Scribd
Upload
6 views

Chapter 2A

Uploaded by

g.m57essa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
6 views

Chapter 2A

Uploaded by

g.m57essa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Functional Forms in

Regression
CH:2
Different Functional Forms
Linearity in Coefficients βi

OLS method is restricted to models that are linear in the Coefficients (Parameters)
Cobb-Douglas Production Function (Double log
function)

• CD production function is non-linear in parameters.


• It can be transformed into a linear model by taking natural logs of both sides

• By adding the error term ui , we obtain the following LRM

• (3) is known as double-log, log-log or constant elasticity model because


both the regressand and regressors are in the log from
• The slope coefficients can be interpreted as partial elasticities β2 and β3
(holding other variables constant)
• Returns to scale of CD function:
Regression of CD Function
• Using cross section data Output, Labor and Capital for 51 states for 2005
• output [value added, thousands of dollars]
• labor [worker hours, in thousands]
• capital [capital expenditure, in thousands of dollars]
Hypothesis testing & goodness of fit CD function:
• all regression coefficients (i.e., elasticities) are individually statistically highly
significant (quite low p values)
• according to F - statistic collectively both factors inputs [labor and capital] are
statistically significant
• quite high R2 [unusual for cross-section data!]
• if we increase labor input by 1%, on average, output goes up by about 0.47%
[holding the capital input constant]
• if we increase the capital input by 1%, on average, the output increases by about
0.52% [holding the labor input constant]
• (β2 + β3 = 0.9896 ≈ 1)it means that a constant returns to scale in 2005
• The original production function:
(Note: 48.79 is approximately the anti-log of 3.8876).
Log-lin or growth models
• measure the rate of growth of real GDP (i.e. GDP adjusted for
inflation) for the USA for the period 1960–2007

• where RGDP stands for real GDP, r is the rate of growth, and t is time
measured chronologically.
• Taking the natural log of both sides of Eq. (2.12), we obtain:
• Equation (2.15) is like any other regression model; the only difference
is that here the regressor is “time”, which takes values of 1, 2, …, 47.
• It is called a semilog model because only one variable (in this case
the regressand) appears in the logarithmic form, whereas the
regressor (time here) is in the level or linear form.
• For descriptive purposes we can call (2.15) a log-lin model .
• It can be estimated by the usual OLS process.
• B2 the slope coefficient and B1 is the intercept coefficient.
Interpretation of the results

• The results show that over the period of 1960–2007 the USA’s real GDP had been
increasing at the rate of 3.15% per year.
• This growth rate is statistically significant, for the estimated t value of about
90.82 is highly significant.
• Taking the anti-log of the intersect C anti-log (7.8756) = 2632.27, this is the
beginning value of real GDP at 1960 $2501.8 billion.
• The rate of growth, r can be computed from
• r = anti-log( B2) –1. Now anti-log ( B2) = 1.03199, Therefore the compound rate of
growth is 0.03199 or about 3.2%

You might also like