Course Name - Econometrics
Course No. – AGECON – 507
Prepared by – Dr. Payal Jaiswal
UNIT IV
Linear Growth
Linear growth refers to a change in size that proceeds at the same rate over time. If growth is
plotted in a diagram and it resembles a straight line, this is called linear growth. Additive
processes produce linear growth. Additive processes occur when the same amount of growth
is added to a system during each time period. If 100 people moved into a small town every
year, and you graphed the population of the town, the graph would look like a straight line
going upward (with positive slope).
Linear growth means that it grows by the same amount in each time step. For example you
might have something that is 5 inches long on Monday morning and then 8 inches long on
Tuesday morning and then 11 inches long on Wednesday morning and so on. So it is growing
by 3 inches a day. If you draw a graph with the days on the horizontal axis and the height in
inches on the vertical axis then the points lie on a line. This is why it is called linear growth.
A line graph has an x-axis and a y-axis. The y-axis is the vertical axis labeled with the variable
being measured. The x-axis is the horizontal axis labeled with the variable that influences the
variable being measured. When you plot any data point, you create an x,y coordinate. The slope
of a line, and therefore linear growth, is calculated using two coordinates: (x1, y1) and (x2, y2).
The formula for calculating slope is:
slope = (y2 - y1) / (x2 - x1)
Compound annual growth rate
Compound Annual Growth Rate (or CAGR) is a widely used measure of growth. It is used to
evaluate anything that can fluctuate in value, such as assets and investments. CAGR takes the
initial investment value and projects an ending investment value while assuming compound
growth over a set period of time.
CAGR is a business and investing specific term for the geometric progression ratio that
provides a constant rate of return over the time period. CAGR is not an accounting term, but it
is often used to describe some element of the business, for example revenue, units delivered,
registered users, etc. CAGR dampens the effect of volatility of periodic returns that can render
arithmetic means irrelevant. It is particularly useful to compare growth rates from various data
sets of common domain such as revenue growth of companies in the same industry or sector.
The Compound Annual Growth Rates are estimated by using the following formula:
The exponential compound annual growth rates are estimated by using log linear functions on
the time series data on area, production and productivity for any commodity. The semi log
exponential functional form is used to analyze the trend in growth rate, which is one of the
appropriate functional forms to estimate the growth rate. That is, the growth rate is estimated
by using the following semi log functional form:
logYt = a + bt............................. (1)
This equation (1) can be elaborated in details as:
Yt= Yo (1+r) t ........................ (i)
Taking log on both sides, we get
Log Yt = Log Yo + t Log (1+r)................. (ii)
Equation (ii) can be rewrite as Y = a + bt............................. (iii)
Where Y= Log Yt ; a = Log Yo ; b= Log (1+r),
In equation (iii)
Yt= area/production/ productivity of commodity, as discussed above
a= constant
t= Time variable in year (1, 2,...n)
b= Regression Coefficient that shows the rate of change or growth rates in a series.
The annual compound growth rate (s) can be worked out by using:
Antilog (b) =Antilog (log (1+r)).
Antilog (b) =1+r
and
r = Antilog b-1
When multiplied by 100, it gives the percentage growth rate in area, production and
productivity of commodity. That is, Compound Annual Growth Rate (CAGR) (%) = r =
(Antilog b -1) x100.
Demand Function:
The quantity of a commodity to be demanded [purchased] by consumers depends mainly on its
own price, cross price [prices of other commodities such as complementary and substitutes]
and income. These factors are referred to as determinants.
Functional relationship between the quantity of commodity demanded and its determinants can
be shown as follows:
Y = f (X1, X2, X3)
Where,
Y = Quantity demanded [purchased]
X1 = Price of the commodity
X2 = Prices of other commodities [complementary/substitutes]
X3 = Income
There will always be an inverse relationship between Y and X1; and a positive relationship
between Y and X2. If it is a substitute and an inverse relationship if it is a complementary
commodity. In the case of normal commodities, there will be a positive relationship between
Y and X3. In the case of inferior commodities, there will be an inverse relationship between Y
and X3. Since four variables are considered in the demand function, the elasticities of Y with
respect to X1, X2 and X3 have been explained.
Linear Demand Function:
The specification of linear demand function for a commodity will be as follows:
Y = b0 + b1X1 + b2X2 + b3X3+ error………………..
Where
Y = Quantity demanded [purchased]
X1 = Prices of a commodity
X2 = Prices of other commodity [complementary/substitutes]
X3= Income
The partial derivative of Y with respect to X1 [marginal effect], keeping X2 and X3 constant,
is ∂ Y/∂ X1 = b1 < 0
The partial derivative of Y with respect to X2, keeping X1 and X3 constant, is
∂ Y / ∂ X2 = b2 ≥ 0
If b2 > 0, then the commodity is a substitutes.
If b2 < 0, then the commodities is a complementary.
The partial derivative of Y with respect to X3, keeping X1 and X3 constant, is
∂ Y / ∂ X3 = b3 ≥ 0
If b3 > 0; then commodity under consideration will be normal; If b3 < 0; then the commodity
under consideration will be inferior.
The partial elasticity of Y with respect to X1, keeping X2 and X3 constant, is
eY.X1 = ∂ Y/∂X1 . X1/Y = b1. X1/Y < 0.
The partial elasticity of Y with respect to X2, keeping X1 and X3 constant, is
eY.X2 = ∂ Y/∂X2 . X2/Y = b2. X2/Y ≥ 0
The partial elasticity of Y with respect to X3, keeping X1 and X2 constant, is
eY.X3 = ∂ Y/ ∂X3. X3 /Y = b3. Y /X3 ≥ 0
In estimating the partial elasticities, one can use either total of X1, X2, X3 and Y or mean value
of X1, X2, X3 and Y or individual values of X1i, X2i, X3i and Yi: In the empirical studies
mean values will be considered to evaluate the price, cross price and income elasticities of
demand. Therefore these elasticities will be referred to as average own price, cross price and
income elasticities. The values of b0 b1, b2 and b3 will be estimated by OLS method with
specific assumptions relating to the distribution of random variable, independent variables and
dependent variable.
The reliability of the above estimates of b0, b1, b2 and b3 can as usual be assessed by taking
the values of standard errors or Y values. If the standard errors are smaller than half of the
regression coefficients of X1, X2 and X3 then the estimates are deemed to be reliable estimates
i.e., X1, X2, X3 variables do have significant impact on the demand for the commodity.
If it is more than half of the regression coefficients, then the estimates of parameters are deemed
to be unreliable estimates i.e., X1, X2 and X3 do not have significant impact on the demand
for the commodity. Similarly, the value of ‘t’ test can be considered to assess the reliability of
estimates of the parameters. If the values of ‘t’ are very high, then the estimates are considered
to be statistically significant.
If they are smaller, then the estimates are considered to be statistically insignificant. Thus, on
the basis of either standard errors or ‘t’ values, the reliability of estimates can be assessed.
Similarly, the overall goodness of the linear equation fitted to time series data can be assessed
by taking the values of both R2 and Adj R2.
If the value of these statistics are close to one, then the linear regression model [equation] fitted
to the observations [data points] will be considered good. If the value of these statistics are
close to zero, then the linear regression model fitted to the observations will be considered not
good.
Thus, the values of R2 and Adj R2 provide us an idea about the extent of variation explained by
the independent variables included in the model in total variation.
Log Linear Demand Function:
If the linear regression model [Linear demand function] is found to be not suitable to the data
points, then other forms of equations have to be attempted. The most popular form of regression
equation is log linear [double-log or Power or constant elasticity model] regression model or
log-linear demand function, whose specification will be as follows:
log Y = log b0 + b1 log X1 + b2 log X2 + b3 log X3 + error
The partial derivative of log Y with respect to log X1 , keeping other variables constant, is
∂ log Y/∂ log X1 = ∂ Y/Y/∂ X1/X1 = b1 which will be less than zero
Similarly, the partial derivative of log Y with respect to log X2, keeping other variables
constant, is
∂ log Y / ∂ log X2 = ∂ Y / Y / ∂ X2/ X2 = ∂ Y / Y / ∂ X2/ X2 = b2 which will be less than or
more than zero
The partial derivative of log Y with respect to log X3, keeping other variables constant, is
∂ log Y / ∂ log X3 = ∂ Y / Y / ∂ X3/ X3 = ∂ Y / Y.X3 / ∂ X3 = b3 which will be less than or
more than zero
b1 is known as the price elasticity of demand. If its value is more than unity, then the
proportionate change in demand will be higher than the proportionate change in price of the
commodity showing elastic nature of the commodity to the changes in own prices, all else
equal. If its value is less than unity then the proportionate change in demand will be smaller
than the proportionate change in own price of the commodity, showing inelastic nature of the
commodity to the changes in prices of the same commodity, all else equal.
If its value is unity then the proportionate change in demand will be equal to the proportionate
change in price of the commodity. If the value of b2, the cross price elasticity of demand, is
more than unity then the proportionate change in demand is higher than the proportionate
change in prices of other commodities [which may be either complementary or substitutes]; If
it’s value is less than unity then the proportionate change in demand will be less than the
proportionate change in prices of other commodities.
If it is equal to unity then the proportionate change in demand will be equal to the proportionate
change in prices of other commodities. If the sign of cross price elasticity of demand is negative
then the commodity under consideration will be complementary. If the sign of cross price
elasticity of demand is positive then the commodity under consideration will be a substitute.
Thus, on the basis of sign of the cross price elasticity of demand, the nature of commodity may
be identified as complementary or substitute. If the value of b3, the income elasticity of
demand, is more than unity then the proportionate change in demand will be higher then the
proportionate change in income showing the elastic nature of the commodity. If its value is less
than unity then the proportionate change in demand will be smaller than the proportionate
change in income showing the relatively inelastic nature of the commodity.
If its value is equal to unity then the proportionate change in demand will be equal to the
proportionate change in income showing the unitary elastic nature of commodity. If the sign of
income elasticity of demand is positive then the commodity under consideration will be
considered normal. If the sign of income elasticity of demand is negative then the commodity
under consideration will be considered as inferior to other commodities.
Thus, on the basis of sign of the income elasticity of demand the commodities can be classified
into normal and inferior. From the linear demand function, own price, cross price and income
elasticities of demand will be evaluated at the mean values of X1, X2, X3 and Y. Therefore,
these elasticities are referred to as average price, cross price and income elasticities of demand.
In the case of linear demand function the elasticities go on changing from point to point, if the
elasticities are evaluated at different values of Y, X1, X2 and X3.
In case of double log demand function the elasticities will be constant. The ultimate choice of
the model for demand function will be based on the value of adjusted R2, the statistical
significance and sign of the regression coefficients. In brief the choice of the results of the
demand function for the analysis, policy making and forecasting will be based on economic,
statistical and econometric criteria.
Demand Function for the Commodities–Estimates of Economic Relationships:
The following hypothetical data [Table 4.1] on physical quantity of the commodity demanded
[Y], its own price [X1],cross price (prices of other commodities)[X2] and income [X3] are
considered for estimating both the linear and log linear demand functions.
The results of the linear demand function [Table 4.2] show- that the regression coefficients of
own price [X1], cross price [X2] and income [X3] are statistically significant. The own price,
cross price and income elasticities of demand for the commodity [estimated at the mean values]
are -0.529 [-3.345 (19.19/121.19)], 0.249 [1.6836 (17.90476/121.1905)] and – 0.0650 [-
0.005478* (1439.57/121.1905) respectively.
The sign of cross elasticity of demand is negative showing that the commodities under
consideration are complimentary. The sign of the income elasticity of demand is also negative
showing that the commodity under consideration is not normal. The price elasticity of demand
is negative and less than unity showing the inelastic nature of the commodity to the changes in
prices of the same commodity, all else equal.
The regression results of the log linear demand function [Table 4.4] based on the data given in
table 4.3 show that the regression coefficient of cross price [X2] is positive but not significant.
The regression coefficient of income [X3] is significantly negative and the regression
coefficient of own price [X1] is negative at low level of confidence.
It should be noted that the choice [linear or log linear form] of the results of the demand
function depends on economic, statistical and econometric norms. In empirical studies, both
the short run and long run demand functions [to estimate both the short run and long run price,
cross and income elasticities of demand] for different commodities are estimated through the
Nerlovian’s Partial Adjustment Mechanism using time services data.
Time series family budget data will be used for estimating demand functions and demand
elasticities. The budget data [concerned at a point of time] will however be not sufficient to
estimate the price elasticities. In view of this, the price and cross price elasticities of demand
for poultry in Mymensingh town, Bangladesh were estimated using panel data.The estimates
of the study are given in Table 4.5.
The signs of the coefficients of the independent variables are positive showing that they are
substitutes. Further the commodity beef was found to be close substitute as the elasticity value
exceeded unity.
The price elasticity of the commodity, poultry is significantly negative evincing the fact that a
one percent increase in the price of that commodity would reduce the demand for that
commodity by more than three percent. It should be noted that the estimation of own price,
cross price and income elasticities of demand for the commodities is not easy if the data points
on the variables are not available in required form.
Dummy Variable
In statistics and econometrics, particularly in regression analysis, a dummy variable is one that
takes only the value 0 or 1 to indicate the absence or presence of some categorical effect that
may be expected to shift the outcome. They can be thought of as numeric stand-ins
for qualitative facts in a regression model, sorting data into mutually
exclusive categories. Dummy variables are used frequently in time series analysis with regime
switching, seasonal analysis and qualitative data applications. Dummy variables are useful
because they enable us to use a single regression equation to represent multiple groups. This
means that we don't need to write out separate equation models for each subgroup. The dummy
variables act like 'switches' that turn various parameters on and off in an equation.
Dummy variables play an important role in the analysis of data, whether they are real-valued
variables, categorical data. The extreme case of representing all the variables (independent and
dependent) as dummy variables provides a high degree of flexibility in selecting a modelling
methodology. In addition to this benefit of flexibility, the elementary statistics (e. g., mean and
standard deviation) for dummy variables have interpretations for probabilistic reasoning,
information theory, set relations, and symbolic logic.
Whether the analytical technique is traditional or experimental, highly complex information
structures can be represented by dummy variables. Examples presented included multiple
regimes, business behaviour, and dynamical systems. There are no hard boundaries between
the relationships of dummy variables in quantitative analysis, sets and logic, and the computer
science concept of data representation in bits. The intelligent use of dummy variables usually
makes the resulting application easier to implement, use, and interpret.
Lagged Variable
A lagged variable is a variable which contains a number of past values of that variable. In
economics the dependence of a variable Y (dependent variable) on another variables(s) X
(explanatory variable) is rarely instantaneous. Vary often, Y responds to X with a lapse of time.
Such a lapse of time is called a lag. A possible source of any problem with the functional form
is the lack of lagged structure in a model. One way of overcoming autocorrelation is to add a
lagged dependent variable to the model. However although lagged variable can produce a better
functional form, we need theoretical reasons for including them.
Proxy Variable
Proxy variable is a variable that is not in itself directly relevant, but that serves in place of an
unobservable or immeasurable variable. In order for a variable to be a good proxy, it must have
a close correlation, not necessarily linear, with the variable of interest. This correlation might
be either positive or negative. Proxy variable must relate to unobserved variable, must correlate
with disturbance, and must not correlate with regressor once disturbance is controlled for.
Example:- In social sciences, proxy measurements are often required to stand in for variables
that cannot be directly measured. This process of standing in is also known
as operationalization. Per-capita GDP is often used as a proxy for measures of standard of
living or quality of life.
Endogenous and exogenous Variables
In an economic model, an exogenous variable is one whose value is determined outside the
model and is imposed on the model, and an exogenous change is a change in an exogenous
variable. In contrast, an endogenous variable is a variable whose value is determined by the
model. An endogenous change is a change in an endogenous variable in response to an
exogenous change that is imposed upon the model. In econometrics, an exogenous variable is
assumed to be fixed in repeated sampling, which means it is a nonstochastic variable. An
implication of this assumption is that the error term in the econometric model is independent
of the exogenous variable.
Examples:
In the simple supply and demand model, a change in consumer tastes is unexplained by the
model and imposes an exogenous change in demand that leads to a change in the endogenous
equilibrium price and the endogenous equilibrium quantity transacted. Here the exogenous
variable is a parameter conveying consumer tastes. Similarly, a change in the consumer's
income is exogenously given, outside the model, and appears in the model as an exogenous
change in demand. In the LM model of interest rate determination, the supply of and demand
for money determine the interest rate contingent on the level of the money supply, so the money
supply is an exogenous variable and the interest rate is an endogenous variable. In a model of
firm behavior with competitive input markets, the prices of inputs are exogenously given, and
the amounts of the inputs to use are endogenous.
Models and sub models
An economic variable can be exogenous in some models and endogenous in others. In
particular this can happen when one model also serves as a component of a broader model. For
example, the IS model of only the goods market derives the market-clearing (and thus
endogenous) level of output depending on the exogenously imposed level of interest rates,
since interest rates affect the physical investment component of the demand for goods. In
contrast, the LM model of only the money market takes income (which identically equals
output) as exogenously given and affecting money demand; here equilibrium of money supply
and money demand endogenously determines the interest rate. But when the IS model and the
LM model are combined to give the IS-LM model, both the interest rate and output are
endogenously determined.
