Title of the paper - Download as PowerPoint by R5Gx2o

VIEWS: 4 PAGES: 19

									Impact Assessment of the Non-
 Tariff Measures (NTM) Upon
  International Lemon Trade
           Daniel Lema, Juan Santini, Ciro Tapia, Daniel
                  Iglesias, and Graciela Ghezan

         Instituto Nacional de Tecnología Agropecuaria
                              (INTA)




VIII International Agribusiness PAA-PENSA Conference - Buenos Aires, Argentina. November 30 - December 2, 2011
          Outline

   Motivation
   Objectives
   Methodology
   Results
   Conclusions
                       Motivation

 Collaborative research project co-financed by the
  European Commission under its 7th Framework
  Program
 ANALYZING THE EFFECTS FROM NON TARIFF
  MEASURES (NTM) IN THE GLOBAL AGRI-FOOD
  TRADE

 FP7 - NTM-IMPACT PROJECT
 http://www.ntm-impact.eu
                   Objectives

 We study the potential effects of the sanitary
    NTM on the lemon trade and on the relative
    performance of Spain and Argentina.
   We analyze the Lemon exports from
    Argentina, Spain, USA, Mexico, South Africa
    and Turkey to their principal trade partners
    for the period 1995-2005.
   SPS measures:
   Medfly
   Citrus canker
Figure 1 : Share in the world production, processing and export of lemon by country
(2000-2006), in %
 80

 70

 60

 50

 40

 30

 20

 10

  0
      Total NH   Total SH    Argentine        Spain           US           Italy   Turkey

                    Production Share     Industrialization Share   Export Share

Source: USDA
Figure 4: EU imports by country (2008)

                 Others
      Brazil      11%
       9%



  Turkey
   13%

      South Africa                       Argentina
         14%                               53%


Source: UN COMTRADE
Figure 5 : Lemon trade flows


                                       Russia
                               Spain
             US




                  Argentine
             Methodology

 Lemon exports from Argentina, Spain,
  USA, Mexico, South Africa and Turkey
  to their principal trade partners The
  data were collected from UNCTAD
  Comtrade database
 Time Frame: 1995-2005.
 Gravity Models
 Estimated using Poisson Pseudo
  Maximum-Likelihood (PPML) method
              Methodology

 Silva and Tenreyro propose the use of
  Poisson Pseudo Maximum-Likelihood
  (PPML) method for estimating gravity
  models for trade.
 The estimates from this method are
  robust, even with different patterns of
  heteroskedasticity.
 In addition to this, this model is
  compatible with the existence of zeros
  in trade data.
 Dependent Variable:
 Lemon exports from Argentina, Spain, USA, Mexico, South
    Africa and Turkey to their principal trade partners. There are
    1089 observations, of which 80% are nonzero.

 Indepent Variables:
 Consumption of lemon of trade partners. (FAOSTAT 2010) for
    the period 1995-2005.
   GDP (current US$) of trade partners (World Bank 2010),
   Tariffs of trade partners – UNCTAD 2010 and Tariff Data (WTO
    2010)
   Border Effects (Centre d’Etudes Prospectives et d’Informations
    Internationales CEPII 2010). The variable takes 1 if the countries
    are contiguous, 0 otherwise.
   Common Languages (CEPII 2010). The variable takes 1 if the
    countries share a common language, 0 otherwise.
   Weighted Distances between the biggest cities of a pair of
    countries (CEPII 2010). (CEPII 2010).
   Real Exchange Rate (indextcr), between pairs of trading partners
 The SPS measure is a dummy variable
 Takes value=1 if the importer country
    imposed a SPS measure (medfly, citrus
    canker) to the exporter country and 0
    otherwise.
   Cases:
   Argentina
   Spain
   All countries
                  Results
 The results regarding the SPS measures
  effects show differences for the global
  case and the two specific cases studied.
 While SPS measures have a negative and
  significant (marginally) effect on global
  trade
 A negative but non significant effect was
  found over Argentine and Spain exports.
                                    Results
Table 3: PPML Estimates: Global Model (All trading countries)
Random-effects Poisson regression               Number of obs      =     1089
Group variable: id                              Number of groups   =       99

Random effects u_i ~ Gamma                      Obs per group: min =       11
                                                               avg =     11.0
                                                               max =       11

                                                Wald chi2(8)       =    29.51
Log likelihood   = -3287.3153                   Prob > chi2        =   0.0003

------------------------------------------------------------------------------
             |   Observed                                    Normal-based
      lnxpos |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      lncons |   .0257268   .0229057     1.12   0.261    -.0191674    .0706211
     lndistw | -.3139063    .0767754    -4.09   0.000    -.4643833   -.1634293
      lnpbim |   .2119473   .0514005     4.12   0.000     .1112042    .3126904
         tax |   .3599841    .299475     1.20   0.229    -.2269762    .9469443
  lnindextcr |   .0330763   .0120095     2.75   0.006     .0095381    .0566145
    frontera | -.1843426    .1474891    -1.25   0.211     -.473416    .1047307
      idioma |   .3754911   .2233703     1.68   0.093    -.0623067    .8132889
         ntm | -.1534569    .0966688    -1.59   0.112    -.3429243    .0360106
    constant |   3.609511   .4948693     7.29   0.000     2.639585    4.579437
-------------+----------------------------------------------------------------
    /lnalpha | -.9036451    .2704514                      -1.43372   -.3735701
-------------+----------------------------------------------------------------
       alpha |   .4050904   .1095573                      .2384203    .6882727
------------------------------------------------------------------------------
Likelihood-ratio test of alpha=0: chibar2(01) = 1958.27 Prob>=chibar2 = 0.000
                                     Results
Table 4: PPML Estimates:Argentina
Random-effects Poisson regression              Number of obs      =       209
Group variable: id                             Number of groups   =        19

Random effects u_i ~ Gamma                     Obs per group: min =        11
                                                              avg =      11.0
                                                              max =        11

                                               Wald chi2(7)       =     36.59
Log likelihood   = -767.19983                  Prob > chi2        =    0.0000

------------------------------------------------------------------------------
      lnxpos |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      lncons |   .0460329   .0616945     0.75   0.456    -.0748862     .166952
     lndistw |   -1.05402   1.102607    -0.96   0.339    -3.215091    1.107051
      lnpbim |   .2007417   .0965483     2.08   0.038     .0115106    .3899729
         tax | -.8077387    1.981838    -0.41   0.684    -4.692069    3.076591
  lnindextcr |   .0527523   .0138129     3.82   0.000     .0256795    .0798251
      idioma |   .2480285    .657186     0.38   0.706    -1.040032    1.536089
         ntm | -.2058255    .2054988    -1.00   0.317    -.6085957    .1969448
    constant |   10.69892   10.36368     1.03   0.302    -9.613526    31.01136
-------------+----------------------------------------------------------------
    /lnalpha | -1.019858    .3383727                     -1.683056   -.3566592
-------------+----------------------------------------------------------------
       alpha |   .3606463   .1220329                      .1858053     .700011
------------------------------------------------------------------------------
Likelihood-ratio test of alpha=0: chibar2(01) =   405.32 Prob>=chibar2 = 0.000
                                     Results
Table 4: PPML Estimates: Spain
Random-effects Poisson regression              Number of obs      =       209
Group variable: id                             Number of groups   =        19

Random effects u_i ~ Gamma                     Obs per group: min =        11
                                                              avg =      11.0
                                                              max =        11

                                               Wald chi2(7)       =     44.79
Log likelihood   = -536.97619                  Prob > chi2        =    0.0000

------------------------------------------------------------------------------
      lnxpos |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      lncons |    .026651   .0246962     1.08   0.281    -.0217526    .0750545
     lndistw | -.3370508    .0534084    -6.31   0.000    -.4417294   -.2323722
      lnpbim |   .0083561   .0320527     0.26   0.794    -.0544661    .0711782
         tax |   5.599403   2.193782     2.55   0.011     1.299669    9.899137
  lnindextcr | -.0095812    .0086122    -1.11   0.266    -.0264608    .0072983
    frontera | -.2026875    .0934552    -2.17   0.030    -.3858564   -.0195186
         ntm | -.0355184      .06763    -0.53   0.599    -.1680708    .0970339
    constant |    4.94005   .4183216    11.81   0.000     4.120155    5.759945
-------------+----------------------------------------------------------------
    /lnalpha |    -5.2532   .7888433                     -6.799305   -3.707096
-------------+----------------------------------------------------------------
       alpha |   .0052308   .0041262                      .0011145    .0245487
------------------------------------------------------------------------------
Likelihood-ratio test of alpha=0: chibar2(01) =     3.35 Prob>=chibar2 = 0.034
                  Results
 We found that the real exchange rate and
  the income elasticity of importer are
  significant variables for Argentine lemon
  exports.
 For Spain exports there are a significant
  and positive border effect (contiguous
  countries) and a significant negative effect
  related to the distance from trade
  partners.
                 Conclusions

 These findings suggest that the SPS
  regulations may be trade-restricting
 Also, that is difficult to isolate specific
  country effects
 Lack of more detailed data: our measure of
  SPS NTM is a dummy variable
                                           Thank you




NTM-IMPACT KOM
             June
Brussels, 29VIII International Agribusiness PAA-PENSA Conference - Buenos Aires, Argentina. November 30 - December 2, 2011
                                                                                                           guy.henry@cirad.fr

								
To top