You are on page 1of 29

ECONOMETRIC

ANALYSIS
Dr Alka Chadha
IIM Trichy
What is Econometrics
• Literally means “measurement in economics”.
• Application of statistical and mathematical methods to the
analysis of economic data, with the purpose of giving
empirical content to economic theories and verifying them
or refuting them.
• An econometric model consists of a set of behavioural
equations derived from an economic model. These
equations involve some observed variables and some
‘error term’.
• The error term is the catchall for all the variables
considered as irrelevant for the study and all unforeseen
events.
Yogurt exercise
0.12

0.1

0.08

0.06

0.04

0.02

0
0 2000 4000 6000 8000 10000 12000
Yogurt demand curve
0.12

0.1

0.08

y = -5E-06x + 0.1192
0.06 R² = 0.6177

0.04

0.02

0
0 2000 4000 6000 8000 10000 12000
Calculating elasticity using regression
• Get time series data on prices, incomes and prices of
related goods, and run a regression to find the
coefficients.
• Estimate the demand equation using these variables and
coefficients of the explanatory variables are the slopes.
• Then estimate the elasticity at various prices and
quantities, or incomes and quantities, or prices of other
goods and quantities using the estimated coefficients.
A behavioural model
• Q = a + bP + u
• This is a statistical or stochastic relationship as opposed
to a deterministic or mathematical relationship.
• A specification of the probability distribution of u says that
the disturbances are independently and normally
distributed with mean zero and variance 𝜎 2 .
• The hypothesis being tested is called the null hypothesis
i.e. b = 0 against the alternative hypothesis that b < 0.
• We can use the estimated demand function for prediction
and policy purposes.
Sources of error term
• Unpredictable element of randomness in human
responses. People do not behave like a machine. In a
relationship between consumption expenditure and family
income, one month they will be on a spending spree, in
another month they will be frugal.
• Effects of omitted variables. Not just family income, other
factors influence consumption like size of family, tastes of
family, spending habits etc.
• Measurement error in the variables. Data may not be
accurate.
Goodness of Fit
• Regression results report R-squared statistic to tell how
closely the regression line fits the data.
• R2 measures the percentage of variations in the
dependent variable that is accounted for by all the
explanatory variables.
• Measures the overall goodness-of-fit of the multiple
regression equation.
• Ranges from 0 to 1.
• If R2 = 0.94, it means that the explanatory variables
explain 94% of the variations in the dependent variable.
Forecasting
• A forecast is a prediction about the values of the
dependent variable, given information about the
explanatory variables.
• We use regression models to predict the value of the
dependent variable beyond the time period over which the
model has been estimated.
Estimating elasticity
Given a linear demand for good X,
Qx= a – bPx + cM + dPy
• Own price elasticity: -b.(Px/Qx)
• Income elasticity: c.(M/Qx)
• Cross-price elasticity: d.(Py/Qx)
Log-linear demand curve
• With a log transformation of the demand equation, the
elasticity is constant along the entire demand curve called
isoelastic demand curve.
log(Qx) = a – b (logPx) + c(logM) + d(logPy)

Now, the parameter –b is the price elasticity, c is the


income elasticity and d the cross-price elasticity.

The natural logarithmic function with base e has the


∆𝑄
property that ∆(log 𝑄) = for any change in log(Q).
𝑄
Similarly for change in other log variables.
Using Excel to run a regression
To run a linear regression in Excel, the data analysis toolpak must be
installed. To verify if it is installed, click Data from the Excel 2010 menu. If
you see the Data Analysis command in the Analysis group (far right), the
data analysis toolpak is already installed.

If there is no Data Analysis item on the ribbon, follow the steps below if it
not already installed.
• Click the File menu option, and then click Options.
• Click Add-Ins, and then in the Manage box (near the bottom), select
Excel Add-ins.
• Click Go.
• In the Add-Ins available section, select the Analysis ToolPak check box,
and then click OK. If Analysis ToolPak is not listed in the Add-Ins
available box, click Browse to locate it. If you get prompted that the
Analysis ToolPak is not currently installed on your computer, click Yes to
install it.
• Click the Data tab and you will see the Data Analysis option on the
Analysis tab at the far right.
Bivariate Analysis
Sales and Advertising Expenditure of a Firm (Rs million)
Year Sales (Y) Advertising Expenditure (X)
2001 44 10
2002 40 9
2003 42 11
2004 46 12
2005 48 11
2006 52 12
2007 54 13
2008 58 13
2009 56 14
2010 60 15
Ordinary Least Squares
• Assume a linear relationship, Y= a + b X
We need to obtain the estimates of a and b i.e. intercept
and slope.
• The value of a is the value of Y when X=0.
• We need to know the co-efficient on X (Advertising) to find
out the responsiveness of Y (Sales) as X changes.
• The null hypothesis is b = 0.
• The estimated coefficient is said to be significant
depending on the t-ratio. Generally, use the 5% or 0,05
significance level is used. This means that the coefficient
is significantly different from zero.
Estimated Results
SUMMARY OUTPUT

Regression Statistics
Multiple R 0.922612
R Square 0.851212
Adjusted R
Square 0.832614
Standard Error 2.860653
Observations 10

Standard
Coefficients Error t Stat P-value
Intercept 7.6 6.332324 1.200191 0.264397
Advertising
Expenditure (X1) 3.533333 0.522281 6.765192 0.000143
Estimated regression equation
• Y= 7.60 + 3.53 X1
At X1=10, Y=7.60 + 3.53*10=42.90
At X1=15, Y=7.60 + 3.53*15=60.55

• The value of a=7.60

• The value of b measures the increase in the firm’s sales


revenues resulting from each unit increase of Rs 1 million
in the advertising expenditure of the firm. Thus, it
measures the marginal effect on Y (Sales) for each unit
change in X (Advertising).
Multivariate Analysis
Sales , Advertising and Quality Control Expenditure (Rs m)
Advertising Expenditure
Year Sales (Y) (X1) Quality Control (X2)
2001 44 10 3
2002 40 9 4
2003 42 11 3
2004 46 12 3
2005 48 11 4
2006 52 12 5
2007 54 13 6
2008 58 13 7
2009 56 14 7
2010 60 15 8
Estimated Results
SUMMARY OUTPUT

Regression Statistics
Multiple R 0.964445
R Square 0.930154
Adjusted R
Square 0.910198
Standard Error 2.095314
Observations 10

Standard
Coefficients Error t Stat P-value
Intercept 17.94366 5.919136 3.031467 0.019075
Advertising
Expenditure (X1) 1.873239 0.703339 2.66335 0.03231
Quality Control
(X2) 1.915493 0.681006 2.812742 0.026043
Estimated equation
• Y= 17.94 + 1.87X1 + 1.92X2

At X1=10 and X2=3,


Y=17.94 + 1.87*10 + 1.92*3=42.40

At X1=15 and X2=8,


Y=17.94 + 1.87*15 + 1.92*8=61.38

• You have to be very careful that the revenue equation is


correctly specified and all variables are taken into
account. If you omit important variables, you are likely to
get spurious results.
Example 1
Year Quantity Price Income

1995 4 24 10
1996 7 20 10
1997 8 17 10
1998 13 17 17
1999 16 10 17
2000 15 15 17
2001 19 12 20
2002 20 9 20
2003 22 5 20
Estimated results
SUMMARY OUTPUT

Regression Statistics
Multiple R 0.996189
R Square 0.992392
Adjusted R
Square 0.989856
Standard
Error 0.632558
Observations 9

Standard
Coefficients Error t Stat P-value
Intercept 8.074682 2.375689 3.39888 0.014516
X Variable 1 -0.4908 0.069976 -7.01378 0.000419
X Variable 2 0.813053 0.093154 8.728099 0.000125
Estimating elasticity
• What is the price elasticity in 1999?
• What is the price elasticity in 2003?
• What is the income elasticity in 1999?
• What will be the demand in 2004 if price rises to 7 and
income rises to 22?
• What will be the change in demand if price is expected to
rise by 5% and income is expected to rise by 10% in
2005?
Interpreting results
• Q = 8.08 – 0.49P + 0.81M
• Price e in 1999 = -0.49(10/16) = -0.31.
• Price e in 2003 = -0.49(22/5) = 0.11.
• Income elasticity in 1999 = 0.81(17/16) = 0.86
• Demand in 2004 = 8.08 - 0.49(7) + 0.81(22) = 22.47

• log(Q) = -0.23 – 0.34log(P) + 1.33log(M)


• Price e = -0.34
• Income e = 1.33
• Net change in demand in 2005 = -0.34(5%) + 1.33(10%)
= 11.6% increase
Are children inferior goods?
• In a developed country like Japan, income of the average
household has persistently increased over the past four
decades.
• The fertility rate of those households has been
decreasing.
• Conclusion: Children are inferior good!
Omitted variables
• You get data on number of children (Y) and per capita
income (X) and run a regression and find a negative
coefficient on income.
• This is a spurious result because of misspecification of
the equation.
• Other important factors are omitted like education levels
of households, percentage of working women, stress,
child-rearing expenses like day-care and school fees.
• It is a paradox because it represents a change in tastes
and preferences of women away from child-rearing
activities to being financially independent.
Estimating Demand
Regression analysis can answer questions like
• How will sales be affected if the firm doubles its
advertising expenditure?
• How much will revenues change after changing price by a
certain amount?
• How much will demand increase if consumers incomes
increase?
• How much will sales fall if rivals reduce their prices,
increase promotion outlays and provide credit incentives?
• What will be the effect of a merger?
Elasticity for food products
• Kumar, P., Kumar, A., Parappurathu, S. and Raju, S. S.,
“Estimation of Demand Elasticity for Food Commodities in
India”, Agricultural Economics Research Review 2011 Vol
24, pp 1–14.
• They find the commodity-wise annual per capita
consumption (kg/person/annum) of different food items
between 1983 and 2004 and find evidence of greater
diversification in food consumption.
Income elasticities
Moderately Middle High
Very poor poor income income All
Rice 0.182 0.102 0.03 -0.025 0.024
Wheat 0.102 0.083 0.07 0.071 0.075
Coarse cereals -0.123 -0.154 -0.141 -0.095 -0.125
Pulses 0.578 0.423 0.279 0.105 0.219
Milk 0.862 0.694 0.539 0.276 0.429
Edible oils 0.703 0.537 0.375 0.156 0.297
Vegetables 0.693 0.518 0.37 0.174 0.259
Fruits 0.753 0.599 0.429 0.282 0.362
Meat, fish and
eggs 1.034 0.9 0.799 0.531 0.669
Sugar 0.337 0.205 0.107 0.01 0.062
Other food
commodities 1.16 1.003 0.911 0.638 0.748
Non food
commodities 2.403 2.421 2.321 1.819 1.993
Price elasticities
Moderately Middle High
Very poor poor income income All
Rice -0.487 -0.377 -0.266 -0.161 -0.247
Wheat -0.48 -0.47 -0.3 -1.611 -0.340
Coarse cereals -0.333 -0.281 -0.196 -0.109 -0.194
Pulses -0.738 -0.667 -0.526 -0.339 -0.453
Milk -0.85 -0.81 -0.708 -0.521 -0.624
Edible oils -0.777 -0.708 -0.591 -0.381 -0.504
Vegetables -0.769 -0.73 -0.6 -0.453 -0.515
Fruits -0.824 -0.777 -0.682 -0.54 -0.595
Meat, fish and
eggs -0.908 -0.897 -0.864 -0.779 -0.821
Sugar -0.643 -0.58 -0.434 -0.255 -0.34
Other food
commodities -0.945 -0.942 -0.933 -0.906 -0.917
Non food
commodities -1.318 -1.28 -1.298 -1.184 -1.237

You might also like