My caption 😄

Innovation-exports behavior at firm-level in developing countries: How this causal relationship changes in acquisition of knowledge or R&D?

Abstract

With micro-level data from a developing country (Chile), I use an econometric causality test (Granger) to study the relationship between the innovation effort and the exports propensity, distinguishing between knowledge acquisition and R&D expenditures. The results show a difference in the causal direction between innovation and exports according to the type of innovation engaged, where the acquisition of knowledge shows a mutual reinforcement (reciprocal causality); rather than R&D expenditure, where causality is found from exports towards innovation (Learning-by-Exporting) and not in the opposite direction. Both, the former idea and the results, are important given that the main source of technological progress in developing countries is related to technological adoption from developed countries, rather that in-house R&D as in developed countries firms. My results show the relevance of the export information mechanism of foreign knowledge, given that it encourages both knowledge acquisition and R&D. This finding shows that the local markets in developing countries—including a medium size and open-to-international-market such as Chile—are still farm reaching from the technological frontier.

Keywords: Exports, Innovation, Learning-by-Exporting, Self-Selection, Mutual Reinforcement, Granger Causality Test, Developing Countries.

Introduction

The availability of new firm-level detailed data encourages a great amount of research, which studies innovation, exports and firm performance. The main findings are consistent with the idea of a positive relationship between exports, productivity and innovation (Bernard & Jensen, 1999); (CrepĂłn, Duguet, & Mairesse, 1998) and (LĂłpez RodrĂ­guez & GarcĂ­a RodrĂ­guez, 2005)]. Related to innovation activities, while some firms engaged in R&D projects to achieve new technology creation (mainly firms from developed countries), most of them merely imitate or adapt existing production techniques to local conditions (Evenson & Westphal, 1995); (UNCTAD, 1999). This idea is particularly important in developing countries; the main source of technological progress is related to technological adoption from developed economies, rather that in-house R&D in developing countries (Hoekman, Maskus, & Saggi, 2004); (UNCTAD, 2004).

Additionally, the innovation concept has gone far away from the R&D activities in the traditional sense, which imply that not every innovation has a technological source. The last (third) edition of the Oslo Manual (OCDE & Eurostat, 2005) defines innovation as the implementation of a new or significantly improved product (good or service) or process, a new marketing method, or a new organizational method in business practices, workplace organization or external relations, where only the first two concepts are related to technological innovations. To be categorized as an innovation, the minimum requirement is that the innovative product, process, marketing method or organizational method must be new (or significantly improved) to the firm.

The acquisition and adaptation of knowledge for catching-up the technology frontier have turned into important factors on economic development, and those factors are the common denominator for successful strategies of development, which were applied to several countries like the Republic of Korea, Finland, Ireland, Singapore, Taiwan, and recently, China and India. This is the main reason why developing countries started looking for strategies to facilitate local firms adapting the knowledge and technology originated in developed countries as their main development strategy (Inter-American Development Bank, 2010). Furthermore, the need to study innovation in a broad definition, and not just restricted to R&D activities, but doing the distinction between knowledge acquisition and R&D, is especially important for developing countries.

How does the innovation-exports causal behaviour in developing countries change when considering R&D or acquisition of knowledge? In order to answer this question, Chile is an interesting, small developing country to study. The country experienced a very successful transformation in the last thirty years, many substantial economic reforms were implemented. For example, measures aimed for boosting competition and becoming open to the global economy. Despite the higher openness, investment in technological innovation in Chile remains small relative to more developed countries. Furthermore, it could be interesting to study other instance of innovations, rather than R&D, as I propose here. And Chile, despite its openness to knowledge from global market, did not have raised R&D innovations. This paper aims to provide evidence, for developing countries, of the causal relationship between the innovation effort and the export propensity, distinguishing between knowledge acquisition and R&D expenditures. We can see this as a contribution regarding previous papers in two main aspects.

First, there are only a few studies that focus on the firm-level export-innovation relationship in developing countries, and those few studies only take into account R&D expenditure as innovation and they do not consider acquisition of knowledge (Benavente, Ortega-Bravo, & González, 2013); (Álvarez, García, & García, 2008); (ƞeker, 2012); (Alvarez & Robertson, 2004) and (Almeida & Fernandes, 2007). In that sense, my paper studies interesting aspects for mainly low-tech firms from developing countries. Firms from developing countries innovate, but not in a sophisticated-disruptive way like those firms from developed economies. They do not have large R&D departments and laboratories; rather, they adapt external knowledge and technology in the form of patents, license, and new machinery for innovation. They perform little improvements to their products in order to achieve new market requirements; they change the shape of their products and not their technical specifications. Furthermore, developing countries firms’ innovations are mainly related to other innovation activities, like acquisition of knowledge, rather than R&D (Hoekman, Maskus, & Saggi, 2004); (UNCTAD, 2004).

Secondly, there is no literature that studies innovation-export causal relationship considering both R&D and acquisition of knowledge separately. On one hand, there is a considerable amount of literature—mainly in developed countries—that study the effects of innovation on firms’ exporting behavior (Barrios, Görg, & Eric, 2003); (Cho & Pucik, 2005); (DĂ­az-DĂ­az, Aguiar, & De SaĂĄ-PĂ©rez, 2008); (KylĂ€heiko, Jantunen, & Puumalainen, 2011); (Vila & Kuster, 2007); (Basile, 2001); (Cassiman & Golovko, 2010) and (Wakelin, 1998). However, on the other hand, some literature has examined the reverse relationship—namely, the effect of exports on firms’ technological resources and innovation (Golovko & Valentini, 2011); (Hitt, Hoskisson, & Hicheon, 1997). Internationalized firms are able to maintain their international competitiveness by acquiring more experience and technological knowledge in foreign markets (Zahra, Ireland, & Hitt, 2000). These papers examine only a single causal direction of innovation-export relationship (Cho & Pucik, 2005); (Damijan, Kostevc, & Polanec, 2010); (KylĂ€heiko, Jantunen, & Puumalainen, 2011), and have not considered the double relationship (Kumar & Saqib, 1996); (Salomon & Shaver, 2005); (Zahra, Ireland, & Hitt, 2000). There are a few notable expectations (Filatotchev & Piesse, 2009); (Golovko & Valentini, 2011); (Monreal-PĂ©rez, AragĂłn-SĂĄnchez, & SĂĄnchez-MarĂ­n, 2012), which jointly examine innovations and exports without defining the causality relation previously. Their study complements that of (Filatotchev & Piesse, 2009) by examining the joint effect of innovation and exports over small and medium-sized enterprises’ growth.

(Filipescu, Prashantham, Rialp, & Rialp, 2013) Filipescu et.al. (2015) studies the double causal effect between a firm’s export and innovation activities, which has been overlooked insofar as they have typically been related to one another unidirectionaly (Pla-Barber & Alegre, 2007); (Vila & Kuster, 2007).

None of these study the relationship between the innovation effort and the exports propensity in developing countries, considering how this causal relationship change under technological adoption or R&D expenditures.

The Model

Following the recommendations of Filipescu et al. (2009), I lagged variables related to innovation by one time period to account for the delay between the R&D investments and the results of these investments. Filipescu et al. (2009) indicated that lagging the innovation variables for longer periods does not significantly influence the results (Filipescu, Rialp, & Rialp, 2009).

However, because expected impacts between innovations and exports and vice versa may not be immediate, we do not expect them to necessarily occur simultaneously (Filipescu, Rialp, & Rialp, 2009), I analyzed their respective effects lagged in one time. Thus, I followed Salomon and Shaver’s (2005) advice to introduce lags into the analysis to reduce possible simultaneity problems (Salomon & Shaver, 2005). Similarly, Baum (2006) considers lags important to improve prospects of valid causal inference. In addition, I include year and subsector dummies for both analyses. This is different, for two primary reasons, from Filipescu, Prashantham, Rialp, & Rialp, (2013) because they consider two lags. First, in their work, they focus on the relation of technological innovations and exports. In my work, I do not consider 2 lags because my innovation definition does not have a technological constrain, so it could be that the purpose of innovation do not have a radical innovation as their purpose (like new product or processes), but achieving an incremental innovation (Pavitt, What makes basic research economically useful?, 1991). Furthermore, one lag could be enough to measure the expected impacts between innovations and exports and vice versa, rather than the two lags for technological innovations. Secondly, there is a data constrains of the survey. INE do not offer a time series panel database. The survey used was constructed with waves of 1 years lagged data, so I have only one lag to construct the granger causality test.

I completed this study by using a pooled cross-sectional data, estimating a Granger test based on one lag. Considering the manufacturing sub-sectors (s), the model specification was established in the following manner for every sub-sector:

Where: - g:Innovation expenditure effort - exp: Export intensity - v:Productivity - l:Labor - k:foreign capital (dummy) - A: Set of categorical variables by year and sub-sector

These innovation and export equations are estimated by means of a Tobit model (Amemiya, 1973), considering that both exporting firms and firms that carry out innovation efforts are censured samples.

The variables are calculated in the following manner: the different amounts of innovation expenditure are calculated according to the aforementioned definition, and afterwards the amount of innovation expenditure per worker is calculated, which is called innovation effort. Exports are considered to be the real value of total exports. This value is divided by the total sales, and the result is the company’s export intensity. Finally, labor productivity is measured by means of the number of sales per worker. I used categorical variables (A) to capture the variance stemming from the year of observation, which is relevant to capture exogenous effects in the model.

Main Results

Given that the focus of this paper is to study the existing causality between exports and innovation, and following the definition of the Granger Causality Test, I report the F-statistics of the variables.

In the next Table, we can see that there is statistically significant Granger causality from exports towards innovation for both R&D and knowledge acquisition where there is a positive relationship.

Table 6: F-statistics, Granger Causality Test: Innovation Effort

variables of the model Acquisition of Knowledge R&D
Export.Intensity_(t-1) + +
Year dummies Yes Yes
Sector dummies Yes Yes
F-stat. 9.43 42.29
Prob> F 0.00 0.00
Obs. 4,942 5,127
Obs. Uncensored 287 1,297
Obs. Censored 4,655 3,830
Estimation tobit tobit

Source: Author’s elaboration based on information from the EIT and ENIA. Control Variables: Labor Productivity, Labor, Foreign Property.

Furthermore, in Table 7 we observe that there is no statistically significant Granger causality from innovation towards exports in the case of R&D. In that sense, innovation expenditure may not have a relevant causal effect on export intensity. The results of the estimations for knowledge acquisition show that there exist causality from innovation towards exports

Table 7: F-statistics, Granger Causality Test: Export Intensity

variables of the model Acquisition of Knowledge R&D
Innov.Effort_(t-1) + +
Year dummies Yes Yes
Sector dummies Yes Yes
F-stat. 5.89 2.40
Prob> F 0.02 0.12
Obs. 4,942 5,127
Obs. Uncensored 1,618 1,995
Obs. Censored 3,324 3,132
Estimation tobit tobit

Source: Author’s elaboration based on information from the EIT and ENIA. Control Variables: Labor Productivity, Labor, Foreign Property.

Consequently, the evidence that I find points first towards a difference in the causal direction between innovation and exports according to the type of innovation effort that was carried out. When considering all R&D expenditures, we see causality from exports towards innovation (Learning-by-Exporting) and not in the opposite direction. However, when considering innovations related to the acquisition of knowledge, there is mutual reinforcement, as there is statistical significance in both directions (reciprocal causality).

The Github project is Here

  drop _all
  set memory 1g
 
  ******************BASES DE DATOS************************************************
  run  "C:\Users\fgreve\Dropbox\WP9\stata\7.do"
  save "C:\Users\fgreve\Dropbox\WP9\stata\7.dta", replace
  drop _all
  
  run  "C:\Users\fgreve\Dropbox\WP9\stata\8.do"
  save "C:\Users\fgreve\Dropbox\WP9\stata\8.dta", replace
  drop _all
  
  *use "C:\Users\fgreve\Dropbox\WP9\stata\7.dta"
  *append using "C:\Users\fgreve\Dropbox\WP9\stata\8.dta"
  
  use "C:\Users\fgreve\Dropbox\WP9\stata\8.dta"
  
  count
  scalar count_all = r(N)
  
  drop if manufactura_dummy==0 
  count
  scalar count_manuf = r(N)
  
  ************VARIABLES***********************************************************
  gen vl          = v/l
  gen VL          = V/L
  gen gl        = g/l
  gen GL        = G/L
  
  gen gv        = g/v
  gen GV        = G/V
  gen ev        = exp/v
  gen EV        = EXP/V
  gen el        = exp/l
  gen EL        = EXP/L
  gen ln_ev     = ln(ev)
  gen ln_EV     = ln(EV)
  
  gen G_mil     = G/1000 
  gen EXP_mil   = EXP/1000
  
  *gen ln_g       = ln(g)
  *gen ln_G       = ln(G)
  gen gD         = cond(g>0,1,0)
  gen GD        = cond(G>0,1,0)
  *gen gl           = g/l
  *gen GL           = G/L
  
  gen g1D         = cond(g1>0,1,0)
  gen G1D       = cond(G1>0,1,0)
  gen g1l           = g1/l
  gen G1L           = G1/L
  
  gen ln_g2       = ln(g2)
  gen ln_G2       = ln(G2)
  gen g2D         = cond(g2>0,1,0)
  gen G2D       = cond(G2>0,1,0)
  gen g2l           = g2/l
  gen G2L           = G2/L
  
  gen ln_g3       = ln(g3)
  gen ln_G3       = ln(G3)
  gen g3D         = cond(g3>0,1,0)
  gen G3D       = cond(G3>0,1,0)
  gen g3l           = g3/l
  gen G3L           = G3/L
  
  
  gen sector_1 = CIIU2_15
  gen sector_2 = CIIU2_20
  gen sector_3 = CIIU2_21
  gen sector_4 = CIIU2_24
  gen sector_5 = CIIU2_27
  gen sector_6 = CIIU2_28
  gen sector_7 = CIIU2_31
  gen sector_8 = CIIU2_99
  
  gen Subsector_=.
  replace Subsector_=1 if sector_1==1
  replace Subsector_=2 if sector_2==1
  replace Subsector_=3 if sector_3==1
  replace Subsector_=4 if sector_4==1
  replace Subsector_=5 if sector_5==1
  replace Subsector_=6 if sector_6==1 
  replace Subsector_=7 if sector_7==1
  replace Subsector_=8 if sector_8==1
  
  recode Subsector_ ///
  (1 = 1 "1") ///
  (2 = 2 "2") ///
  (3 = 3 "3") ///
  (4 = 4 "4") ///
  (5 = 5 "5") ///
  (6 = 6 "6") ///
  (7 = 7 "7") ///
  (8 = 8 "8"), ///
  gen(Subsector) 
  
  gen enc_7 = 0
  gen enc_8 = 0
  replace enc_7=1 if encuesta==7
  replace enc_8=1 if encuesta==8
  
  gen gasto = .
  replace gasto =  0 if G>0
  replace gasto =  1 if G1>0
  replace gasto =  2 if G2>0
  replace gasto =  3 if G3>0
  
  
  forv i = 1/15 {
  gen region`i' =cond(region==`i',1,0)
  }
  
  gen  n_GD       = cond(GD==0,1,0)
  gen  n_G1D      = cond(G1D==0,1,0)
  gen  n_G2D      = cond(G2D==0,1,0)
  gen  n_G3D      = cond(G3D==0,1,0)
  gen  n_EXPD     = cond(EXPD==0,1,0)
  gen  n_k_ext    = cond(k_ext==0,1,0)
  
  gen exp_innov      = GD*EXPD
  gen exp_innov1     = G1D*EXPD
  gen exp_innov2     = G2D*EXPD
  gen exp_innov3     = G3D*EXPD
  
  
  gen GD_x100        = GD*100
  gen n_GD_x100      = n_GD*100
  gen G1D_x100       = G1D*100
  gen n_G1D_x100     = n_G1D*100
  gen G2D_x100       = G2D*100
  gen n_G2D_x100     = n_G2D*100
  gen G3D_x100       = G3D*100
  gen n_G3D_x100     = n_G3D*100
  
  gen k_ext_x100     = k_ext*100
  gen n_k_ext_x100   = n_k_ext*100
  gen EXPD_x100      = EXPD*100
  gen n_EXPD_x100    = n_EXPD*100
  gen INN            = factor
  gen exp_innov_x100 = exp_innov*100
  gen exp_innov1_x100= exp_innov1*100
  gen exp_innov2_x100= exp_innov2*100
  gen exp_innov3_x100= exp_innov3*100
  
  gen l1 = cond(l<50,1,0)
  gen l2 = cond(l<200,1,0) - l1
  gen l3 = cond(200<l,1,0) 
  
  gen l1_x100 = l1*100
  gen l2_x100 = l2*100
  gen l3_x100 = l3*100
  
  *****LIMPIEZA DE DATOS**********************************************************
  count
  scalar count_clean0 = r(N)
  
  drop if l==. | L==. | l==0 | L==0
  drop if v==. | V==. | v==0 | V==0
  drop if exp>v | EXP>V | g>v | g>V 
  
  count
  scalar count_clean1 = r(N)
  
  ************LABELS**************************************************************
  label variable v                  "Sales"    
  label variable V                  "Sales"
  
  label variable g                  "Innov.Exp"    
  label variable G                  "Innov.Exp"
  
  label variable exp                "Export"  
  label variable EXP                "Export"
  
  label variable l                    "Labor" 
  label variable L                    "Labor"
  
  label variable GD                   "Innov.Expend^d [general]"  
  label variable gD                   "Innov.Expend^d [general]"
  label variable G1D                  "Innov.Expend^d [tech]"  
  label variable g1D                  "Innov.Expend^d [tech]"
  label variable G2D                  "Innov.Expend^d [know]"  
  label variable g2D                  "Innov.Expend^d [know]"
  label variable G3D                  "Innov.Expend^d [train]"  
  label variable g3D                  "Innov.Expend^d [train]"
    
  label variable expD                 "Export^d" 
  label variable k_ext                "Foreign.Property^d"
  label variable EXPD                   "Export^d"
  
  label variable vl                 "Labor Productivity"
  label variable VL                 "Labor Productivity"
  label variable G_mil              "Gasto.Innov"
  label variable EXP_mil                "Export"
  
  label variable GL                     "Innov.Exp.Worker"
  label variable gl                     "Innov.Exp.Worker"
  
  label variable GL                     "Innov.Effort [general]"
  label variable gl                     "Innov.Effort [general]"
  label variable G1L                    "Innov.Effort [tech]"
  label variable g1l                    "Innov.Effort [tech]"
  label variable G2L                    "Innov.Effort [know]"
  label variable g2l                    "Innov.Effort [know]"
  label variable G3L                    "Innov.Effort [train]"
  label variable g3l                    "Innov.Effort [train]"
  
  label variable EV                     "Export.Intensity"
  label variable ev                     "Export.Intensity"
  label variable EL                     "Export.worker"
  label variable el                     "Export.worker"
  
  global encuesta                   "enc_7 enc_8"
  global sector                         "CIIU2_15 CIIU2_20 CIIU2_21 CIIU2_24 CIIU2_27 CIIU2_28 CIIU2_31 "
  global region                     "region1 region2 region3 region4 region5 region6 region7 region8 region9 region10 region11 region12 region13 region14 " 
  
  label variable GD_x100                "Innov.Expend [general]"
  label variable G1D_x100           "Innov.Expend [tech]"
  label variable G2D_x100           "Innov.Expend [know]"
  label variable G3D_x100           "Innov.Expend [train]"
  
  label variable k_ext_x100             "Foreign.Property"
  label variable EXPD_x100          "Export."
  label variable INN                    "Represented Firms"
  
  label variable exp_innov_x100         "Innov [general]. and Export."
  label variable exp_innov1_x100        "Innov [tech] and Export."
  label variable exp_innov2_x100        "Innov [know] and Export."
  label variable exp_innov3_x100        "Innov [train] and Export."
  
  label variable l1_x100                "Small firms"
  label variable l2_x100                "medium firms"
  label variable l3_x100                "large firms"
  
  ********************************************************************************
  gen lnV   = log(V)
  gen lnEXP = log(EXP)
  gen lnG   = log(G)
  gen lnL   = log(L)
      
  label variable lnV "Sales"
  label variable lnG "Innov.Exp"
  label variable lnEXP "Export"
  label variable lnL "Labor"
  
  global vars "lnV lnEXP lnG lnL"
  //EST.DESCRIPTIVA X SECTOR
  eststo A: quiet estpost su $vars , detail
  eststo N: quiet estpost su $vars if GD== 0, detail
  eststo S: quiet estpost su $vars if GD== 1, detail
  forv i = 1/3 {
  eststo E`i': quiet estpost su $vars if gasto==`i' , detail
  }
  esttab A N S E1 E2 E3 /// 
  using "C:\Users\fgreve\Dropbox\WP9\doc\estdesc0.rtf", ///
  mlabel("all-sample" "no-expend" "expend" "tech" "know" "train") ///
  title("Summary Stats Mean, by type (logarithmic values)") ///
  replace collabels(none)  label nogaps nonumbers /// 
  addnotes("Source: Authors elaboration.") ///
  cells(mean(fmt(%12.1fc)))
  
  ********************************************************************************
  gen lnVL = log(VL) 
  gen lnEL = log(EL)
  gen lnGL = log(GL)
  
  label variable lnVL "Labor Productivity"
  label variable lnEL "Export.worker"
  label variable lnGL "Innov.Exp.Worker"
  
  global vars "lnVL lnEL lnGL"
  //EST.DESCRIPTIVA X SECTOR
  eststo A: quiet estpost su $vars , detail
  eststo N: quiet estpost su $vars if GD== 0, detail
  eststo S: quiet estpost su $vars if GD== 1, detail
  forv i = 1/3 {
  eststo E`i': quiet estpost su $vars if gasto==`i' , detail
  }
  esttab A N S E1 E2 E3 /// 
  using "C:\Users\fgreve\Dropbox\WP9\doc\estdesc1.rtf", ///
  mlabel("all-sample" "no-expend" "expend" "tech" "know" "train") ///
  title("Summary Stats Mean, by type (logarithmic values)") ///
  replace collabels(none)  label nogaps nonumbers /// 
  addnotes("Source: Authors elaboration.") ///
  cells(mean(fmt(%12.1fc)))
  
 ********************************************************************************
  gen EV100 = EV*100
  gen GV100 = GV*100
  
  label variable EV100 "Export.Intensity"
  label variable GV100 "Innov.Intensity"
  
  global vars "EV100 GV100"
  //EST.DESCRIPTIVA X SECTOR
  eststo A: quiet estpost su $vars , detail
  eststo N: quiet estpost su $vars if GD== 0, detail
  eststo S: quiet estpost su $vars if GD== 1, detail
  forv i = 1/3 {
  eststo E`i': quiet estpost su $vars if gasto==`i' , detail
  }
  esttab A N S E1 E2 E3 /// 
  using "C:\Users\fgreve\Dropbox\WP9\doc\estdesc2.rtf", ///
  mlabel("all-sample" "no-expend" "expend" "tech" "know" "train") ///
  title("Summary Stats Mean, by type (percent values)") ///
  replace collabels(none)  label nogaps nonumbers /// 
  addnotes("Source: Authors elaboration.") ///
  cells(mean(fmt(%12.1fc)))
  
  ********************************************************************************
  global vars "EXPD_x100 k_ext_x100 l1_x100 l2_x100 l3_x100"
  //EST.DESCRIPTIVA X SECTOR
  eststo A: quiet estpost su $vars , detail
  eststo N: quiet estpost su $vars if GD== 0, detail
  eststo S: quiet estpost su $vars if GD== 1, detail
  forv i = 1/3 {
  eststo E`i': quiet estpost su $vars if gasto==`i' , detail
  }
  esttab A N S E1 E2 E3 /// 
  using "C:\Users\fgreve\Dropbox\WP9\doc\estdesc3.rtf", ///
  mlabel("all-sample" "no-expend" "expend" "tech" "know" "train") ///
  title("Summary Stats, by type (%)") ///
  replace collabels(none)  label nogaps nonumbers /// 
  addnotes("Source: Authors elaboration.") ///
  cells(mean(fmt(%12.1fc)))
  
  *****CAUSALIDAD*****************************************************************
  *****GASTO EN INNOVACION GENERAL************************************************
  global control 
  *"l k_ext"
  label variable ev          "\$Export.Intensity_{t-1}$"
  label variable EV          "\$Export.Intensity_{t}$"
  
  *****CAUSALIDAD EXPORTACION SOBRE VENTAS
  gen g_l = gl
  gen G_L = GL
  label variable G_L          "\$Innov.Effort_{t}$"
  label variable g_l          "\$Innov.Effort_{t-1}$"
  quiet tobit EV  ev G_L g_l $control  $sector $region, ll(0)
  quiet test (_b[g_l]=0) (_b[G_L]=0) 
  eststo c , add(Fc_g r(F) p_g r(p) )
  
  drop g_l G_L
  gen g_l = g1l
  gen G_L = G1L
  label variable G_L          "\$Innov.Effort_{t}$"
  label variable g_l          "\$Innov.Effort_{t-1}$"
  quiet tobit EV  ev G_L g_l g2D g3D $control  $sector $region, ll(0)
  quiet test (_b[g_l]=0) (_b[G_L]=0) 
  eststo c1 , add(Fc_g r(F) p_g r(p) )
  
  drop g_l G_L
  gen g_l = g2l
  gen G_L = G2L
  label variable G_L          "\$Innov.Effort_{t}$"
  label variable g_l          "\$Innov.Effort_{t-1}$"
  quiet tobit EV  ev G_L g_l g1D g3D $control  $sector $region, ll(0)
  quiet test (_b[g_l]=0) (_b[G_L]=0) 
  eststo c2 , add(Fc_g r(F) p_g r(p) )
  
  drop g_l G_L
  gen g_l = g3l
  gen G_L = G3L
  label variable G_L          "\$Innov.Effort_{t}$"
  label variable g_l          "\$Innov.Effort_{t-1}$"
  quiet tobit EV  ev G_L g_l g1D g2D $control  $sector $region, ll(0)
  quiet test (_b[g_l]=0) (_b[G_L]=0) 
  eststo c3 , add(Fc_g r(F) p_g r(p) )
  
  esttab c c1 c2 c3 ///
  using "C:\Users\fgreve\Dropbox\WP9\doc\GrangerExportacion.rtf", ///
  keep(G_L g_l) ///cells(none) suprime coeficientes de la regresion
  cells(_sign) ///
  nogaps collabels(none) ///
  mlabel("general" "tech" "know" "train") /// 
  indicate( "FE Other Expend. = g1D g3D" "FE Sector = CIIU2_15" "FE Geog = region1") ///
  star type not label nonumbers replace /// 
  star(* 0.10 ** 0.05 *** 0.01) ///
  substitute(_ _  $  $  %  \% ) ///
  title("F-statistics, Granger Causality Test: Export Intensity") ///
  stats(Fc_g p_g  N N_unc N_lc cmd, /// 
  fmt(%15.2fc %15.2fc %15.0fc %15.0fc %15.0fc ) ///
  labels("F-stat." "Prob $>$ F" "Obs." "Obs.Uncensured" "Obs.Censured" "Estimation")) ///              
  addnotes("Source: Author's elaboration based on information from the EIT and Enia." ///
  "Control Var: Labor, Foreign Property.")
  
  
  *****CAUSALIDAD GASTO EN INNOVACIԎ
  quiet tobit GL gl EV ev $control  $sector $region, ll(0)
  quiet test (_b[ev]=0) (_b[EV]=0) 
  eststo c , add(Fc_ev r(F) p_ev r(p) )
  
  quiet tobit G1L g1l EV ev g2D g3D $control $sector $region, ll(0)
  quiet test (_b[ev]=0) (_b[EV]=0) 
  eststo c1 , add(Fc_ev r(F) p_ev r(p) )
  
  quiet tobit G2L g2l EV ev g1D g3D $control $sector $region, ll(0)
  quiet test (_b[ev]=0) (_b[EV]=0) 
  eststo c2 , add(Fc_ev r(F) p_ev r(p) )
  
  quiet tobit G3L g3l EV ev g1D g2D $control $sector $region, ll(0)
  quiet test (_b[ev]=0) (_b[EV]=0) 
  eststo c3 , add(Fc_ev r(F) p_ev r(p) )
  
  esttab c c1 c2 c3  ///
  using "C:\Users\fgreve\Dropbox\WP9\doc\GrangerGasto.rtf", ///
  keep(EV ev) ///cells(none) suprime coeficientes de la regresion
  cells(_sign) ///
  nogaps collabels(none) /// 
  indicate( "FE Other Expend. = g1D g3D" "FE Sector = CIIU2_15" "FE Geog = region1") ///
  mlabel("general" "tech" "know" "train") ///  
  star type not label nonumbers replace /// 
  star(* 0.10 ** 0.05 *** 0.01) ///
  substitute(_ _  $  $  %  \% ) ///
  title("F-statistics, Granger Causality Test: Innovation Effort") ///
  stats(Fc_ev p_ev  N N_unc N_lc cmd, /// 
  fmt(%15.2fc %15.2fc %15.0fc %15.0fc %15.0fc ) ///
  labels("F-estat." "Prob $>$ F" "Obs." "Obs.Uncensured" "Obs.Censured" "Estimation")) ///             
  addnotes("Source: Author's elaboration based on information from the EIT and Enia." ///
  "Control Var: Labor, Foreign Property.")
UpDog logo  Proudly hosted with UpDog.