nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
Since the opening of trade in Mexico in 1994, foreign direct investment (FDI) has played an important role in the economic growth and development of Mexican economy, reporting a constant and positive flow since then. However, rampant insecurity along with slow economic growth in Mexican states have had a negative effect on investment flow in some regions. Using available information from the Secretaría de Economía and the National Institute of Statistics and Geography (Inegi, by its acronym in Spanish), for all 32 Mexican states, we aim to answer the following questions: Has FDI decreased in the states that are considered insecure? In which states do total FDI, new investment and reinvestment have a constant and positive flow? Is past performance of neighboring states’ investment, important in determining investment flow towards a given state?
Using the Spatial Vector Autoregressions technique (SpVAR) this work shows the effect of FDI, economic growth and insecurity on the 32 federal entities of Mexico. By analyzing the states that have received more FDI and states with the highest insecurity indicators, the work shows the causality between variables, and the impact of these variables among states.
There are discussions about the effects of spillovers from large transnational corporations or from FDI over Mexican states. Research from Gutiérrez-Portilla et al., (2016) consider that the only benefit they produce is less precarious jobs and higher wages than those from local businesses. In contrast with previous studies, this work shows that the effects of FDI, production growth and insecurity between states are far more important than the benefits within a state. Using the SpVAR methodology, the impacts received by an entity from its neighbors are quantified. Also, our study presents measurements of the benefits that a state produces within itself and on its neighbors when it receives FDI. Additionally, our findings show that the regions concentrating most investments are not the same as those with the highest insecurity indexes, which could be an indication of dispersion of investment to safer regions instead of concentration.
Using a spatial methodology, we show that the externalities generated by the variables considered in this work are greater than the impacts within a state. This implies the need to analyze whether the entity that receives FDI flows benefits its neighboring entities in a positive way by transferring part of its growth (i.e. push-out effect), or if the entity gets benefits from the increase of FDI of its neighbors (i.e. push-in effect) (Márquez et al., 2010; 2014). This also implies the need to analyze the internal and external links that investment produces within the country. These links are not quantified, it could be thought as if there is no Granger causality between insecurity and FDI. This work shows that in fact, there is Granger causality from the spatial perspective. Although, there is a weak causalitybetween economic growth and FDI, this can be explained by the slowtotal economic growth of the country, which cannot attract enough foreigninvestment.
The main aim of our research is to show the tendencies of FDI and its relationship to states’ economic growth and insecurity. The most important contribution lies in demonstrating that insecurity in neighboring states has a positive effect on the concentration of these investments in regions with the potential to attract them. It is also shown that states which fail to attract FDI are benefited when neighbors attract this investment and are affected when neighboring states have high levels of insecurity.
The hypothesis to be tested is that “security is fundamental to attract new investments, given that investors will invest in a certain state only if they feel secure about recovering their invested capital. Not only security in a particular state is important, but also the security offered by its neighbors; this is, it is not enough that security policies are applied in states that have the potential to attract FDI, a regional policy must be set in place, because the benefits will spill over the entire surrounding region”.
The structure of the present work is as follows: the first section presents an introduction; the second section discusses some of the theoretical arguments about the advantages of FDI and empirical evidence from the literature; the third section is dedicated to the model specification and the data used with the spatial time series methodology; the fourth section shows the empirical evidence under different scenarios; finally, the fifth section contains the main conclusions of our work.
1. Theoretical discussion about FDI
Specialized literature (Banco Interamericano de Desarrollo (BID), 2018; Blonigen and Feenstra, 1997) differentiates two types of FDI BID: a) Horizontal FDI that consists in doubling the domestic activities on a foreign country in a way that some activities like assembly and production take place in both the headquarters’ country, and the foreign countries. The objective of horizontal FDI is to have production be as close as possible to the final consumers and clients in order to reduce transportation costs. It is also called “seeking-market FDI”; b) Vertical FDI is characterized by activities that are dispersed geographically by different functions, in such a way that some activities, like research and development (R&D) are done in the country of origin, but others, such as assembly and production, take place in foreign countries. This type of FDI has the objective of minimizing production costs by doing different activities in different countries where the costs are lower (productive fragmentation).It is known as “efficiency-seeking FDI” and it gives place to global valuechains. Therefore, the intensive activities of R&D will take place in acountry with a great amount of highly skilled workers, while the assemblyand other activities of lower added value will take place in a country withabundance of lower skilled workers (BID, 2018).
There are other classifications of FDI in terms of the motivation of companies looking for locating their investment outside their country of origin. Among them are the following: a) Export platform FDI, is when a company seeks being close to its clients. Hence, the subsidiary is settled in a country from a region where it wants to supply its good or services, and from there it exports to different countries in that region (BID, 2018), b) Tariff jumping FDI is motivated by the wish of avoiding customs payments, and other commercial barriers (Blonigen and Feenstra, 1997, cited in BID, 2018). In this case, the companies settle their subsidiaries in countries where there are trade agreements that allow to avoid those barriers, or to reduce international commerce’s costs. Finally, c) technology-driven FDI is motivated by the interest of learning from other companies. In this way, they benefit indirectly from the technological knowledge of other companies. Multinational companies look for locating in places where the knowledge ecosystems are advanced (BID, 2018).
In this context, it is evident that foreign direct investment (FDI) is an important source of funding for developing countries. According to Dussel Peters (2004), FDI is of great importance for the economic growth due to the technology and knowledge transfer. Ashby and Ramos (2013) mention that between 2004 and 2010, Mexico was the 16th recipient of international investment behind Australia and Singapore, and ahead of India. This flow of FDI was a result of the North American Free Trade Agreement (NAFTA, now USMCA) coming into effect (Dussel Peters, 2004; Torres-Preciado et al., 2017). The NAFTA attracted international investors to establish in the north of Mexico due to the low labor costs, leading to attend North American market in a more competitive way. Other reasons that attracted FDI to Mexico was the privatization wave that the Mexican government did before and during the commercial liberalization. State-owned entities were passed on to private international investors.
There are several factors that influence the FDI flows. Daniele and Marani (2011) consider that the quality of the legal and institutional framework tends to influence the amount of FDI received by a country. Investment flows can also be influenced by the diversity of political and institutional systems (Globerman and Storer, 2009); meaning that the institutions are important for these investments because of the following reasons: “first ofall, according to studies about long term growth prospects, efficient institutionsimprove productivity prospects, and this attracts investors” (Danieleand Marani, 2011: 133). Secondly, a poor institutional framework meansthat there will be additional costs for multinationals increasing their totalcosts affecting their international competitiveness.
FDI is done by highly competitive multinational companies, they assimilate internal costs in a country, but when these costs are high or numerous it affects their investment decision making. According to Daniele and Marani (2011) some of the factors that attract FDI are: 1) a great potential of domestic and foreign markets; 2) high population density; 3) presence of foreign capital (signal effect); 4) good accessibility and infrastructure (since government investment is an important factor for attracting FDI); 5) high skilled and specialized human capital as well as R&D expenditure; 6) presence of economic clusters determined by a numerous amount of competitors, clients and suppliers of the same sector as the multinationals (Pelegrín and Bolancé, 2008) .
There are several factors that determine FDI flows to different regions. For instance, Lindsey Blanton and Blanton (2007) consider that the size of the market is the most common factor for attracting FDI. A country with a big market has a higher probability of attracting FDI, since it allows economies of scale in terms of production and distribution. In addition to the latter, Barry et al., (2003) mention that economists have long acknowledged the importance and benefits of clustered economies for the location of companies. The implications of clusters have been analyzed widely in the growing literature of the “new Economic Geography” (Krugman and Venables, 1996).
The advantages of the location refer to the peculiarities of a particular location that makes it more attractive to FDI. However, “far from promoting economic development of regional economies, they help to increase regional inequalities”. These inequalities are exacerbated when some regions have insecurity problems and lack of FDI drivers. Here, “advantages of location refer to the peculiarities of a particular place that makes it more attractive for foreign investment” (Gutiérrez-Portilla et al., 2016: 71).
Hence, the advantages of a good location are due to the search of markets, resources, efficiency and strategic assets. According to Dunning (1993), this means that the search of a new market is to take advantage of economies of scale in each region. Meanwhile, the search of resources allows to access low-cost natural resources and that these contribute to improve international competitiveness. While the search for efficiency allows to promote and improve labor division and specialization of assets of the multinational companies. Finally, the search of strategic assets allowsto protect and keep the companies’ advantages and to curb their competitors(Dunning, 1993).
Among the benefits that FDI produces, Guzmán Anaya (2013: 236) considers that “FDI has positive effects in host countries given that they produce spillovers in the local companies that contribute to increase their productivity. The transmission channels of the productivity spillover are the workers mobility, the competition effect, and the demonstration effect”.
Appleyard and Field (2003) highlight five of the multiple benefits of FDI: 1. Increase of competitiveness of the host country, since the arrival of new companies helps to reduce the pressure on prices, and to improve the efficiency in resource allocation promoting the competitiveness of local companies and promoting a learning process. When multinational companies begin to settle, it allows access to more competitive goods. 2. Creation of human capital: Multinational companies introduce new management, marketing, and organizational practices, which demands training of personnel. Thus, multinational companies contribute to the creation of human capital in the host country, meaning that the productivity improves and provides better salaries than local companies. 3. Incentives for exports: The multinationals produce more for the international markets than for the domestic one since they are more focused on exports. This allows the domestic companies that are looking to export their products to incorporate a learning system. 4. Use of technology: FDI favors the arrival of modern technology, especially all that is related to new varieties of raw materials, high technology products, and new productive processes which cannot be obtained by means of financial investment or commerce of goods and services. Moreover, as the multinational companies establish relations with local companies through purchase of raw materials or other products, and flows of work force from company to company, they create positive externalities that contribute to the technical progress of the host country.
However, the ability to absorb the potential benefits of FDI is determined by the level of spillover that prevails between and within regions. Without these spillovers, the multinationals would function as a bubble, only benefitting their own workers. Nevertheless, a region with the above advantages to attract FDI that also has an insecurity problem as envisaged in Mexico, where kidnapping, homicide, armed robbery, etc. occur, the entry of this type of investment could be limited and regional inequalities could beincreased.
It is important to emphasize that investors may “mimic” the investment decisions of colleagues from other regions. DeCoster and Strange (1993) have pointed out that:
even if these efficiency reasons are not prevalent, firms may find it rational to agglomerate spatially. If there is uncertainty about locations in which to invest, investors may exhibit a tendency to imitate each other’s location decisions. In their specific model, this arises because investors locating in a good location provide a signal to other investors, and to banks which provide the funds for investments (DeCoster and Strange, 1993: 283).
The best way to analyze these statements is to confirm if it is the new investment or the reinvestment that has flowed to more entities.
Neighboring entities play a relevant role in FDI, because according to Ashby and Ramos (2013), crime organizations invest resources in activities that reduce the efficacy of State’s deterrence against them using violence and threats (Fiorentini, 1995). Thus, the main effect of organized crime is to increase the costs of doing businesses (Hallward-Driemeier and Stewart, 2004; Daniele and Marani, 2011), not only by locating in insecure places, but also by locating in secure entities with insecure neighbors that where supply chains can be interrupted (Barnes and Oloruntoba, 2005; Czinkota et al., 2005; Globerman and Storer, 2009; Branzei and Abdelnour, 2010). Hence, it is important to analyze if the new investment received by Mexican entities than reinvestment. If in an entity reinvestment is growing more, it implies that the multinationals settled in there have learned to manage the insecurity climate and can reinvest their profits.
In general, the insecurity creates a climate of uncertainty and the possibility of expelling investment. For instance, the local demand may decrease due to emigration or relocation of businesses (Ashby and Ramos, 2013; Greenbaum et al., 2007). However, criminal activity may lead to emigration (Ashby and Ramos, 2013; Berry and Levitt, 1999), to decrease consumption per capita (Mejía and Restrepo, 2010) or to cause relocation or permanent closures of companies (Greenbaum et al., 2007). Small companies are more vulnerable to be target of organized crime through extorsion (Daniele and Marani, 2011).
Different regions have different intrinsic advantages to attract FDI, despite their variety they seem reduced by each region’s problems. The flows of investment will be directed towards regions with more potential and factors that can attract them. Although multinationals do not care if the neighboring entities have the same potential or if at least, they offer less security for the entrance and exit of their final products, if a region does not have attractiveness to international investors and there are insecurity problems, the investment flow will be poor and scarce. Under these circumstances, as Gutiérrez-Portilla et al., (2016) remark, the FDI tends to increase the regional disparities instead of promoting regional harmony. Those inequalities will aggravate when regions have insecurity problems and lack of attractiveness to FDI.
2. Spatial distribution of insecurity and FDI
2.1. Data description
There are several studies that have analyzed FDI and its relationship with insecurity in Mexico. Most of them have related it with the homicide rates as a proxy variable of organized crime activity. Ashby and Ramos (2013) consider that individuals are less likely to report non-violent crimes, such as theft or other violent crimes like kidnapping or extorsion which can put the victims’ lives in danger. Fajnzylber et al., (2002) state that most crimes are under-reported, and that the problem is more serious in countries with less reliable judicial systems (as it occurs in Mexico). For this reason, it is argued that homicides are less under-reported (Ashby and Ramos, 2013; Fajnzylber et al., 2002). Hence, the reported homicide rate is the most used variable instead of theft or kidnapping, which does not imply that it is the most appropriate indicator but the less biased, and it has been used in studies like the one by Carbajal Suárez and Vergara González (2015).
We, too, used the homicide index as a proxy variable for insecurity. This variable was built as the number of homicides for every hundred thousand inhabitants for each one of the 32 states in Mexico. This is official information published by the Executive Secretariat of the National Security System (SESNSP by its acronym in Spanish) in 2021. Since Inegi does not publish quarterly data for the gross domestic product (GDP) of each state, we used the Quarterly Indicator of State Economic Activity (ITAEE by its acronym in Spanish) as a proxy for state GDP. The ITAEE represented economic growth for states in the country. Finally, the FDI variable was the FDI flows towards Mexico by state, whichis official information published by the Secretaría de Economía and it is classified as new investments and reinvestments made by international companies in different states of the country. All data is quarterly for the 32 states in Mexico during the period 1999:01-2019:04, in constant prices of 2019.
2.2. Exploratory analysis of FDI and insecurity
Of total FDI of 2019, Mexico City was the entity which attracted the highest percentage (23%), followed by Nuevo Leon (9.6%), State of Mexico (8.7%), Puebla (6%), Jalisco (4.6%), Coahuila and Chihuahua (4%), respectively. Meanwhile, the states with the highest homicide rate in the same year were Guanajuato (7.6%), Jalisco, the State of Mexico (6.9% each) and Chiapas (5.3%). Even though the State of Mexico and Jalisco are insecure states and received FDI flows, they did not have thehighest percentages of attraction of FDI.
Graph 1 shows the spatial behavior of the FDI by components in 1999 and 2019. It shows that in 1999 (left hand side), the percentage of new investments (PEIEDNI) was higher in Morelos (Mor), Oaxaca (Oax), Quintana Roo (QRoo), Sinaloa (Sin), Aguascalientes (Ags), Baja California Sur (Bcs), and Durango (Dgo). While reinvestments (PEIEDRI) were higher in Hidalgo (Hgo), Chiapas (Chis), Campeche (Camp), and Veracruz (Ver). The configuration of investments at state level changed in 2019, the highest FDI component was reinvestment, which implies that the already established companies were the ones keeping their investment in the same state, and that the flows of new investment were lower than in 1999, as it is the case of Guanajuato (Gto), Sonora (Son), Veracruz (Ver), and Aguascalientes (Ags). The fact that reinvestment has been greater than new investment means that somehow, multinational companies have learnt to manage the prevailing insecurity in states like Guanajuato.
However, this situation limits the entry of new investment given that it has shifted to less insecure entities such as Nayarit (Nay), Queretaro (Qro), Quintana Roo (QRoo), and Baja California Sur (Bcs), where the percentage of new investment is higher than in the aforementionedr states.
Source: author’s own elaboration based on data from Secretaría de Economía (2021).
Graph 1 Spatial distribution of new FDI (PEIEDNI), reinvestment (PEIEDRI) and accounts (PEIEDCE) between companies in 1999 and 2019
Figure 1 illustrates the spatial distribution of insecurity by state, where the states shaded with squares have more insecurity, while the states shaded with diagonal stripes have lower insecurity, and the dotted entities have a medium level of insecurity. The highest flows of FDI in 2019 came from the US with 37%, Spain with almost 12%, Germany 10%, Canada 9%, and Italy, Japan and Belgium 4% each.
Source: author’s own elaboration based on data from Secretaría de Economía (2021), QGIS (2020).
Figure 1 Spatial distribution of insecurity and FDI from USA, Spain and Germany, 2019
The concentration of FDI from the three countries that invested more in Mexico in 2019 is distributed in the following states:
US region: Investment coming from the US in 2019 is illustrated in figure 1, where the white bar has the highest concentration in Mexico City (23%), Nuevo Leon (10%), State of Mexico and Chihuahua (7.5% each), Baja California and Tamaulipas (5% each).
Spain region: Investment coming from Spain, indicated in figure 1 with a gray bar, is concentrated in Mexico City (25%), Queretaro (10%), State of Mexico and Guanajuato (7% each), Jalisco and Nuevo Leon (6% each).
Germany region: Investment from Germany, indicated in figure 1with a black bar, is concentrated in Mexico City and the State of Mexico(almost 13%), Morelos (almost 7%), and San Luis Potosi (6%).
Insecurity and its past behavior affect the flow of new investment in Mexican economy. Ashby and Ramos (2013) mention that some multinational companies have learned to manage insecurity, as it is shown above. Mostly they are companies based on natural resources like oil extraction and mining since they are limited by the locations. That is also the case of agriculture, only that this has more flexibility (Ellison and Glaeser, 1999). Oetzel et al., (2007) emphasize that although the increase of delinquency discourages foreign investment, multinational companies (MNC) in primary sectors which characterized by high sunk costs and long-term investment horizons can participate in the risk management in a continuous way. Evidence suggests that multinational companies of globalized old extraction industries have accumulated an important experience in dealing with difficult regional conditions (Ashby and Ramos, 2013), including violent conflicts and delinquency (Bennett, 2002), like in Guanajuato, where reinvestment is more important than new investment.
3. Methodology and main results
The empirical evidence of this work is based on the approaches by Márquez et al., (2010; 2014); Lesage and Cashell (2015), and Torres-Preciado (2017), who have used spatial autoregressive vector models. The work of Márquez et al., (2010; 2014) used this methodology to estimate the growth of Spanish economy, while the latter applied it to forecast the growth of manufacturing employment in Mexico. In this case, the focus is on the flow of FDI and insecurity of the 32 federal entities of Mexico. Andrés Rosales et al., (2021) used the same methodology to analyze the relationship between economic growth and public spending in the states of Mexico. In this case, the focus is on the flow of FDI and insecurity of the 32 federal entities of Mexico. The importance of using this technique is that it allows to quantify positive effects of insecurity and growth over other states. The model specification is the following:
Where the subindex i refers to the entity, t is time, and LFDIT refers to the logarithm of total foreign direct investment. LFDIT* is the average of the logarithm of foreign direct investment of the neighboring entities; ITAE is the rate of growth of the entity’s economic activity. ITAE* is the average growth rate of the economic activity of the neighboring entities. INSEC is the logarithm of the insecurity index, while INSEC* is the average of the logarithm of the insecurity index in the neighboring entities. While the Gammas are the parameters of the model.
It may be questionable that the analysis of the determinants of FDI flows is limited to the effect of the insecurity variable and to economic growth of states, especially because available literature offers several other explaining variables.1 These two variables were chosen following the conventional structure of similar reviewed studies that mostly use them. Other variables were included because the SpVAR model became unstable. However, given the hypothesis to be tested and the contribution to be made to the literature, this work highlights the effect of neighborhood for each state, the flows of FDI and the effects on economic growth.
In the spatial framework, one can determine if the lag of the externalvariables adds
valuable information about the determination of the localvariables (Márquez et al., 2014). According to Equation 1, if the previous values of
LFDI*i,t, ITAE*i,t,
INSEC*i,t help to explain the future values of
LFDIi,t, the value of the parameters
Then it can be said that there is a push-in effect, which implies thatneighbors of a
particular entity affect that entity’s behavior. If previousvalues of
LFDIi,t explain the behavior of
LFDI*i,t, ITAE*i,t,
INSEC*i,t, the parameters
The definition of the spatial weights matrix (W) is of great importancefor the spatial estimation (Quintana and Andrés-Rosales, 2014). Accordingto these authors, W is a positive square matrix with dimension thatdepends on the sample size. It describes the interaction of spatial unitsbetween entities. By definition, wij = 1, if states i and j are neighbors and0 otherwise. Although there are other types of vicinities that could beincluded which are not limited to contiguity. It is important to highlightthat a normalized matrix is specified here in the same way that traditionalspatial modeling literature does, which implies that the rows add to one,representing a spatial smoothing of the impacts of the neighboring regions and where autocontiguity is not allowed or it is equal to zero (Anselin,1988).
3.1. Spatial Granger Causality
The VAR model can be used to examine to which extent the variables in the system are determined by its past values, which is known as Granger causality (Granger, 1969). Enriched by the framework of spatial VAR (SpVAR), the spatial spreading and causality between variables and regions can be determined (Márquez et al., 2014). These new considerations were named Spatial Granger causality analysis by Márquez et al., (2014); Kuethe and Pede (2011), and it is a new way to show the collateral spatial effects within the context of traditional VAR.
On the one hand, regarding causality generated by neighbors of an entity shown as Push-in effect in columns two to four of table 1, we found that the growth in production (ITAE) of the neighbors of Aguascalientes, Baja California, Guerrero, Jalisco, Nuevo Leon and Sonora, affects the FDI of these states. While the FDI of the neighbors of Baja California Norte, Baja California Sur, Mexico City, and Oaxaca affects the FDI of these listed states. On the other hand, the increase in insecurity of the neighbors of Baja California Norte, Chihuahua, Oaxaca, Puebla, Sonora and Tamaulipas affects the FDI of these entities.
The push-out effect implies that a particular state affects the economic growth (ITAE*), the FDI* and the insecurity (INSEC*) of its neighbors by modifying its FDI component. This effect is shown in columns five to seven of Table 1, where on one hand, if the foreign investment is modified in Aguascalientes, Colima, Morelos and Sonora, it affects the economic growth of their neighbors. On the other hand, the effect of FDI of Aguascalientes, Coahuila, Guanajuato, Puebla, and Tabasco affects the FDI of their neighboring entities. Finally, there is Granger causality between FDI with respect to insecurity of the neighbors of Baja California, Baja California Sur, Guanajuato, Puebla, Queretaro, San Luis Potosi, and Zacatecas. This implies that while there is a greater flow of foreign investment in these entities, the insecurity among their neighbors will increase or decrease.
Columns eight and nine of Table 1 show that the effect that changes in insecurity levels have on direct foreign investment within the states, were statistically significant for Baja California, Campeche, Hidalgo, Estado de Mexico, Sinaloa, Tlaxcala, and Zacatecas. On the other hand, changes in FDI of Baja California, Coahuila, State of Mexico, Morelos, Nayarit, Quintana Roo, Tamaulipas and Yucatan, affect the insecurity within these states. The latter is an indication that in the aforementioned states, some investments attract insecurity and vice versa.
Table 1 Spatial Granger Causality Tests
| 1999:01 to 2019:04 | Effects between entities | Within entity effect | ||||||
|---|---|---|---|---|---|---|---|---|
| Push-in effect | Push-out effect | |||||||
| Contiguity Matrix (1) | ITAE* does not cause LFDI (2) | LFDI* does not cause LFDI (3) | INSEC* does not cause LFDI (4) | LFDI does not cause ITAE* (5) | LFDI does not cause LFDI* (6) | LFDI does not cause INSEC* (7) | INSEC does not cause LFDI (8) | LFDI does not cause INSEC (9) |
| Aguascalientes | 2.80(0.02) | 0.76(0.57) | 0.13(0.98) | 2.32(0.05) | 3.51(0.00) | 1.34(0.25) | 0.23(0.94) | 0.38(0.85) |
| Baja California | 4.21(0.04) | 3.52(0.06) | 3.30(0.07) | 0.07 (0.78) | 0.44(0.50) | 2.87(0.09) | 18.91(0.00) | 5.71(0.01) |
| Baja California Sur | 0.69(0.50) | 3.98(0.02) | 0.80(0.45) | 1.72 (0.18) | 1.55(0.21) | 4.39(0.01) | 1.25(0.29) | 0.93(0.39) |
| Campeche | 1.56(0.21) | 0.24(0.78) | 0.41(0.66) | 1.44 (0.24) | 0.84(0.43) | 0.72(0.48) | 3.66(0.03) | 1.70(0.19) |
| Mexico City | 0.45(0.77) | 2.58(0.04) | 0.88(0.47) | 0.08(0.98) | 0.25(0.90) | 1.42(0.23) | 0.33(0.85) | 0.81(0.52) |
| Chihuahua | 1.12(0.35) | 0.77(0.57) | 2.32(0.05) | 0.31(0.90) | 1.10(0.36) | 1.44(0.21) | 0.21(0.95) | 0.19(0.96) |
| Chiapas | 1.43(0.22) | 1.15(0.34) | 0.79(0.55) | 0.33(0.89) | 0.40(0.84) | 0.08(0.99) | 1.31(0.26) | 0.54(0.74) |
| Coahuila | 0.84(0.52) | 0.96(0.44) | 1.11(0.36) | 0.36(0.87) | 5.49(0.00) | 1.34(0.25) | 0.46(0.79) | 2.25(0.05) |
| Colima | 0.08(0.77) | 0.22(0.63) | 1.50(0.22) | 3.34(0.07) | 0.00(0.98) | 0.01(0.91) | 3.05(0.08) | 1.04(0.31) |
| Durango | 0.06(0.99) | 1.57(0.18) | 0.67(0.64) | 0.90(0.48) | 1.16(0.33) | 1.64(0.16) | 1.36(0.25) | 0.81(0.54) |
| Guerrero | 2.92(0.01) | 1.41(0.23) | 0.36(0.87) | 1.18(0.32) | 0.50(0.77) | 1.05(0.39) | 0.44(0.81) | 1.91(0.10) |
| Guanajuato | 0.97(0.43) | 0.61(0.68) | 2.25(0.05) | 0.14(0.98) | 4.28(0.00) | 2.94(0.01) | 1.53(0.19) | 0.35(0.88) |
| Hidalgo | 0.84(0.52) | 1.17(0.33) | 0.66(0.65) | 1.44(0.22) | 0.85(0.51) | 1.52(0.19) | 3.26(0.01) | 0.51(0.76) |
| Jalisco | 2.48(0.04) | 1.76(0.13) | 0.13(0.98) | 0.60(0.69) | 0.48(0.78) | 0.52(0.75) | 0.31(0.90) | 0.72(0.60) |
| State of Mexico | 0.45(0.50) | 0.04(0.84) | 7.61(0.00) | 0.18(0.66) | 1.63(0.20) | 1.52(0.22) | 5.28(0.02) | 3.16(0.07) |
| Michoacan | 3.44(0.06) | 0.61(0.43) | 0.03(0.84) | 0.67(0.41) | 2.30(0.13) | 0.17(0.68) | 0.02(0.88) | 0.60(0.43) |
| Morelos | 0.68(0.60) | 1.25(0.29) | 0.99(0.41) | 2.43(0.05) | 1.73(0.15) | 0.69(0.59) | 0.79(0.53) | 2.24(0.07) |
| Nayarit | 0.58(0.70) | 1.57(0.17) | 1.74(0.13) | 0.44(0.81) | 0.21(0.95) | 1.53(0.19) | 0.90(0.48) | 2.47(0.04) |
| Nuevo Leon | 2.28(0.08) | 0.93(0.42) | 1.28(0.28) | 0.10(0.95) | 1.38(0.25) | 0.74(0.52) | 0.40(0.74) | 1.11(0.34) |
| Oaxaca | 0.36(0.86) | 4.16(0.00) | 2.60(0.03) | 1.21(0.31) | 1.54(0.18) | 0.52(0.75) | 0.28(0.91) | 1.08(0.37) |
| Puebla | 0.67(0.64) | 0.45(0.80) | 3.34(0.00) | 0.43(0.82) | 3.78(0.00) | 2.50(0.03) | 1.97(0.09) | 1.24(0.29) |
| Queretaro | 0.24(0.86) | 0.29(0.82) | 1.59(0.19) | 0.85(0.47) | 1.84(0.14) | 2.73(0.05) | 0.73(0.53) | 1.53(0.21) |
| Quintana Roo | 0.56(0.72) | 1.23(0.30) | 0.88(0.49) | 1.61(0.16) | 0.23(0.94) | 0.52(0.76) | 0.34(0.88) | 2.33(0.05) |
| Sinaloa | 0.13(0.71) | 1.55(0.21) | 1.46(0.23) | 0.40(0.52) | 0.06(0.79) | 0.00(0.97) | 2.97(0.08) | 0.66(0.41) |
| San Luis Potosi | 1.97(0.09) | 0.83(0.53) | 1.74(0.13) | 0.46(0.79) | 0.15(0.97) | 2.06(0.08) | 0.37(0.86) | 1.22(0.30) |
| Sonora | 3.66(0.05) | 1.13(0.29) | 4.29(0.04) | 5.89(0.01) | 2.43(0.12) | 0.20(0.65) | 1.99(0.16) | 2.66(0.10) |
| Tabasco | 0.45(0.81) | 1.07(0.38) | 0.31(0.90) | 0.85(0.51) | 3.09(0.01) | 0.35(0.87) | 1.25(0.29) | 0.60(0.69) |
| Tamaulipas | 0.89(0.49) | 1.24(0.29) | 2.79(0.02) | 1.09(0.37) | 0.63(0.67) | 1.92(0.10) | 1.87(0.11) | 2.09(0.07) |
| Tlaxcala | 0.17(0.97) | 1.72(0.14) | 1.29(0.27) | 0.93(0.46) | 1.22(0.30) | 0.70(0.62) | 3.98(0.00) | 1.14(0.34) |
| Veracruz | 0.28(0.91) | 1.29(0.27) | 0.53(0.75) | 0.35(0.87) | 1.18(0.32) | 1.04(0.39) | 1.09(0.37) | 0.59(0.70) |
| Yucatan | 0.42(0.83) | 0.28(0.92) | 0.81(0.54) | 1.16(0.33) | 0.43(0.82) | 1.30(0.27) | 1.05(0.39) | 2.21(0.06) |
| Zacatecas | 0.82(0.53) | 1.27(0.28) | 1.59(0.17) | 0.93(0.46) | 1.69(0.14) | 1.98(0.09) | 2.99(0.01) | 1.18(0.32) |
Null hypothesisA* does not Granger cause B. P-value shown in brackets. Variables FDI, NFDI and RFDI are in natural logarithmic scale. For transformation: Ln(x) = ln(x+0.0001) if x = 0, and ln(x) if x >0.
Source: author’s own ellaboration based on data from Inegi (2021) and Secretaría de Economía (2021).
3.2. Spatial Impulse response analysis
The SpVAR was estimated for each federal entity. To begin, the optimal number lags were identified, and the stability tests were applied. Results of characteristic roots indicate that the SpVAR satisfied the stability conditions.2
In the estimation of SpVAR models, the impulse-response functionprovides us with cyclic information about the behavior and interdependencebetween regions (Márquez et al., 2014). With impulse-responsefunctions one can quantify the impacts that a region has on its neighbors(push-out spillovers), and the impact that the neighbors have on a specificregion (push-in spillovers).
Table 2 shows the quantification of the effects generated by gross domestic product growth at state level considering the contemporary effect. According to columns 2-4 of table 2, after a minor shock in neighbor states, Campeche generates and increase of 1.93%, 1.12% and 2.17% in total foreign direct investment (FDI*), new investment (NFDI*) and reinvestment (RFDI*), respectively. Chiapas is the state with the greatest increases when its neighbors modify the variables mentioned above (3.02%, 3.58% and 4.37%, respectively). Tabasco and Nuevo Leon are the most benefitted entities when new investment in their neighbors is modified, (4.37% and 4.07%, respectively), whereas reinvestment done by neighbors greatly benefits the reinvestment of Chiapas (4.37%), Guanajuato (4.32%), Oaxaca (4.42%) and Tabasco (4.64%).
Columns 5-10 in table 2 show that a small increasing shock of production (ITAE*) and insecurity (INSEC*) of the neighbors of an entity, does not have a significant effect in FDI, NFDI and RFDI of states, since the impact is less than or equal to 1.5%. The negative value observed for Yucatan (column 6) means that the economic growth of neighbors decreases new investments in the state. The latter is an indication that the economic growth and insecurity are not the most important components that affect the total investment, new investment, and reinvestment, except for a few entities like Coahuila (0.71%), Guanajuato (0.77%), Hidalgo (1.29%), where the insecurity of their neighbors positively affects the flow of total FDI towards them.
Table 2 Generalized impulse-response by federal entity and their respective neighbors
| 1999:01 to 2019:04 | Effects between entities | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Push-in spillovers | Push-out spillovers | |||||||||||||||||
| From FDI*, NFDI*, RFDI* to | From ITAE* to | From INSEC* to | From FDI, NFDI, RFDI to | From FDI, NFDI, RFDI to | From FDI, NFDI, RFDI to | |||||||||||||
| Federal entity | FDI (2) | NFDI (3) | RFDI (4) | FDI (5) | NFDI (6) | RFDI (7) | FDI (8) | NFDI (9) | RFDI (10) | FDI* (11) | NFDI* (12) | RFDI* (13) | ITAE* (14.a, 14.b, 14.c) | INSEC* (15.a, 15.b, 15.c) | ||||
| Aguascalientes1 | 0.10 | 1.97*** | 2.39*** | -1.29** | -0.71 | 0.66 | -1.45** | -0.95 | -0.41 | 0.06 | 1.55*** | 3.02*** | -0.35** | -0.26 | 0.35 | -0.01 | -0.01 | -0.01 |
| Baja California1 | 0.09** | 0.31 | 2.6*** | -0.02 | -0.11 | -0.27 | -0.03 | -0.93 | 0.11 | 0.66** | 0.21 | 2.73*** | -0.13 | -0.06 | -0.18 | 0.00 | -0.01 | 0.00 |
| Baja California Sur2 | 0.13** | 0.06 | 1.87*** | -0.03 | -0.19*** | 0.15 | 0.06 | 0.05 | 0.97 | 0.10** | 0.41 | 1.16*** | -0.18 | -0.9*** | 0.10 | 0.00 | 0.00 | 0.01 |
| Campeche2 | 1.93*** | 1.12*** | 2.17*** | -0.77 | -0.88 | -0.10 | -0.57 | -0.89 | -0.11 | 0.85 | 0.85 | 2.29*** | -0.20 | -0.32 | -0.05 | 0.00 | -0.01 | 0.00 |
| Mexico City1 | 0.13 | 1.57 | 3.11*** | -0.31 | -0.06 | -0.29 | 0.12 | 0.52 | -0.37 | 0.13 | 1.55 | 2.45*** | -0.25 | -0.04 | -0.15 | 0.00 | 0.01 | 0.00 |
| Chihuahua1 | 1.41 | 3.07*** | 2.07*** | 0.09 | -0.01 | 0.39 | 0.36 | 0.42 | 0.52 | 1.71*** | 1.97 | 1.52 | 0.08 | 0.00 | 0.18 | 0.01 | 0.00 | 0.01 |
| Chiapas2 | 3.02 | 3.58*** | 4.37 | -0.24 | -0.44 | -0.25 | 0.76 | -0.06 | -1.08 | 2.21*** | 2.86 | 3.31 | -0.06 | -0.11 | -0.07 | 0.01 | 0.00 | -0.01 |
| Coahuila2 | 0.98*** | 2.37*** | 0.83*** | 0.32 | 0.71 | -0.32 | 0.71** | 1.06** | -0.39 | 0.81*** | 2.12*** | 0.68*** | 0.14 | 0.22 | -0.14 | 0.01 | 0.01 | -0.01 |
| Colima1 | 1.07 | 1.15** | 0.88 | 0.00 | 0.62 | -0.05 | 1.13 | 1.16 | 0.12 | 0.72 | 0.98** | 1.1*** | 0.00 | 0.23 | -0.04 | 0.01 | 0.01 | 0.00 |
| Durango1 | 1.40*** | 2.70*** | 1.73 | -0.16 | -0.93 | 0.10 | -0.25 | 0.77 | -0.36 | 0.53*** | 1.52*** | 1.32*** | -0.06 | -0.26 | 0.04 | 0.00 | 0.01 | 0.00 |
| Guerrero1 | 0.95 | 0.98** | 3.15*** | -0.49 | -0.47 | 1.09*** | -0.03 | 0.97** | -0.01 | 0.47 | 0.76** | 2.21 | -0.16 | -0.17 | 0.37 | 0.00 | 0.01 | 0.00 |
| Guanajuato1 | 0.64 | 2.42 | 4.32 | -0.20 | -0.51 | -0.65 | 0.77** | 0.68** | -0.08 | 0.38 | 2.52 | 2.94 | -0.09 | -0.23 | -0.18 | 0.01 | 0.01 | 0.00 |
| Hidalgo3 | 3.53 | 3.29*** | 1.00 | -0.62 | 0.11 | -0.18 | 1.29** | 0.49 | -0.32 | 1.03*** | 1.73 | 0.83*** | -0.18 | 0.03 | -0.09 | 0.01 | 0.00 | 0.00 |
| Jalisco1 | 0.12 | 2.46*** | 3.85 | -0.97 | -0.75 | -1.70 | -0.11 | 0.36 | -0.15 | 0.05 | 1.36*** | 2.14*** | -0.31 | -0.25 | -0.43 | 0.00 | 0.00 | 0.00 |
| State of Mexico2 | 0.40*** | 3.25 | 2.02*** | 0.11 | 1.55*** | -0.01 | 0.05 | 1.74*** | -0.30 | 1.21 | 2.25*** | 1.54 | 0.24 | 0.55*** | 0.00 | 0.00 | 0.02 | 0.00 |
| Michoacan1 | 1.06 | 2.63*** | 4.35 | 0.10 | 0.15 | -0.16 | -0.63 | -0.14 | 0.50 | 0.36 | 1.36 | 2.96*** | 0.03 | 0.04 | -0.05 | -0.01 | 0.00 | 0.00 |
| Morelos2 | 1.55*** | 2.17 | 3.09*** | -0.85 | -0.05 | -0.06 | -0.16 | 1.13 | -0.05 | 0.48*** | 1.19*** | 2.08 | -0.21 | -0.01 | -0.02 | 0.00 | 0.01 | 0.00 |
| Nayarit2 | -0.09 | 0.06 | 2.00 | 0.00 | -0.03 | 0.36 | 0.04 | 0.03 | -0.32 | -0.34 | 0.10 | 1.72*** | -0.01 | -0.02 | 0.17 | 0.00 | 0.00 | 0.00 |
| Nuevo Leon1 | 1.81*** | 4.07*** | 0.39 | -0.37 | -0.18 | 0.17 | 0.34 | 1.34 | -0.67 | 0.63*** | 2.15*** | 0.30 | -0.13 | -0.04 | 0.08 | 0.00 | 0.01 | -0.01 |
| Oaxaca1 | 0.75 | 2.74*** | 4.42*** | 0.56 | -0.62 | -0.16 | 0.92 | 0.50 | 0.61 | 0.39 | 1.63*** | 2.96*** | 0.16 | -0.17 | -0.04 | 0.01 | 0.00 | 0.00 |
| Puebla2 | -0.22 | 2.97*** | 3.38 | 0.49 | 0.14 | 0.09 | -0.30 | 0.18 | 0.03 | -0.15 | 2.21*** | 2.8*** | 0.19 | 0.04 | 0.03 | 0.00 | 0.00 | 0.00 |
| Queretaro1 | 0.09 | 1.13 | 3.01*** | -0.14 | 0.30 | 1.29*** | 0.07 | 0.26 | -0.41 | 0.08 | 0.84*** | 1.87*** | -0.09 | 0.10 | 0.43*** | 0.00 | 0.00 | 0.00 |
| Quintana Roo1 | -0.03 | 0.42 | 1.56*** | -0.19 | 0.29 | -0.14 | 0.16 | 0.18 | -0.16 | -0.04 | 0.57 | 1.53*** | -0.13 | 0.19 | -0.07 | 0.00 | 0.00 | 0.00 |
| Sinaloa1 | 2.09 | 3.43*** | 2.93 | -0.07 | -0.50 | -0.86 | 0.52 | 1.24*** | -0.01 | 1.19*** | 2.29*** | 1.77*** | -0.03 | -0.16 | -0.29 | 0.01 | 0.01 | 0.00 |
| San Luis Potosi2 | 0.96** | 2.92*** | 2.59*** | 0.51 | 0.82 | 1.72*** | -0.61 | -0.45 | -1.42*** | 0.56** | 2.00*** | 1.23 | 0.19 | 0.25 | 0.44 | -0.01 | 0.00 | -0.01 |
| Sonora2 | 0.39 | 2.45 | 3.12*** | 0.16 | 0.28 | 1.11 | 0.54 | -0.03 | -0.45 | 0.15 | 1.5*** | 1.68 | 0.07 | 0.11 | 0.45 | 0.00 | 0.00 | 0.00 |
| Tabasco1 | 3.07*** | 4.37*** | 4.64*** | -0.33 | -0.48 | -0.78 | -0.89 | -0.57 | -0.33 | 1.89*** | 3.41*** | 3.40*** | -0.09 | -0.15 | -0.24 | -0.01 | -0.01 | 0.00 |
| Tamaulipas1 | 0.21 | 2.83*** | 2.70*** | -0.02 | -0.79 | 0.73 | 0.15** | 2.41*** | -0.24 | 0.88*** | 1.86*** | 1.85 | -0.04 | -0.21 | 0.28 | 0.01 | 0.02 | 0.00 |
| Tlaxcala1 | 1.85 | 2.93 | 3.74*** | 0.57 | 0.68 | 0.52 | 0.06 | 0.49 | 0.27 | 1.12*** | 2.58*** | 2.76*** | 0.25 | 0.28 | 0.21 | 0.00 | 0.00 | 0.00 |
| Veracruz1 | 3.13*** | 4.18*** | 1.1*** | -0.66 | -0.35 | 1.63*** | 0.63 | -0.08 | -0.44 | 1.58*** | 2.68 | 0.74*** | -0.19 | -0.09 | 0.53*** | 0.01 | 0.00 | 0.00 |
| Yucatan1 | 0.25 | 1.44*** | 2.67*** | 0.41 | -5.00*** | -0.30 | -0.03 | 0.31 | 0.46 | 0.25 | 0.80*** | 1.69 | -0.07 | -0.23 | -0.14 | 0.00 | 0.00 | 0.00 |
| Zacatecas1 | 0.10 | 2.52*** | 3.46*** | 0.74 | 0.03 | 0.67 | 0.30 | 0.86 | -0.19 | 0.03 | 1.17*** | 1.98 | 0.13 | 0.01 | 0.22 | 0.00 | 0.01 | 0.00 |
**10% statistical significance
***5% statistical significance.
Numbers following the name of the entity in column one indicate: 1 for 1-lag test; 2 indicates a 2-lag test, and 3 indicates a 3-lag test.
Source: author’s own ellaboration based on data from Inegi (2021) and Secretaría de Economía (2021).
It is worth noting that, contrary to other studies, the SpVAR methodology used here allows to capture the bidirectional relationship between the considered variables, instead of limiting the real explanation of the independent variable due to the unidirectional relationship modeled, for instance, by fixed effect panel data and dynamic panel data methodologies, such as that used Ramos and Ashby (2017), who analyze the effect of violent crimes on FDI in Mexican states, and who show that the lag of the variable is important to explain future FDI performance. In the same manner, Madrazo (2009) applied a panel data methodology to study the effect of violent crime on FDI; both studies found a negative relationship of violent crimes with FDI, but they do not provide any evidence on how a region’s neighbors might affect the regional FDI.
Columns two to four of table 3 show a minor disturbance in FDI, NFDI and RFDI, which positively affects these variables in time. The impact of these variables in time is more important than the impact in other variables like growth or insecurity (these impacts are not included due to a lack of space). The effects of insecurity and economic growth of entities shown in column 5 of table 3 (columns 5-10), have a small effect on the different components of investment. Some entities have a negative impact, like Colima, Puebla, Tabasco and Zacatecas (-2.16%, -1.39%,-1.95%, and -1.30%, respectively). Negative effects indicate that the past insecurity has negatively influenced investment in these states.
Table 3 Generalized impulse-response within federal entities
| 1999:01 to 2019:04 | Effects within entities | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| From FDI, NFDI and RFDI to | From INSEC to | From ITAE to | |||||||
| Federal entity (1) | FDI (2) | NFDI (3) | RFDI (4) | FDI (5) | NFDI (6) | RFDI (7) | FDI (8) | NFDI (9) | RFDI (10) |
| Aguascalientes1 | 7.08*** | 5.55 | 3.88*** | 1.08 | 0.59 | 0.47 | -0.21 | -0.19 | 0.60 |
| Baja California1 | 0.49 | 5.17*** | 4.48*** | 0.02 | 0.20 | 0.37 | 0.06 | -0.21 | 0.63 |
| Baja California Sur2 | 0.69*** | 0.71*** | 5.21*** | 0.07 | -0.00 | -0.23 | -0.01 | -0.03 | 0.57 |
| Campeche1 | 6.94*** | 4.98*** | 3.74 | -0.60 | -1.11*** | 0.04 | -0.04 | -0.94** | -0.66 |
| Mexico City3 | 3.32 | 4.04*** | 5.28*** | -0.25 | 0.13 | -1.94*** | -0.41 | -0.20 | -0.28 |
| Chihuahua1 | 2.7*** | 6.68*** | 4.86 | 0.12 | 0.11 | -0.02 | 0.12 | 0.69 | 0.56 |
| Chiapas1 | 5.36 | 6.05 | 5.30*** | 1.03** | 1.03 | -0.23 | -0.31 | 0.36 | 0.85 |
| Coahuila2 | 3.83 | 5.16 | 3.63 | 0.65 | 1.00** | -0.02 | 0.23 | 0.08 | -0.40 |
| Colima1 | 6.23 | 5.74*** | 3.11*** | -2.16*** | -0.51 | 0.37 | -0.02 | 0.93 | -0.22 |
| Durango2 | 5.86 | 6.45*** | 4.85*** | -0.41 | -0.67 | 0.21 | 0.65 | 1.63 | 0.37 |
| Guerrero1 | 5.50*** | 5.10*** | 4.65 | 0.03 | -0.09 | 0.10 | -0.46 | 0.25 | 0.82 |
| Guanajuato2 | 4.04*** | 3.47 | 6.11*** | 0.71 | 0.73 | -0.65 | -0.13 | 0.04 | 0.23 |
| Hidalgo1 | 6.30 | 7.38 | 3.58 | 0.79 | 1.22 | -0.44 | -0.25 | -0.49 | 0.25 |
| Jalisco1 | 5.38*** | 5.39*** | 6.90*** | -0.08 | -0.01 | 0.79 | -0.75 | -0.23 | -1.27** |
| State of Mexico2 | 0.83*** | 4.92*** | 5.03*** | -0.14 | 0.70 | -0.92** | -0.00 | 1.08*** | -0.49 |
| Michoacan1 | 6.22*** | 5.70*** | 5.46*** | -0.08 | -0.30 | 0.81 | -0.32 | -0.70 | -0.41 |
| Morelos2 | 6.78*** | 6.40*** | 4.82*** | 0.50 | 0.87 | 0.22 | -0.59 | -0.45 | 0.05 |
| Nayarit1 | 0.84*** | 2.30*** | 3.36*** | -0.00 | 0.14 | -0.85*** | 0.03 | 0.04 | -0.59 |
| Nuevo Leon2 | 6.01*** | 7.91*** | 4.49*** | 0.68 | -0.96 | 0.08 | 0.20 | -0.35 | 0.18 |
| Oaxaca1 | 6.12*** | 6.59*** | 6.44*** | -0.41 | -0.17 | 1.69*** | 0.42 | -1.03 | -0.41 |
| Puebla1 | 4.67*** | 5.49*** | 4.61*** | -1.39*** | 0.34 | -0.50 | -0.02 | 0.64 | 0.58 |
| Queretaro2 | 2.99*** | 4.98*** | 5.00*** | -0.08 | 1.11*** | -0.26 | 0.82*** | 1.12*** | 1.97*** |
| Quintana Roo2 | 2.91*** | 2.92*** | 3.87*** | -0.03 | 0.01 | -0.19 | -0.07 | -0.38 | -0.30 |
| Sinaloa2 | 4.58*** | 5.64*** | 5.70*** | 0.12 | 1.32*** | -0.73 | 0.55 | -0.44 | -0.22 |
| San Luis Potosi1 | 4.51*** | 5.75*** | 6.03*** | 0.14 | 0.57 | 1.76*** | 0.76 | 1.43*** | 1.56*** |
| Sonora2 | 6.11*** | 7.4*** | 6.97*** | 0.92 | 0.72 | 0.98 | -0.34 | 0.92 | 1.13 |
| Tabasco1 | 6.46*** | 5.80*** | 6.13*** | -1.95*** | -0.61 | -0.86 | -0.21 | -0.77 | 0.23 |
| Tamaulipas1 | 0.76*** | 7.31*** | 5.09*** | -0.01 | -0.82 | 0.09 | -0.08 | -0.85 | -0.33 |
| Tlaxcala2 | 4.64*** | 5.03*** | 5.08*** | 0.11 | 0.05 | 0.83 | -0.08 | 0.73 | -0.83 |
| Veracruz1 | 5.16*** | 6.17*** | 4.75*** | 0.48 | -0.16 | 0.06 | -0.06 | 0.20 | 0.57 |
| Yucatan1 | 3.74*** | 5.32*** | 5.60*** | 0.01 | -0.05 | -0.45 | 0.11 | 1.11** | -0.57 |
| Zacatecas1 | 6.77*** | 6.06*** | 5.49*** | -1.39** | -0.84 | -0.44 | 0.77 | -0.78 | 0.23 |
**Indicates 10% statistical significance
***Indicates 5% statistical significance
Numbers following the name of the entity in column 1 indicate: 1 for 1-lag test; 2 indicates a 2-lag test, and 3 indicates a 3-lag test.
Source: author’s based on data from Inegi (2021) and Secretaría de Economía (2021).
Conclusions
Foreign Direct Investment (FDI) had a differentiated effect on federal entities in Mexico. This work has shown, through the spatial VAR technique, that reinvestments in the neighbor states largely determine reinvestment in a particular state, generating spillovers throughout the different states in the country. Additionally, performance of new investments is determined by past performance of new investments within the state; this is, if a state received new FDI in the past, its likelihood of seeing more new investment increases, as it was found for Hidalgo, Nuevo León, Sonora and Tamaulipas.
Stylized facts showed that neither economic growth nor insecurity are major determinants of new investments or reinvestments, but these are explained by other factors. In other words, both insecurity and economic growth have failed to be relevant variables to explain FDI.
We have also shown that FDI generates benefits both within the state and to its neighbors, and that total foreign direct investment, reinvestment and new investment are mostly explained by their own lags, but also by past performance of these investments in neighboring entities.
It is worth noting that the states’ own lags throughout time are larger than neighbors’ lags or the effect that a state has over its neighbors. Regarding the effect of variables themselves in time, which have an influence on future investment within entities, we found a 5.3% for new investment, 4.8% for reinvestment and 4.3% for total investment. This implies that past performance is more important for new investments within states than it is for neighboring states.
Average push-out spillovers of states towards their neighbors are larger in reinvestments (2%) than in new investments (1.6%) or total foreign direct investments (0.6%). However, push-in spillovers are much larger than push-out spillovers; for instance, when reinvestment of a neighbor changes, the spillover in the state is of 2.7%, higher than 0.7%, which is the spillover of a state on its neighbors. These results match those of Márquez et al., (2014; 2010) and Andrés-Rosales et al., (2021) in their effects. On the other hand, when new investment of a neighbor changes, the push-in spillover is 2.3%, also 0.7% higher than a state’s push-out spillover on neighbors. Finally, the effect of an increase in total FDI in a state is 1% over its neighbors total FDI. What we can conclude is that reinvestment has had a larger effect among states than new investment. These results are hardly comparable with other studies because the SpVAR has been scarcely used for FDI research, but they are somewhat comparable with other research topics that have used the same technique.
With our research, we have shown that both economic growth and insecurity have not largely affected total FDI flows, as also found by Ashby and Ramos (2013) in the sense that companies have learnt to cope with insecurity and are still investing in different states. However, contrary to these authors, we only found a Granger causality in some entities for the Mexican economy. When quantifying these effects in percentage terms and determining the effect of insecurity on FDI, we did not find them significant, and in those entities where the impact was positive, this was less than 2.4% as observed for Tamaulipas, or 1.7% for the State of Mexico. In the rest of the entities, we found no evidence of what the authors found, and this is what is special about spatial analysis, that it cannot be generalized to the whole country, which is what macroeconomic studies do, while spatial analysis quantifies in a more punctual way the behavior of the problem in question.
Madrazo (2009) considers a hypothetical scenario: if Mexico had a zero-homicide rate for every 100 thousand inhabitants during the period 1998-2006, it would have received around 94 extra dollars per capita in FDI each year or around 9,396,720 extra dollars in FDI each year for any city with a population of over 100,000. Nevertheless, we have shown in this study that this amount would be off the mark, given that if a city keeps its insecurity under control, but its neighbors do not, this will have a negative influence over FDI flows to the city. We agree with Ramos and Ashby (2013) in that it is likely that foreign investors have heterogeneous capabilities to evaluate and cope with the high levels of organized crime in host states, which influences the likelihood of investing in them. This may explain the positive effects of insecurity on FDI in some states, but this is not the rule for every entity.
Finally, it is noteworthy that the present research could be analyzed at sectoral and municipal levels, which ould improve the study, since the state level continues to be an aggregate level and treating economic sectors altogether repeats the error of macroeconomic analysis. This will be done in future studies to understand the nature and performance of investment in the country in a better way, and to be able to recommend public policies that address these issues.
