Page 125 - Azerbaijan State University of Economics
P. 125

THE                 JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.82, # 2, 2025, pp. 117-137
                                                                             
                                                                                2
                                                                      2
                                  (   ,    , … ,    ) =    +     + ⋯ +    = ∑     →       
                                                            2
                                                     2
                                                                                 
                                                                        
                                                
                                                     1
                                                           2
                                    1
                                       2
                                                                            =1
                    must be performed using alternative techniques. Since the objective function     is
                    nonlinear  in  parameters,  applying  Fermat’s  theorem  becomes  impractical.  Taking
                    partial  derivatives  with  respect  to  the parameters,  setting  them equal  to  zero,  and
                    solving the resulting system of equations is often highly complex or even infeasible.
                    Thus,  the  minimization  problem  is  typically  addressed  using  approximate  and
                    numerical procedures.

                    Among  numerical  algorithms  for  minimizing      are  the  Marquardt  method  (a
                    modification of the Newton–Gauss algorithm), Powell’s version of the least squares
                    method,  the  hybrid  method,  and  others.  In  this  study,  we  employ  the  Marquardt
                    algorithm (Marquardt, 1963), which is recognized as a modification of the Newton–
                    Gauss method (Björck, 1996), Powell’s least squares approach, the hybrid method,
                    and the method introduced by Levenberg (Levenberg, 1944).

                    In  our  study,  Armington  and  CET  functions  were  evaluated  using  the  Marquardt
                    method based on a program specifically developed for this study in Mathcad (Hasanli,
                    Y.; Sadik-Zada, E.; Ismayilova, S.; Rahimli, G.; Ismayilova, F. (2023)).

                    DATA COLLECTİON AND PROCESSİNG
                    As mentioned, the Armington function enables the estimation of how imported and
                    domestically produced products substitute for one another. We have estimated the
                    Armington function for the Azerbaijan economy across the oil, non-oil and service
                    sectors. For this purpose, the dependence of total demand for the products of each
                    commodity or service group on the quantities of domestic and imported products in
                    that group was evaluated econometrically.

                    The  statistical  data  used  for  the  econometric  evaluation  were  collected  from  the
                    Balance of Payments and the System of National Accounts (SNA) for 2009–2021.
                    Accordingly, from the Balance of Payments tables, the imports of the oil and gas
                    sector were considered as the imports of the oil sector, while imports of other sectors
                    were considered as the imports of the non-oil sector. For the imports of the service
                    sector, the receipts from the balance of services in the Balance of Payments were used.
                    To determine the volume of domestic production sold on the local market, the value
                    of exports was subtracted from the total output of each sector. Export data were taken
                    from the Balance of Payments, while total output was obtained from the total output
                    tables  for  types  of  economic  activities  and  aggregated  across  the  three  sectors
                    considered in the study. Additionally, the indicators obtained from the Balance of
                    Payments were converted into manat using the exchange rate series for the respective
                    years. The total demand for each group of goods and services across the country is



                                                           125
   120   121   122   123   124   125   126   127   128   129   130