nfuence ths demand. We wanted to estmate demand for Ce phone by coege students, what varabe are key to affect. Probabe answers can be 1 Prce of the product. 6 Prce of substtute products 2 Taste 7 Prce of compementary products 3 Preferences 8 Income of the consumer. 4 Fashons 9 -- Popuaton 5 -- Advertsement 10 Future expectaton of prce etc. Ideay a varabes that are key to nfuence the demand shoud be ncuded In the regresson anayss. However the varabe ncuded n the anayss are based on avaabty of data and cost of generatng new data and ts reabty. Two types of data are used n regresson anayss. 1) Tme seres data. It gves nformaton about a varabe over a perod of tme. 2) Cross sectona data. It provdes nformaton on a gven varabe for a gven perod of tme. Suppose we have gathered the nformaton from dfferent coeges n Inda on foowng. 1 Average no of ce phones sod per months 2 Average charges for one mnute 3 annua semester tuton fees 4 Post pad charge card rates 5 Locaton of the campus The reasons for seectng these varabes are based on economc theory of demand. Some tmes researchers are requred to use ther creatvty n comng up wth varabes whch measures varabe about whch reabe data can not be gathered. Here tuton fee s such varabe whch s proxy for ncome of a consumer. The ocaton s ncude to determne whether the demand for ce phones s affected by the number of avaabe substtutes for ce phones such as pubc ca offce ( PCO ), operatng nsde the campus. The assumpton s that the coeges n urban areas may have more PCOs from whch to make cas convenenty, and ths mght adversey affects the students demand for ce phones. Y Ouantty, X1- Prce of ce phones (n Rs.00), X2 Tuton (In Rs. 000 ) X3 Prce of post pad charge card n pasa, X4 Locaton. Co ege Y X1 X 2 X3 X 4 1 10 10 0 1 4 10 0 1 2 12 10 0 1 6 95 1 3 13 90 8 11 0 1 4 14 95 7 90 1 5 9 11 0 1 1 10 0 0 6 8 12 5 5 10 0 0 7 4 12 5 1 2 12 5 1 8 3 15 0 1 0 15 0 0 9 15 80 1 8 10 0 1 10 12 80 1 2 90 1 11 13 90 6 80 1 12 14 10 0 5 75 1 13 12 10 0 1 2 10 0 1 14 10 11 0 1 0 12 5 0 15 10 12 5 1 4 13 0 0 16 12 11 0 1 5 80 1 17 11 15 0 1 6 90 0 18 12 10 0 1 2 95 1 19 10 15 1 10 0 0 2 0 20 8 15 0 1 0 90 0 21 9 15 0 1 3 95 0 22 10 12 5 1 5 10 0 1 23 11 12 5 1 6 95 1 24 12 10 0 1 7 10 0 0 25 13 75 1 0 10 0 1 26 10 10 0 1 2 11 0 1 27 9 11 0 6 12 5 0 28 8 12 5 1 0 90 0 29 8 15 0 8 80 0 30 5 15 0 1 0 95 0
Usng ths data, we then express the regresson equaton to be estmated n the foowng near equaton. Y = a + b1X1 + b2X2 + b3X3 + b4X4. Y Ouantty of ce phones demanded. A Constant or Y ntercepts. X1 Average prce of a mobe phone mode. X2 Annua tuton fees (n thousands) X3 Average prce of a ca for one mnute (In Pasa) X4 - Locaton of campus. (1 f ocated n an urban area, 0 f otherwse) b1, b2, b3, b4, - Coeffcents of the X varabes measurng the mpact of the varabes on the demand for Ce phones. Among the software packages used by Economst to conduct a regresson anayss of the demand for a good or servces are SPSS and other software. Usng regresson functon n Exce, we obtaned the foowng regresson equaton. Y = 26.67 0.088X1 + 0.138X2 0. 076X3 0.544X4 (0.018) (0.087) (0.020) (0. 884)
(Standard error of coeffcents are sted n parentheses) Standard error of Y estmate s 1.64 R = 0.717, Ad|usted R = 0.67 F = 15.8 Positive sign represents direct relationship and negative sign represents inverse relationship. Each estimated coefficient tells us how much the demand for cell phones will change relative to a unit change in each of the explanatory variables. For example, b1 of 0.0 indicates that a unit change in price will result in a change in demand for cell phones of 0.0 in the opposite direction. Prce of ce phone (X1) = Rs. 100 (.e., Re. 10,000) Annua semester coege tuton (X2) = 14 (.e., Rs. 14,000) Prce of a prepad mobe card servce=110(X3) (.e., Rs. 1.10) Locaton of campus (X4) = urban area (.e., X4=1) Therefore nsertng these vaues n to the estmated equaton gves us Y = 26.67 0.088(100) + 0.138(14) - ).076(110) 0.544(1) = 10.898 or 11 rounded. Prce eastcty of demand = - 0.88* 100 / 10.898 = - 0.807 whch shows that demand s reatvey neastc. Smary ncome eastcty of demand s +0.177 whch shows that demand s reatvey neastc. Statstca evauaton of regresson resuts. Our regresson resuts are based on a sampe data. How confdent are we that these resuts are truy refectve of the popuaton of coege students? The basc test of statstca sgnfcance s t test. The conventon n economc research s to seect 0.05 eve of sgnfcance. !his means you can be "#$ confident that the results obtained from the sample are representative of the population. %omputed &t' value is greater than table value we say that estimate is significant. T vaue s cacuated by dvdng estmated coeffcent of ndependent by ts standard error. A smpe and usefu way to hande the crtca eve s to use the rue of 2. Ths means that f the absoute vaue of t s greater than 2, we can concude that the estmated coeffcent s sgnfcant at 0.05 eves. Another mportant statstca ndcator used to evauate the regresson resut s the coeffcent of determnaton or R. We aso cacuate R2 . This measure shows the % of the variation in the dependent variable accounted for by the variation in all the explanatory variables in the regression equation. Ths measure can be as ow as zero and as hgh as 1.0. One ndcates that a varatons n the dependent varabes can be accounted for by the expanatory varabes. Zero ndcatng that the varatons n the dependent varabes are not accounted for by the varaton n the expanatory varabes. Normay arger the number of varabes n equaton hgher the vaue of R. Hence we have another aternatve measure caed ad|usted R. It takes care of more number of varabe. In ths case ad|usted R s 0.67 hence equaton s sgnfcant. Another test caed F test s aso used n con|uncton wth the above test. It aso tes us whether entre equaton s statstcay sgnfcant or not. If cacuated F vaue s greater than tabe vaue equaton s sgnfcant. Demand forecastng can be done by many ways. Some of them are 1 Expert opnon --- a) |ury of executve opnon; b) Deph Method 2 Opnon po and market research 3 Survey of spendng pan 4 Economc Indcators 5 Pro|ectons 6 Econometrc Mode Economc Indcators: a) Leadng Indcators 1 average hours manufacturng 2 Inta cams of unempoyment nsurance 3 Manufacturers new orders for consumer goods and materas 4 Index of consumers expectaton 5Money suppy 6 Stock prces 6 Budng permts a b) Concdent Indcators 1 Empoyee on non agrcutura pay ro 2 Industra producton 3 Manufacturng and trade sae c) Laggng Indcator 1 Average duraton of unempoyment 2 change n abour cost per unt of output 3 Average prme rate charge by bank 4 commerca or ndustra oan outstandng