Forum: help

RE: micEconSNQP [ Reply ] By: Daniel Muluwork Atsbeha on 2015-02-27 12:36	[forum:41952]
Dear Arne, Can you suggest a way to cite micEconSNQP? The following is what I extracted from the documentation of the package: Henningsen, A. 2014. Production analysis with the Symmetric Normalized Quadratic (SNQ) profit function in R: Package: micEconSNQP: Institution? Thanks. Daniel

RE: micEconSNQP [ Reply ] By: Daniel Atsbeha on 2014-01-06 19:46	[forum:40277]
Dear Arne, Happy New Year!! Many thanks again for your answer. Best regards, Daniel

RE: micEconSNQP [ Reply ]
By: Arne Henningsen on 2013-12-29 19:21

[forum:40239]

Dear Daniel

You are right that the derived netput equations are homogeneous of degree zero, meaning that only relative prices matter. Thus, if you change *all* prices (or quantities) by the *same* factor (e.g. converting all prices from US dollars to Canadian dollars), the elasticities remain unchanged. However, if you change the units of measurement of a *single* input (e.g. converting the labour input from working hours to working days), the elasticities will change.

Best regards,
Arne

RE: micEconSNQP [ Reply ]
By: Daniel Atsbeha on 2013-12-17 14:38

[forum:40201]

Dear Arne,
Again, I have one question about the symmetric normalized quadratic profit function and micEconSNQP.

The help file for micEconSNQP on page 12 says "please note that the SNQ Profit function is not invariant to units of measurement so that different scaling factors result in different estimates of elasticities". Using different scaling factors in the example codes also produced different elasticity estimates.

However, if the derived netput equations are homogeneous of degree zero, meaning that only relative prices matter, I wonder why units of measurements matter in micEconSNQP. In a latter paper Kohli (1994) also says measurement units don’t matter on footnote 7 of page 589. Do you know why micEconSNQP behave sensitively to measurement units? I will appreciate any clues. Thanks.

Reference Kohli, U. 1994. Canadian Imports and Exports by Origin and Destination: A Semi-Flexible Approach. Canadian Journal of Economics 27(3): 580 - 603.

Daniel

RE: micEconSNQP [ Reply ] By: Barbara Hutniczak on 2013-03-04 03:35	[forum:39032]
Thanks for the answer, I'll do my best! Best regards Barbara Hutniczak

RE: micEconSNQP [ Reply ]
By: Arne Henningsen on 2013-03-01 23:11

[forum:38994]

Dear Barbara

I did not implement the imposition of concavity in the quasi-fixed inputs. I am not sure whether it is possible to simultaneously impose convexity in prices and concavity in the quasi-fixed inputs within a single (second) step by using the current method. As I am currently not working with SNQP profit functions, I have no plans to implement this feature. However, you are more than welcome to implement this feature yourself.

Best regards,
Arne

RE: micEconSNQP [ Reply ] By: Barbara Hutniczak on 2013-02-24 08:08	[forum:38940]
I have a further question regarding imposing constraints on the SNQP function. Is there any way I could impose concavity on fixed factor coefficient matrix?

RE: micEconSNQP [ Reply ] By: Daniel Muluwork Atsbeha on 2012-01-30 14:47	[forum:5502]
Thats great !! Thanks Arne. Best regards, Daniel

RE: micEconSNQP [ Reply ]
By: Arne Henningsen on 2012-01-28 05:41

[forum:5494]

Dear Daniel

Yes, there is a way to constrain the alpha_i parameters to be zero. You can use argument "restrict.matrix", which snqProfitEst() passes to systemfit(). Please note that snqProfitEst() uses argument "restrict.regMat" of systemfit() to impose the cross-equation restrictions and that "restrict.matrix" is applied to the parameters after applying "restrict.regMat". You can obtain the coefficients that are internally estimated after applying "restrict.regMat" by following command:

coef( estResult$est, modified.regMat = TRUE )

where "estResult" is the object returned by snqProfitEst(). If you have n netputs (n equations), the first n coefficients of this vector correspond to the n intercepts (alpha_1, ..., alpha_n).

The following code first estimates the model that is used in the example section of the documentation of snqProfitEst() and then re-estimates this model with all intercepts (alpha_1, alpha_2, alpha_3) restricted to be zero. Finally, a likelihood ratio test is used to test the restriction. Of course, restricting all alpha_i to be zero does not make sense without the within transformation (as in this example). Therefore, it is not surprising that the restriction is rejected in the example below.

library( "micEconSNQP" )
data( germanFarms )
germanFarms$qOutput <- germanFarms$vOutput / germanFarms$pOutput
germanFarms$qVarInput <- -germanFarms$vVarInput / germanFarms$pVarInput
germanFarms$qLabor <- -germanFarms$qLabor
priceNames <- c( "pOutput", "pVarInput", "pLabor" )
quantNames <- c( "qOutput", "qVarInput", "qLabor" )
germanFarms$time <- c( 0:19 )

# estimate unrestricted model
estResult <- snqProfitEst( priceNames, quantNames, fixNames = c("land","time"),
data = germanFarms )
estResult

# create restriction matrix for imposing three restrictions
rMat <- matrix( 0, nrow = 3,
ncol = length( coef( estResult$est, modified.regMat = TRUE ) ))
# first restriction: alpha_1 = 0
rMat[ 1, 1 ] <- 1
# second restriction: alpha_2 = 0
rMat[ 2, 2 ] <- 1
# third restriction: alpha_3 = 0
rMat[ 3, 3 ] <- 1

# estimate restricted model
estResult2 <- snqProfitEst( priceNames, quantNames, fixNames = c("land","time"),
data = germanFarms, restrict.matrix = rMat )
estResult2

# test restriction with an LR test
lrtest( estResult$est, estResult2$est )

micEconSNQP [ Reply ]
By: Daniel Muluwork Atsbeha on 2012-01-27 17:04

[forum:5490]

Hello R users,

I have one question about micEconSNQP. Is there a way to constrain the alpha parameters (i.e. the constants in the netput equations) to zero?
I am interested in this because I have panel data on which the within transformation is applied to avoid estimation of farm fixed effects.
Given the transformation, I need to constrain the alpha parameters into zero. Is there a way to do this?

Thank you in advance.

Daniel