Disaggregation using Insee's method
[yhf,retal] = etalcalinsee(namey,arg1, ,argn)
namey= low frequency series
arg1, ,argn= strings that can be
.'alpha =xx' where w is significance level for the constant default 0.1, that is 10%)
.'ny_prov =n' where n is the number of known annual data that are ignored in the diseggregation process
.'mod=''rw''' if the residuals are supposed to be a RW or 'mod=''ar''' if the residuals are supposed to be an AR(1) (default: model with maximum likelihood is chosen)
.'bench=''xxx''' where xxx is the name of the series the user wants to compare the result with (for instance to compare the method with another one)
.'noprint' if the user does not want to print the results
yhf= the disaggregated series
retal= a results tlist with:
- retal('meth') = 'Insee''s disaggregation'
- retal('y annu') = annual series
- retal('y estim') = annual transformed series (that is differentiated if residuals follow a RW, levle if not)
- retal('x annu') = a (N x 1) vector of annual data
- retal('x estim') = a (N x 1) vector of transformed annual data
- retal('x trim') = a (n x k) matrix of exogenous high frequency indicators
- retal('nobs') = # of observations in the regression
- retal('resid estim') = a (n x 1) vector of regression residuals
- retal('beta') = estimated coefficients of the annual regression
- retal('tstat') = t stats
- retal('pvalue') = pvalue of the betas
- retal('prescte') = boolean indicating the presence or absence of a constant in the regression
- retal('llike') = the log-likelihood
- retal('rho') = autcorrelation coefficient of residuals:
. if rho = ]-1,1[ then the model is estimated in levels
. if rho = 1 then the model is estimated in differences
- retal('trend') = trend
. retal('trend') = 1 if retal('prescte') = %t and rho = 1
. retal('trend') = 2 if retal('prescte') = %t and rho ~= 1
. retal('trend') = 0 in other cases
- retal('prov y annu') = provisional value for y (and therefore not used in estimation)
- retal('x annu aug') = (N x l) matrix of regressors, including the constant or trend if necessary
- retal('x trim aug') = (n x l) matrix of regressors, including the constant or trend if necessary
- retal('yhat annu') = (N x 1) adjusted annual y
- retal('resid annu') = (N x 1)annual residual
- retal('resid annu report') = annual residual extended to the incomplete year with the estimated model
- retal('prov resid annu') = annual residual extended using the last residual ("provisional year")
- retal('high freq yhat') = high frequency adjusted y
- retal('prov high freq yhat') = high frequency adjusted y extended using the last residuals ("provisional years")
- retal('high freq resid') = high frequency residual in TS form
- retal('prov high freq resid') = high frequency residual extended using the last residuals ("provisional years") in TS form
- retal('prests') = boolean indicating the presence or absence of a time series in the regression
- retal('namey') = name of the y variable
- retal('namex') = name of the x variables
bounds() load(GROCERDIR+'\data\EtalCalInsee.dat'); [yhf,retal] = etalcalinsee('consameu','cammeu','bench=''consmmeu''','s=12','alpha=0.15') ; | ![]() | ![]() |