, it is used in niimerou: ple to understand and calculate st widely emogeaphi ed of any vital statistics rates. As an index of and public health problems 3. Since the entire population of the region is exposed to the risk of mortality, C.D.R. defined in (9-4) is a probability rate giving the probability thala person_belonging to the given population will die in the given period. Demerits. Most serious drawback of crude death rate-is that it completely ignores age and sex distribution) of the population. Experience shows that mortality is different in different segments of the population; Children in the early ages of their-life, and the older generation are exposed to higher risk of mortality as compared to younger people. Moreover, mortality rate is also different for females irrespective of age groups, than their male counterparts. C.D.R. is not suitabie for comparing the mortality in two places or same place in two periods unless (i) the populations of the places being compared have more or-less the same age and sex distribution, or (ii) two perinds are not too. distant, since in.a stable latge community, age-sex structure ofthe population’ shows very little change. Remark. We can compute the crude death rate f sand females separately. For. example, “efude death’ rate for males in-the given region during the given period is given by the formula Male death." 9g "Ds ag Male popilation -° 2000. “mp, * 1,000. (985 Secale | C.D formales where "D is the aumber“of male deaths and !*P the niale population in the given region duving the given period. Similarly Female deaths ‘D 7 C.D.R. for females: = Fe Population % 1,000 ="Fp x 1,000 (95a) Por usually les between and | 30 per-thousandland femdle C.D.R. is generally less than male C.D.R, eae Specific Death Rates (SDR) In order to arrive at a more useful figure than CD.R., we must take into account the fact that the mortality pattern is differerit in different segments of the population. Various segments may be age, sex, occupation, religion, community, social status, etc. For example, the people engaged in infant or child welfare work would be interested to know thé. mortality condition in, the age groups below 1 year, 1-4 years, 5-9 years, ete. ; those engaged in maternal health programmes would'like to know the number of deaths occurring among women in the reproductive period (usually 15 to 49 years); insurance duthorities would be interested in the mortality pattern at different ages of the population. oh Poca Geath rate computed for a particular specified section of the population is termed as specific death rate (S.D.R.JJS.D.R. for given geographical region during a given period is definedas: Total number of deaths in the specified section of the population in the given period Total population of the specified section in the same period S.D.R. xk (9-6) where k = 1,000 usually. Usually S.D.R omputed specific to (i) age and (ii) sex.98 . FUNDAMENTALS OF APPLIED STATISTICS : . FUNDAMENTALS OF APPLIED STATISTICS 3.D.R.). To formulate ideas mathematically let Age-Specitic Death Rate (Ap »D, = Number of deaths in the age-group (ex + 0) ie., number of deaths among the pe s than x +n, ina given region during a given period, ¢(say) als ‘Then the aye-specific death rate for the age-group x tox +n, usually denoted by ym, is given by fons with age x or more but I population of the age-group x to (x +n). D, ay = “p, * 1000 OT Taking » = 1, we get the annual age- S.D.R. given by m, = 2e,21,000 : (9-8) To be more specific, the age-S.D.R. for males is given by = =i x 1,060 . ian (9:9), 'P . am Ps . where ;'P, is the tutinber-of males in the population in the age-group x tox +n and""D, is the number of deaths amongst this population. : ; Similarly, the age ~ $:D.R, for females is given by the formula ; te f © Tm, = «1,000 7 - (9494) Formulae (9-9) and (9-92) give the death rates specific to both age and sex. Remarks 1: Specific death rates reveal more glaring facts about various segments of a population than the CDR. If the death rate is high in a particular age-group, except old age, preventive measures can be taken to improve upon the situation. The SD’s are extremely helpful in planning-and research. 2. Notations . Regarding symbols used in Section 9-4-2 and the following Sections, the Lower suffix denotes the beginning of the particular age-interval and the lower prefix, the'width of the interval, the upper prefix, if any, denotes the sex and upper suffix, if any, the particular community or place or period under consideration, ac : 3. The symbol , P,-may be interpreted as the total number of persons between ages x. and x47 =I on last birthday (1. 6. d). In particular P, would mean the population-with age x on 1.b.d.. Simi interpretation may be given to other symbols. 4. The age-specific mortality rates for whole of India (rural and urban combined) for years 1971 to 1996 are given in Table 9.4, on pagé 9-9. From the table we observe that with the passage of time, there is a steady decline in thie mortality rates for‘almost all the age-groups up to 55-59. This seems to bé due to advanced medical technology, better medical facilities and general awareness about health consciousness among the people of India. Merits 1. The death rates specific to age and sex overcome the drawback of , since they are tpmbuted by taking into consideration the age and sex composition of the population. By eliminating the variation in the death rates due to age-sex distribution of the population $.D.R’ Appropriate measures of the relative mortality situation in the regions, smpeie ot General analytical purposes, the death rate specific for age and sex is one of the most important and widely applicable type of death rates. TE also supplies one of the essential components required for computation of net reproduction rate (cf § 983) and construction oflife table (ey. ¢ 9s putation 0 and construction of life table . ir 's provide moreee VITAL STATISTICS ———————_— 9-9 TABLE 94: AGE SPECIFIC MORTALITY RATES~ALL INDIA (RURAL AND URBAN COMBINED) Age-Group| 1971*| 1976*| 1981 | 1986] 1987] 1988 1989] 1990 | 1991 @| 1992 @| 1993 @| 1994 e| 1995 e| 1996 oa” [S19] S10 [412/366 35.2 | 393299) 263] 265 | 265 | 257 | 239 | 242 | 239 5-9 | 47 | 48] 40] 33] 33] 32] 28] 25) 27 | 29 | 22 | 21 | os |] 29 wort | 20} 24] 17) 16) a5} is fis}i} oa | va | va | ve | aa | as 19 | a4] 27 | 24) 23) 22) aa} ar}ar) ar! 22] 19 | rs | is | i 20-24 | 36 | 34] 31) 29] 28] 28] 29] 27] 28 | 28 | 26 | 25 | a4 | 23 2-29 | 37) 39) 32] 30] a7] 26) 28} 26] 31 | 27 | 27 | 27 | o66| 25 30-34 | 46] 45] 40] 33/32] 30] 29] 30] a1 | 32 | 28 | 32 | 29 | 29 36-39 | 57] 48} 44] 42) 37] 40] 36) 36] 39 | 38 | 34 |-35 |-a4 | 39 40-44 | 67] 72 | 58] 56] 53| 54] 50] 51] 48 | 51 | 45 | ae | a7 | 49 45-49 | 95 | 95 | 85] 78) 76] 78] 74) 77| 74 | 75 | 67 | 72 | 68 | 67 50-54 | 168 | 16.2] 13.2}126| 11.8] 122] 10] 12] 13 | 5 | it2 | 110 | 103 | 109 55-59 | 212] 23.6 | 206) 178] 17.9) 14 170] 178 | 176 | 178 | i76 | 160 | 147 | 15.7 60-64 | 349 | 40.3 33.0] 313] 30.7| 29.7] 277| 259| 285 | 286 | 275 | 270 | 247 | 267 65-69 | 48.4 | 51.4 | 46.4 | 44.9] 42.3 | 45.0] 426] 425 | 416 | 438 | 403 | 381 | 359 | 389 ~ 10-74 |109.3] 99.6 | 97.4] 91.0] 89.4] 93.8| 854] 85.1 | 914 | 915 | 876 | a77| 572 | 615 75-79 | NA | NA.]| NA} NA| NA} NA] NA| NA] NA | NA | NA | NA | 809 | 89.5 so-s¢ | NA] NA | NA|NA|-NA| WA] NA| NA] NA | Na | wa | wa | 19.9] 1204 853 | NA | NA | NA| NA] NA| NA| NA] NA|.NA | NA | NA | NA | 208-4| 1822| (alleges [149 [75.0 [aes [ni] 109 [uo fis er se [ior | 33 [aa | 90 | 90 "Sotirce : Office’of the Registrar Genera, India, Sample Registration System ~ : Excludes Bihar and West Bengal. @ : Excludes Jammu & Kashmir (NA:Not available ; From 1995 onwards, age’group extent Demerits 1. However, S.D.R.’s are not.of much utility for. overall compaiigon of mortality conditions prevailing in two different regions, say, A and B. For ‘example, it might happen that for certain age-grouips the mortality pattern for region A is greater than that for B but for thé’ others: the case may be opposite. Hence it will not be possible to draw general conclusion regarding the overall mortality pattern‘in region A_as compared to the region B. In order to draw some valid conclusions, the different age or/and sex specific death rates imust be combined to give a single figure, reflecting the true picture of mortality in the region. 2. Moreover, in addition to age and sex distribution of the Population social, occupational and topographical factors come into operation causing what: is called differential mortality. completely. ignore these factors. In order to eliminate such spurious effects, dised death rateDare computed. 2 Infant Mortality Rate (LMLR. e infant mortality rate is defined as the chance of dying ofa newly born infant within a year ofits life, under the given mortality condition. Notations : a ; Dj : Number of deaths (excluding foetal deaths) among the children between the age- group 0-1 ie., the number of deaths among the children of age 0 on last birthday (1.b.d.) among the residents of a region during the calendar year z. Bi : Total number of live births reported in the same region within the same calendar year 2. ‘The infant mortality rate during the calendar year z, usually denoted by is, , is given by : 28 49-10) Es BeNDAMENTALS OF APPLIED STATISTIC: FUNDAMENTALS 9-10 , - SDR for age 0Lb.d. 6 Lbd. Vs. LLM. The LM.R. and the age - SDR for age 0 l.b.d. we theemane h or, However, the denominator in age-SDR at age 0 l.6.d. includes ait te patie iar Golo ear Usually ‘Ue cttimatos af he population by age are not available eaten callable it ts a well known fact that the tolal number of infants below 1 year are very much under enumerated, Hence, the age-SDR for age 0 bd. (rm) is highly overstated. Accordingly, LMLR. is regarded as a more useful measure of infant mortalily as compared with mp. 2. An obvious advantage of LM.R. is thal il does not require the data of population censuses or estimates. The [.M.R. can be obtained for any population, any region (area) and for any time period, wided the data for infant deaths (Dj) and the live births (Ii) for that period are available. However, ne is not Lrue of age -SDR for age. 0 U.b.d. because the estimates of the population 0 /.b.d. may riot. be available for small area. , 3. Strictly speaking 1.M.R. is not a probability rate. For instance, for the infants born in any month, say, November of the calendar year (2 ~ 1) and dying in the month, say , May of the calendar year 2, the death is recorded in the numerator but birth is not recorded in the denominator of (9:10). Similarly , for the infants born in any month, say August of the calendar year z and dying in, say, March of the year (+1), the death is not recorded in the numerator but the birth is recorded in the denominator of (9-10). Alternatively, we can say that a child born on Ist February of calendar year z is exposed to the risk of death in that year for'11 months, while a child born on Ist November of the same year z is exposed to this risk for only two months. 4. In most of the countries all over the world, it is observed that J.M.R. (the risk of death under 1 year of age) is higher-than-the mortality rates for any othet age groups of the life span, except at the, very old ages. Moreover, unlike the.old-age deaths, the’infant-deaths-are very responsive to the improvements in the environmental and medica! conditions. Thus, the I-M-R:.is regarded as a very’ sensitive index of the health conditioris of a:community’or country and réflects’ any changes.in its health standards. é 5. N’setious drawback of LM.Ris the under-registration of live births’ Also the definitions of live ‘country from time.to time. Quite often, the infants who are bora alive but die imimediately after birth are usually recorded as ‘dead’. This results in under-registration of live births and over-registration leaths, leading t statement of [.M.R. than: its actual value: Thus, by improving the birth registration-system,.we can Tower the I.M.R. without saving a single life. Table 9-5 gives the infant mortality rates for rural, urban and combined India for some € years from 1971 to 1999. : Table 9:5(A) gives the IMR for India for 1998-1999 according to the sex of the child and the educational backgrourid of the mother. : Table 9-5(B) gives the IMR for SAARC Countries’ for the years 1996 and 97 for comparative studies. From Table 9-5, we observe that there is a steady decline in the LMAR: over the years € 19971 to 1999, reflecting upon the improvement in the environmental and medical conditions in India. Table 9-5(A) suggests that the education of the mothers plays an important role in reducing the IMR. This is perhaps due to the general awareness about the child care and health, among the educated mothers. : Remarks. 1.Age - SDE births and still births vary from country to, country. and for the sa: é « TABLE 9.5: INFANT MORTALITY RATES BY TABLE 9.5(A) : INFANT MORTALITY. RATE RURAL AND URBAN AREAS ~ 1999 BY BACKGROUND CHARACTERISTIO. € (PROVISIONAL - INDIA) 1998-99 — INDIA" Year Rural___Urban Combined Background Infant j 197i 138 82 129 Characteristic Mortality § | 1976 ~ 139 80 129 Rate | | aot 40 81 130 Sex of Child \ | 1978 137 " 127 Male ug © 1 1979 130 n 120 Female ecWiTAL sranistics ice oti 1981, 119 62 10 Mother's education | | 1982 iM 65 105 illiterate 86.5, 1983 14 105 e < Middle complete 58.5 | | 1984 ua 104 school complete ca 1985 107 ov nd above | 1 1986 105 96 National Family Health Survey | 104 95, 1998-99 International Institute for 102 4 Population Sciences, Mumbai 9s 58 a1 He | 86 50 80 9.5 (B) INFANT MORTALITY RATE | 87 53 . 80 FOR SAARC COUNTRIES | 85 Infant Mortality Bes Bd SAARC Countries fale | i994 80 0 tive births) | | 1995" 80 {1996 1 India 73 7 | 1997 1. Bangladesh 83 ai | | 19985 “11 Bhutan 90 87 [1999 5 Maldive 54 53 | Source : Office of the Registrar General, India, Nepal 82 5 4 I (Sample Registration system) Pakistan 7 “| 95 95 | I Excludes Jammu & Kashmir and Mizoram SriLanka 1 we} :Bitinate at. the national. lovl: ticladas .. Source’: Human Development Report, 1998 Jand (Rural) due to part-teceipt of returns. (UNDP) vas Standardised Death Rates. Using (9-4) and (9-8), the crude death rates in terms of agé-Specific death rates for two regions A and B are given respectively by : De Dims Pye mt = Sp 1000 = Sp -A9-11) DP ime Pb and mb = 5px 1,000 = Sp 5— TPE (9-110) The expressions in (9-11) and (9-11) are nothing but. the weighted arithmetic means of the age-S.D.R., the weights being the corresponding ‘populations inthe age-groups. We observe that even if age-S.D.R’s are same, ie., a Pe =P; Py P,P me=m,eWx, m*#my, since in general, ie, since the age-distributions of the populations in the two regions A and B are not identical. This drawback is removed if the same set of weights is used in (9-11) and (9-11 a) for computing the weighted average of the age-S.D.R’s. This is what is done in standardised death rates (STDR) or adjusted death rates, used with a prefix to identify the basis of adjustment as, for example, age-adjusted death rates and so on. We discuss below the two methods of age-adjustments, which are in common use.9-12. FUNDAMENTALS OF APPLIED STATISTI FUNDAMENTALS OF APPLIED STATIST! Direct Method of standardisation. This method consists in weighting the age specific death rates not by the corresponding population of the area to which they refer (as is done in C.D.R.) but by the population distribution of another region chosen as a standard. Thus, if P.* is the number of persons in the age-group x tox #1 in the standard population, then the standardised deate rates for the regions A and B are given respectively by : Tm Pe Im, P, (SIDR, = *pe~ and (STP Rn = ER (9-12) These-age adjusted death rates for regions A and B respectively are nothing but the crude death rates that would be observed in the standard population if it were subject to the age-S.D.R. of the regions A and B Reiarks 1.'The death rate may similarly be adjusted for other factors also such as sex, rate, etc., and may be interpreted accordingly. ; 2. Standardisation of age and sex together. Let "mg and fm, be the age $.D.R.s at age x (I.b.d.) corresponding to male and female population respectively in the region A and lel ™ Py and f P; be the corresponding male and female populations in the standard population. Then the age-sex adjusted (standaridised) death rate for.the, rman Ai given by the formula © poe De ing IPS bo . Tee Pay (9125) Similarly we can obtain STDRs adjusted for any other two factors, ey .g., age-race, sex-race and so on.. Merits I. Standardised death rates ate readily comprehensible and easy to calculate, 2. Age-adjusted death rates ‘are Comparable sinée thiéy eliminate the difference caused by ¢ the different distributions of the age-specific population for regions A and B . The difference in these death rates givés a true picture of the difference in moftality in the two regions. Demerits. The main drawback of the method is:the choice of a standard population’. The ¢ choice of this ‘standard’ is bound to affect the magnitude of the resulting adjusted rates and may change their relative positions with respect to each other. This bias may, however, be eliminated by taking the standard population as actual population (or life-table stationary € population) of a bigger region of which A and B are subsets. For instance, to compare the mortality pattern in the States of Punjab and Rajasthan, the standard chosen may be the population of Northern-India or whole of India. However, there is no. generally accepted standard population for international comparisons. Indirect Standardisation. In computing the standardised death rates by formula: (9-12) it is necessary to know the number of persons and the age-specific death rates for different age-groups. Quite often we have a population classified by age but the age-S.D.R.’s may not be known. However, the total number of deaths and hence C.D.R. may be known.'In such case, we use the indirect method of standarisation which consists in multiplying the crude death rate of the region A, say, by adjustment factor ‘C’ measuring the relative ‘mortaity’ proneness of the population of the region such that the result is equal to the standaridised death rate. Thus the problem is to find C such that CDRxC =S.T.D.R. DmePe Lima Ps TmePs Tmo, > Spe * ce o 6255s SaVITAL STATISTICS . 9-13 Since mf are not usually known, an approximate value of C is obtained on replacing m,* by m; , the age - S.D.Rs for the standard population, thus giving Em, Py" mg P," rp * EPS “ER Finally, the (indirect) standardised death rate for region A is given by [Irom (9-13)] C= =(C.D.R. for standard population) (9-134) STDR for A = (C.D.R. for A) x é --(9-136) Remark. Actually there is no point in comparing the two methods of standardisation. We use indirect standardisation as an approximation to direct standardisation only when the necessary data for the latter is not available. The two methods will be exactly equivalent if the age-S.D.R. for the given population happen to be proportional to the SDRs of the standard population. Both the methods of direct and indirect standatisation are subject to the obvious criticism that the mortality indices so obtained depend on the age-sex composition of the standard population used and as Such the greater gains (losses) in mortality reduction obtained at younger (older) age are not adequately accounted for, TABLE 96 Age-group._| A 7 Pee Example 9-1. Compute the |" (Wears). Population ...Deaths “|Population, Deaths crude and: standardised déath| Under 10 30,000 600°" }..~12,000 372 «4 rates of the two populations A]- 10 —20 12,000 240 30,000 660 and. B, -regdiding -A as|..20—40 | 50,000 1250 "62,000" 1612 standard population, from the | -4o— 60 | _ 30,000 “"jo50° "| ‘i5,000. 525 “| data in Table 9-6. ‘Above 60 | 10,000 500 |. 3,000 180 | Solution. TABLE 9.7: COMPUTATION OF C.D.R. AND S.7.0.R. _._ Population A ‘Population B ‘Age-group | Population | Deaths, | Death rate| Population] Deaths | Death rate} yz pt (Years) Pf Ds | per 1,000°} - pb Ds | per 1,000 a i me ‘ m> | “Under 10] — 20,000. 600 30... 32,000 |" 372 31 6,20,000 io-20| 12,000 240 20 30,000 660 22 2,64,000 20-40] 50,000 1250 25 + | ., 62,000 1612 | °° 26°). 13,00,000) 40-60) 30,000 1050 35 15,000 525 35 10,50,000 Above 60 | 10,000 500 50 3,000 180 60 6,00,000 Total | .1,22,000 | 3,640 1,22,000 | 3,849 38,34,000 Crude Death Rates : i a 1000 = 2549 x 1,000 = 298 CDR. for population A = $5-5% 1,000 = 799.999% bi 0 = 294 DDS i x 3349 - CD.R for population B = £p5%1,000 = 759,999 % 1000 = 27-49.14 SDR for A STDR for B= * FUNDAMENTALS OF APPLIED STA1(S FUNDAMENTALS eee ard population, C.D.R. for A = 2! Yin, LP We, thus conclude that death rate in population B is greater than in population A Example 9-2. Bstimate the standardised death rates for the two countries from the data in Table 9-8, TABLE 9.8 Age group (in years) Death Rate per 1000 Standardised Country A Country B Population (in lakhs) - 0-4 20-00 5:00 100. 5-14 1-00 050 200 15-24 140 1-00 190 25-34. 2°00. 100 ” 180 35 ~ 44 3307 2:00 120 45-54 7-00 “** ‘5:00:. 100 55 ~ 64. "15-00 12:00 “70 ‘ 65-74 2-40-00. (35:00. 30 75 and above “120-00 110-00. & 10... Solution. : TABLE 9-84: CALCULATIONS FOR STANDARDISED.OEATH RATES f is Deatli rate per 1,000 Standardised Ageeroup |~“CountryA | CountiyB | Population mg Pg me Ps (Years) me me” ou 20-0 50 100 2,000.0 500-0 5-14 10 05 200 200-0 100-0 15-24 14 10 190, 266-0 190-0 25-34 30 10 180 3600 1300 | € 35 = 44 33 20 120 3960 | 240.0 45-54 70 50 100 00-0 500-0 55-64 150 120 0 1,050.0 ado | |B 65-74 40-0 35-0 30 1,200-0 1,050-0 ‘75 and above 1200 110.0 10 1,200.0 1,100.0 Total 1000 1372 4,700 |@ ‘Standardised Death Rate for CountryA = (STDR), mare = 7372 8 bps Standardised Death Rate for Country B = (STDR)g = tare “ = 47 Example 9:3, Find the standardised death rate data given in Table 9.9. by Direct and Indirect methods for the 8VITAL: STATISTICS 9-15 TABLE 9.9 zl “Standard Population Population A Age Population Specific Death | Population ‘Specific Death (in 000) Rate fin’ 000) Rate | P O-5 eee ee eee tee 8 5-15 10 13 4 15-50 27 15 9 50 and above 5 10 59 Solution TABLE : 9-9(a): COMPUTATION OF STOR BY DIRECT AND INDIRECT METHODS (aaa "[__ Standard Population Population A maps | me Pe I e Pf me | Peme | Pe me Pame 0-5 8,000 50 | 400,000] 12,000]. 48 |-. 576,000]. 384,000] 600,000 5-15 10,000 | 15° | 150,000] 13,000} 14 | 182,000] 140,000] 195,000 15-50 27,000 | 19 | 270,000} 15,000 9 |. 135,000] 243,000] 150,000 | 50-and above 5,000 60 300,000} 10,000 .-59 590,000} 295,000] 600,000 {Total 50,000 ~--] 11,20,000 50,000 ~=|-14,83,000| 10,62,000 15,45;000- Direct Method ; oem Eine Pt -10,62;000 ; (STDR),°=. SPE = 750,000 = 21-24 : Indireét Method : 7 14,83,000 Fo.000 = 2255 Adjustment Factor 11,20,000 50,000 ~_ Ym, P,* = 50,000 * 15,45,000 ~ 0-249 f (STDR)q =C x(CDR)y= 0°7249 x 99-68 = 21-5005. ee aaatit TABLE OR LIFE TABLE : The life table gives the life history of a hypothetical group or cohort as it is gradually diminished by deaths{ It is a conventional method of expressing the most fundamental and essential facts about the age distribution of mortality in m and is a powerful tool for measuring the probability of life and death of various age sectors) A life table provides answers to the following questions : (@ How will a group of infants all born at the same time and experiencing unchanging mortality conditions throughout the life time, gradually die out ? i) When in the course of time all these infants die, what would be the average longevity per person ? (iii) What is the probability that persons of specified age will survive a specified number of years ? (iv) How many persons, out of selected number of persons living at some initial age, survive on the average to each attained age ?9-16 FUNDAMENTALS OF APPLIED STATISTICS € The life table thus asummary of Lhe mortality experience of any population group during a given period and is a very effective and comprehensive method for providing concise ‘ measures of the longevily of that population ‘The data for constructing a life table are the census data and death registration data. Life tables are generally constructed for various sections of the people which, as experience ¢ shows, have sharply different patterns of mortality. Thus there are life tables constructed for ” different races. occupational groups and sex. Life tables are as well constructed on regional basis and other faclors accounting differential mortality. We give below a detailed discussion of the various terms and factors that enter in the construction of the Life Tables. Notations and Terminology : § 4,-is the number of persons living at any specified age x in any year out of an assumed nuniber of births, say, /p tisually'called the cohort or radix of the life table. dis the number of persons among the /, persons (attaining a precise age x) who die before reaching the age (x'+1). Obviously, we have dy a ly—ly gp ae Abgttie lin . Og € where A is thé-differencce operator . 4px is the ptobability that-a.person‘aged x'Survives up to age x + n.. Thus if J,,°is:the number of persons living at age (x + n) in any year, then _€ Pr, ae So bean be De 2 (G15) he CB In particular, if we take n= 1, we have | Ty, Wei] \ps a Ol ag which gives the probability that a person aged x will survive ill his.next. birthday. \ p, , is the complementary probability of survival, ie.,q,ris the probability that pergon of exact age x will die within one year following the attainment of that age-"Thus, we, d,” have / a= = Oy ” Note. Also, by definition : ‘ beey She dy=ly(1- Ge) (From (9-16 by > bent = be Ps 9.16 a) § L, is the number of years lived in the aggregate by the cohort of lp persons between age and (x sT) or L, may be interpreted as the average size of the cohort between ages x andg (@ #1). Thus, if deaths are assiimed to be uniformily distributed over the whale year or equivalently, if we assume the linearity of /, ,, for ¢€ (0, 1J, then we get 8 1 Le] leapt and iy,=l,—td, cs 0 1 98 b= | Ge-td,)at « aVITAL STATISTICS : : 9-17 ale ft [ode | 222 | =1,-$d,= ly —$ (le teat) i (9D | Lane a ills then . (9-174) From (9-17) , on using (*), we also get Ly =k amy (9-17 b) T, is the total number of years lived by the cohort ly a after attaining the age x, Le., 7’, the total future life time of the /, persons who reach age x. Thus, we have T, = Ly they yt beset ... (9-18) Thus if w is the highest age at which any survivors are recorded in the mortality table, ie. if l,,=0, then : .. (9:18 a) (9-185) *‘ ffexpectation of Life. The Curate. Expectation. of life, usually, denoted bye, ‘gives the average number of complete years of life lived by’ the cohort lp.after age x by each of “pe sons attaining that age. The complete Expectation‘of tife, usually: denoted ‘by-ey8 measu the average number of years a person of given age x can be expected to live under prevailing mortality conditions. It gives the humber of'years of lifeentirely completed includes the fraction of the year survived in the year in which death occurs,.which on the average can be taken to be (1/2) year. Thus we have” e9 =e, + i : -.(9-23) Since total number of years lived by /, persons of age x is given by : = lege dt, 70° the complete expectation of life of a person attaining age x is obtained from fe following relation : E = --(9-23a) le eo”, the expectation of life at age 0, is the average age at death or the average longevity of a person belonging to a given community. ey1. deel x+f 243 4. m,= 2-4 x J. Trak t+ hy thyyat.. ca ao ke Beam) =m => T..1=T,-L, _ |g zplitt a Ae oh)9-33 VITAL STATISTI os of Life: Tables, Although the basic objective of life tables is to Five ace picture of the age distribution of mortality in a given population group, it has ee i widely ina large number of spheres. Today life table is widely accepted as impor i aa material in demographic and public health studies. In the words of William Farr, ite able is the ‘Biometer’ of the population. We enumerate below some of important applications life tables. 1. For Use by Actuaries in Insurance. Lif 0 : all questions concerning the duration of human life. ‘These tables, based on the scientific a of statistical methods, are the key stone or the pivot on which the whole.science of life assurance hinges. Life tables form the basis for détermining the rates of premiums to be paid by persons of different age groups, for various amount of life assurance. Life tables provide the actuarial seience with a sound foundation, converting the insurance business from a mere gambling in human lives to the ability to offer well calculated safeguard in the event of death. 2. For Population Projections. fe tables are indispensable for the solution of Life tables are used by demographers to devise measares such as ‘Net Reproduction Rate’ to study the rate of growth of population . They have also, been used in preparation of population projections by age and sex, i... in estimating what the size of the populasion.will be at some future date: 3. For Coniparison:of Differént Populations. Lifé tables for two or more different groups of population may be used for the relative’ comiparisan ‘of varidus- measures. of. mortality suchas death tate, expectation of life at various ages, etc; Of particular interest is the comparison ofe,%,:thé’average longevity for members 6f a population. : 4, Life tables are as well used bythe governmient and’ the private establishments for determining the’ rates of retirement benefits to be given to its employees or for formulating “various programmes for retired persons. ot 5. Since a life table depicts the distribution of the people according to age and sex, it is extremely. useful in planning in, respect of education and for‘ predicting the. school going population if connection with school building programmes. 6. Life tables are also used : (i) For.making policies and programmes relating to publi i ing pd ublic health, b: and public administration fa publi healih by fhe governs (ii) To evaluate the impact of family welfare programmes on the population growth (iii) For estimating the probable 1 ws probable number of r of future widows and orphans in a (iv) For computing the aj i i an puting proximate size of future labour force and military forces,. FERTILITY : In demography, the word fertility is used in relation to the actual production of children or ‘occurrence of births, specially live births’. Fertility must be distinguished from fecundity which refers to the capacity to bear children. In fact, fecundity provides an upper bound for fertility. As a measure of the rate of growth of populiti n various fertility-rates -are computed. eNO oe In the following’ sections we shall discuss briefly some of the important rates which are usually comptited to have an idea about the fertility in the relevant section of the population. 9-7‘. Crude Birth Rate (C.B.R.). This is the simplest of all the measures of fertility and consists in relating the number of live births to the total population. This provides an index of the relative speed at which additions are being made through child birth. The fertility pattern of the above mentioned measure is given by crude birth rate (C.B.R.) defined as follows : : a Crude Birth Rate = pe xk ... (9-82) where Bt =Total number of live births in the given region or locality during a given period, say ¢. Pt =Total population of the given region during the period ¢. k =Aconstant, usually 1000.VITAL_STATISTICS 9.45 Merits. Lis simple, easy to calculate umber of births (B9 and the tolal si Knowledge of these figures for different se Demerits 1.'The crude birth rate, though reliable since it completely ignores the x distribution of the population. 2. C.B.R. is nol. a probability ratio, since the whole population * cannot be regarded as exposed to the risk of producing children. In fact, only the females and only those between the child bearing age group (usually 15 to 49 years) are exposed to risk and as such whole of the male population and the female population outside the child-bearing age should be excluded from P!, the denominator in (9-82). Moreover, even among the females who are exposed to risk, the risk varies from one age group to another, a woman under 30 is certainly under greater risk as compared to a woman over 40. 3. As a consequence of variation of climatic conditions in various countries, the child. bearing age-groups are not identical in all the countries. In tropical, countries, the period starts at an apparent earlier date than in-countries with-cokd weather. Accordingly, crude birth rate does not enable us to compare the fertility situations in different eountries. 4. Crude birth rate assumes that women in all the ages have the same fertility, an assumption which is not true since younger women have, in general higher fertility: than elderly women. C.B.R. thius-gives us’'an estimate of-a-hetrogeneotis figureand is unsuitable for corparative-studiés : 5..The level ofcrude birth rate is determined by 4 humber-of factors such as age atid sex distribution ofthe population, fertility of-the population, sex ratio, marriage rate,-migration, family planning measures and, so on: Thus a relatively high crude birth rate may be observed in a population with a favourable age ’rid'sex structure even though fertility is low, i.e., a population with large proportién of the individuals in the age-group 15—49 years will havea. high crude birth rate, other things remaining same te Remarks 1. C.B.R. usually lies between 10 and.55 per thousand. 2. Since it is only a live birth that augments the existing popula in measuring fertility, thus excluding still births. : 3. The Table 9-15 gives C.B.R., C.D. infant mortality rate (I.M.R.) for rural, u:ban and combined india for different years. Also see Table 9-17A on Page 9-49. TABLE 9-15. CBR, COR AND IMR FOR RURAL, URBAN-AND COMBINED INDIA 1970 1980 1990 -1998...1999.:.2000.. 2001. 2002 2003 Birth rate (Per ’000 population) nd readily comprehensible. It is based only on the of the population (P) and does not necessitate thé of the community or the population. e, is only a crude measure of fertility and n, only live births are'considered Rural 389 346 317-289 280 27-6: 276 271 266 Urban 29-7 281 247 U5 WO 28 2-7 203 199 Combined 36-8 333 302 272 265 26:0 258 25-4 25-0 Death rate (Per '000 population) ; Rural ; 173.135 105 96 97 94 93 Qi 87 Urban 102 80 68 65 66 63 63 63 61 Combined 157 124 97 89 90 87 85 84 81 Infant mortality rate (Per ‘000 live births) Rural = 14 86 OCOT:sC‘TT?:;C*C*CTS:Cié‘SASC*«‘CTNC;é«iWS Urban 6 50 45 «45 44 44 42 40 Combined - 14. 80 71 72 ~— 70 —B8GCSC ‘Secondary Source : Monthly Abstract of Statistics (C.S.0.) Volume 56, Number 10, October 2003.