In this paper, we use trends in GDP gaps per capita or relative GDP per capita indicator as an aggregate alternative measure to show regional per capita income inequality between the states of Mexico. The indicator compares each state's gross domestic product per capita (GDPpc) with the annual average of each year and 2019; this allows for evaluating growth performance before and after the COVID-19 pandemic. The econometric method is a spatial dynamic panel, and the convergence–divergence model is conditioned to the behavior of the fiscal policy followed in each state: The fiscal policy indicators are current expenditure and public investment. In general, it is concluded that the amounts of current spending and public investment, in an active fiscal policy context, tend to favor states with higher relative per capita income, which is why an increase in regional inequality is observed. However, it tends to affect low-income regions more, with higher ratios of current spending and investment to GDP, which reduces regional inequality. Therefore, the short-term effects of active fiscal policy to reduce regional disparities can be neutralized.
Notes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
This work analyzes the role of fiscal policy in the process of greater inequality in regional per capita income in Mexico from 1989 to 2021. The divergence (inequality) of per capita income by state in Mexico increased with the trade liberalization of the mid-eighties, which promoted the accelerated growth of the north and center of the country while causing the stagnation of the poorest states located in the south (Acevedo and Medina 2001; Díaz-Bautista 2003; Diaz et al. 2017; German-Soto et al. 2020; Rodríguez et al. 2021); in 1994 the ratio of the gross domestic product per capita (GDPpc) of Mexico City compared to that of Chiapas was 4, by 2019 it had reached a maximum of 7.4. At the same time, current expenditures have been the most critical component of total fiscal expenditures, reducing the amount and relevance of public investment in regional economies, especially in states with lower per capita income. In the state of Chiapas, current spending was 26.4%, and public investment was 0.9% as a percentage of gross domestic product (GDP) in 2021.
The federal rules for allocating resources to the regions in Mexico result in the most significant allocations of the components of regional fiscal spending (current and investment) in the states with a higher per capita income, while they tend to be lower in the poorest states. The allocation of current and investment spending in the richest states has caused per capita income inequality between states to increase.
Advertisement
The picture is different if we focus on the importance of current spending and public investment in each of the regions, measured by their ratio to GDP, where the regularity shows that these ratios to GDP of current expenditure or public investment tend to be higher (lower) in regional economies with lower (higher) per capita income. From this, we infer that an active fiscal policy will have a more significant effect on per capita income in the poorest regional economies due to a higher ratio of fiscal expenditure to GDP, which would reduce inequalities in per capita income between states in Mexico.
The literature on the subject has marginalized the discussion of the role of fiscal policy in the processes of economic convergence or divergence; the predominant view has been the neoclassical one, under which public spending is distorting and tends to generate inequality because it increases growth but reduces the taxes necessary to finance it (Barro 1990, 1991; Sala-i_Martin 2000).
In this research, we use a novel method to analyze the role of fiscal policy in Mexico, the dynamic spatial panel. This method has been recognized to be superior to others, given its predictive capacity that incorporates space and time (Billé et al. 2023). It also allows us to address a deficiency of the usual growth models, which, assuming independence between the regions studied, have a bias of omitting a relevant variable by not considering the spatial interdependence of the observation units in time. Another novel aspect of this work on convergence and divergence studies in Mexico is the construction of an indicator for the gap’s GDP per capita (relative GDP per capita) to measure income inequality and for the analysis of the relative dynamics (Rey 2001; Rey and Sastré 2015). The relative GDP per capita indicator is built as a double ratio of GDP, first about the average for 2019 to use the COVID-19 pandemic as a comparison threshold and, second, to the average of each year. This double relationship of GDP per capita operates as a measure of inequality in two ways: One analyzes the dispersion of income for the COVID-19 pandemic, and another uses the increasing or decreasing trend of the regional GDP per capita gap to show changes in regional income inequalities. It is important to note that the trends of the gaps in GDP per capita or the relative per capita indicator are an alternative measure to show income inequality considering the factorial components (salaries of employees, gross operating surplus, taxes, and subsidies) among the states (Mendoza and Villagra 2023), and therefore, it is an aggregate approach in comparison with the traditional measures that make use of income inequality of households (Hlasny 2021; Scott et al. 2017) or with variants in which is used a spatial version of the Gini index (Quintana and Salas 2023).
The document is structured in four sections. The first section discusses the theoretical basis of the inequality analysis and fiscal policy's role. The second section presents the stylized data. The third section specifies the dynamic spatial panel model used and its results. In the last section, we offer final considerations.
Advertisement
2 Theoretical discussion
Spatial or territorial inequality has been recently addressed by a good number of studies that consider that, within countries, there are regions “left behind” (Martin et al. 2021). This phenomenon has affected traditionally disadvantaged regions and those once very prosperous. In this line, Autor et al. (2013) find that previously prosperous areas of the USA have suffered severe job losses and wage declines due to competing with Chinese imports. Nunn et al. (2018), using data from more than a century in Europe, also confirm this type of process and conclude that regional inequality operates in a traditional U-shaped way, which coincides with what was found by Piketty and Saez (2003). The same is true for Latin America, as stated by Gasparini and Cruces (2021), who warn that this is the most unequal region in the world.
Faced with the processes of inequality, Solow's neoclassical model of economic growth (1956) came back in the economic literature to try to establish a convergence process between rich and poor regions in the long run. This triggered many empirical studies in the 1990s to verify the existence of such a hypothesis (Barro and Sala-i-Martin 1995). The first results showed the possibility of beta convergence when poor regions grow more than rich ones and sigma convergence when the dispersion of per capita income between said regions tends to decrease (Sala-i-Martin 1996). This idea was questioned, especially that of beta convergence, because it assumes that all regions tend to converge to the same stationary state, giving rise to the conditional convergence concept (Mankiw et al. 1992).
By postulating that the stationary states are not the same for rich and poor regions, the conditional convergence hypothesis proposes evaluating the correlation of the growth rate with the initial per capita income levels, but conditional on the different stationary states. How conditionality is achieved is not very clear, and for this purpose, trial and error tests have been used with numerous proxy variables for the stationary states, such as investment rates (Cho 1996), demographic factors (Kelley and Schmidt 1995), foreign direct investment, commercial liberalization (Dollar and Kraay 2003), democracy and governance (Rivera-Batiz 2002), and human capital (Mankiw et al. 1992; Islam 1995) among many other.
One aspect to highlight is that conditional models have tended to minimize the government's importance in the growth and convergence processes. They have considered that the size of the government does not have much significance, and they have even found a negative relationship between public spending and growth (Landau 1983; Barro 1991). However, when the role of the government is taken into account through fiscal transfers, the results are not negative, for example in the European case, (Checherita et al. 2009) find a convergence process due to net fiscal transfers, which they call “impoverishing” because the poor regions that receive fiscal transfers are less impoverished than the rich regions that pay for these transfers, there is then a redistributive effect through fiscal policy.
Fiscal policy can positively affect income redistribution and indirectly affect growth and the economic cycle (Capella-Ramos et al. 2020). However, it can also have distorting effects known as the “beggar-thy-neighbor effect,” “moral hazard,” and “flypaper effects,” whose magnitudes depend on the design of transfers, institutional quality, and technological capacity (Capella-Ramos et al. 2020). There is also evidence that fiscal decentralization tends to drive convergence processes, while intergovernmental transfers that seek to cover the income gap with their own resources and expenses increase regional inequalities (Stossberg et al. 2016).
At the international level, the results of the relationship between fiscal policy and growth are not conclusive in demonstrating positive effects (Cashin 1995; Devarajan et al. 1996; Bleaney et al. 2001). Only when fiscal spending is disaggregated is positive evidence found in some cases, such as education, infrastructure, and public protection (Benos 2008). Generally, it is considered that fiscal transfers have a specific aim and not only the general objective of redistributing income, but they also impact real convergence (Capella-Ramos et al. 2020).
Despite the relevance of fiscal policy in convergence processes, studies of its effect in the Mexican case are practically nonexistent. Mexican regional convergence studies have focused on exposure to foreign trade due to the North American Free Trade Agreement, where the states most exposed to globalization accelerate their convergence rate (Acevedo and Medina 2001; Díaz-Bautista 2003; Diaz et al. 2017; German-Soto et al. 2020; Rodríguez et al. 2021). Some convergence studies focus on closing productivity gaps (Rodriguez-Gamez and Cabrera-Pereyra 2019; López and Cermeño 2016; Asuad and Quintana 2007, 2010; Esquivel and Messmacher 2002; Carrion-i-Silvestre and Germán-Soto 2009; Cabral and Mollick 2012; Rodriguez-Gamez et al 2019). In the few studies on the effects of fiscal policy in Mexico, the results of Asuad et al. (2007) show a negative relationship between the GDPs per capita of the country's states and federal transfers, which confirms an effect of productive inefficiency. On the other hand, Andrés-Rosales et al. (2021) find that public spending does not influence regional convergence processes due to the sharp reduction in public investment in the period 1999–2019.
This work analyzes the impacts of fiscal policy from a new perspective, in which fiscal transfers to the states are approximated by current spending and public investment. This means that not all spending or all fiscal transfers to states impact growth. Another distinctive aspect regarding previous studies is that the dependent variable of the convergence models is built as a double ratio for the average of the states in 2019 and weighted to the average of each year in the period 1989–2021, which allows the expression of regional convergence or divergence both to the pre-Covid-19 pandemic period as well as concerning the states averages for each year.
3 Data and measurements
With the State and Municipal Public Finance statistics of the INEGI, three fundamental indicators were estimated: fiscal net total spending (gnc), current spending (gc), and public investment (inv). The fiscal net total spending is total gross expenditures minus the cost of public debt; current spending is composed of personal services, materials and supplies, general services, transfers, assignments, subsidies and other aid, resources assigned to municipalities, and other expenditures; and public investment includes movable, immovable and intangible assets, public investment, and financial investments and other provisions; it is essential to clarify that the public investment of the states is, as a general rule, a more minor component than the federal public investment. The fiscal indicators were originally in thousands of current pesos. They were converted to billions of constant 2013 pesos with the implicit deflators of the GDP from the accounts by the states of INEGI. The ratios of the fiscal indicators to GDP were computed by dividing by the GDP in billions of constant 2013 pesos per state from the accounts per federal entity of the INEGI and multiplying by one hundred. GDP per capita was obtained with GDP in billions of 2013 pesos divided by the total population of each state.
The analysis does not consider the cases of two states, Tabasco and Campeche, since the country's main oil infrastructure is installed there, which biases the growth data, though this does not mean that they directly benefit from the oil revenues generated (Esquivel and Messmacher 2002; Rodríguez et al. 2021; Mendoza and Viallagra 2023).
To estimate the income inequality variable, we first follow the procedure suggested by Rey (2001) and Rey and Sastré (2015) to estimate relative GDP per capita. Firstly, we compute the GDP per capita relative to the average for the year 2019 (\({\text{rgdp}}_{2019it}\)):
where the average for 2019 is \({\text{mgdp}}_{it = 2019} = 1/30\mathop \sum \limits_{i = 1}^{30} {\text{pcgdp}}_{it = 2019}\) and the average for each year of the first weighting is obtained as \({\text{mrgdp}}_{it} = 1/30\mathop \sum \limits_{i = 1}^{30} {\text{rgdp}}_{2019it}\).
The year 2019 was taken as a comparison point to consider the situation of the states before the 2020 pandemic and thus to be able to evaluate the impact of COVID-19.
At the same time, the inequality indicator allows us to evaluate how the GDP per capita data deviates from the average in each year, which accounts for the inequality or convergence concerning the average performance.
Table 1 shows the variables estimated for each indicator. The data highlight the slow economic growth of the states in the period from 1989 to 2021 and the decline in 50% of the states that present negative growth rates for the period. Regarding the direction of total net fiscal spending and current fiscal spending as a proportion of GDP, the data show that it has been channeled mainly toward the poorest regions of the Mexican southeast, where Chiapas, Guerrero, and Oaxaca, which are states with sharp drops in their economic performance, present increases in GDP ratios from 20 to 27.3% in relative total net fiscal expenditure and from 19.6 to 26.4% in relative current fiscal spending.
Table 1
Regional fiscal spending and relative GDP per capita, 1989–2021
Source: Authors with different data sources from INEGI
State
Relative GDP per capita (rgdp)
Average growth
Ratio of fiscal net total spending to GDP (gn_gdp)
Average growth
Ratio of fiscal current spending to GDP (gc_gdp)
Average growth
Ratio of public investment to GDP (inv_gdp)
Average growth
1989
2021
1989–2021
1989
2021
1989–2021
1989
2021
1989–2021
1989
2021
1989–2021
Basic statistics
Aguascalientes
0.8
1.1
1.3
2.8
8.2
6.3
2.2
7.4
7.7
0.6
0.8
1.1
Baja California
1.5
1.4
− 0.4
4.8
7.4
1.8
4.7
7.3
1.8
0.1
0.2
1.3
Baja California Sur
1.5
1.4
− 0.1
1.9
8.3
10.8
1.8
8.2
11.6
0.1
0.1
− 0.4
Coahuila
1.3
1.4
0.2
0.5
2.5
12.5
0.4
2.4
15.7
0.1
0.1
0.4
Colima
1.2
1.0
− 0.5
1.9
12.1
17.0
1.6
11.9
20.3
0.3
0.2
− 0.8
Chiapas
0.6
0.4
− 1.3
2.7
27.3
29.2
2.3
26.4
33.6
0.4
0.9
4.0
Chihuahua
0.8
1.2
1.4
1.7
8.0
11.6
1.4
7.9
15.2
0.4
0.1
− 2.2
Mexico City
2.1
2.6
0.9
2.6
5.8
3.9
2.1
5.4
5.3
0.6
0.3
− 1.4
Durango
0.8
0.8
0.0
1.6
12.3
21.4
1.4
11.6
23.8
0.2
0.7
7.0
Guanajuato
0.7
0.9
0.7
1.3
9.0
19.4
0.9
8.7
27.8
0.4
0.4
− 0.1
Guerrero
0.6
0.5
− 0.7
2.8
21.4
21.5
2.7
21.0
22.0
0.1
0.4
8.7
Hidalgo
0.9
0.7
− 0.8
0.9
13.0
43.4
0.8
12.6
45.4
0.1
0.4
18.1
Jalisco
1.2
1.1
− 0.1
2.6
7.2
5.7
2.4
6.6
5.8
0.2
0.6
4.3
Mexico State
0.8
0.7
− 0.4
1.5
14.0
26.0
1.4
13.0
26.6
0.1
1.0
19.6
Michoacan
0.7
0.7
0.0
1.2
13.8
33.4
1.0
13.0
37.1
0.2
0.8
11.2
Morelos
0.9
0.7
− 0.5
1.8
12.4
18.6
1.8
12.1
18.8
0.1
0.2
9.4
Nayarit
0.9
0.7
− 0.7
2.9
14.8
13.1
2.5
14.5
15.9
0.5
0.3
− 1.3
Nuevo León
1.6
1.9
0.4
3.9
5.0
0.9
3.5
5.0
1.4
0.5
0.1
− 2.7
Oaxaca
0.6
0.5
− 0.6
22.2
20.0
− 0.3
22.0
19.6
− 0.3
0.3
0.4
1.5
Puebla
0.6
0.6
0.1
1.8
12.9
19.8
1.4
12.4
25.2
0.4
0.5
0.7
Queretaro
1.1
1.3
0.7
1.5
7.2
12.7
1.2
7.1
15.6
0.2
0.1
− 1.9
Quintana Roo
1.2
1.2
− 0.1
1.7
9.3
14.8
1.6
9.0
14.6
0.0
0.3
24.4
San Luis Potosi
0.8
1.0
0.5
1.3
8.9
19.2
0.9
8.6
28.4
0.4
0.2
− 1.4
Sinaloa
1.0
0.9
− 0.2
1.7
10.6
16.7
1.3
10.2
21.5
0.4
0.4
0.1
Sonora
1.4
1.5
0.2
2.1
7.4
8.0
1.2
7.2
15.4
0.9
0.2
− 2.6
Tamaulipas
1.1
1.0
− 0.3
1.8
9.0
13.0
1.3
8.2
17.2
0.5
0.8
1.9
Tlaxcala
0.8
0.5
− 1.2
2.9
17.1
15.6
2.2
16.5
21.1
0.7
0.6
− 0.5
Veracruz
0.9
0.7
− 0.8
1.9
13.1
19.1
1.5
12.7
24.7
0.4
0.4
− 0.2
Yucatan
0.8
0.9
0.5
2.0
10.2
13.7
1.6
10.0
16.8
0.3
0.2
− 1.0
Zacatecas
0.7
0.7
0.3
2.1
14.0
18.8
1.4
13.9
28.2
0.6
0.2
− 2.2
The states of Campeche and Tabasco are not included due to their oil profile that biases regional trends
What contrasts most in the data in Table 1 is the slow dynamic of public investment and its sharp decline in two of the country's most economically significant states, Mexico City and Nuevo León. The best performance of public investment happened in Quintana Roo, Mexico State, and Hidalgo; in the former, the current government is carrying out one of its largest investment projects—the so-called Maya Train—and in the other two states, there are important industrial relocation processes.
3.1 Regional relative GDP per capita trend, 1989–2021
We calculated the standard deviation of the relative GDP per capita to obtain an indicator for sigma-type convergence. The data in Fig. 1 confirm that between 1989 and 2021, there is a straightforward process of divergence or increase in dispersion between the country's states. An important fact is that the decrease in short-term inequality occurs in periods of economic crisis: the banking crisis in 1994, the economic slowdown of 2001, the financial crisis from 2007 to 2009, and the crisis caused by the COVID-19 pandemic in 2020. This situation shows that income levels deteriorate during the global economic crisis, especially for the wealthiest states linked to the international economy, which would explain part of the decrease in inequality; the fiscal policy would explain another part following a counter-cyclical process from 2009 onward.
Fig. 1
Fiscal spending and regional income inequality trends, 1989–2021.
Source: Authors with different data sources from INEGI. Notes: The ratio of fiscal net total spending to GDP is calculated as the ratio of the sum of the fiscal net total spending (net of debt payment) to the sum of the GDP per state. The states of Campeche and Tabasco are not included in the data because their oil activity causes biases in regional trends
×
In the country's states, sigma convergence and divergence processes occur, as shown in Fig. 2. There are two very contrasting groups of states: that of the wealthiest states, such as Mexico City and Nuevo León, which, by moving above the average, cause greater inequality, and the poorest states in the country, Chiapas and Oaxaca, which by reducing their per capita GDP also increase inequality, the latter being the primary beneficiaries of fiscal transfers. The data also suggest that the relationship between fiscal spending and inequality is fragile, as shown in Fig. 3. Mexico City and Chiapas are two states that receive high fiscal expenditures. However, the GDP per capita in Mexico City is very high, while in Chiapas, it is very low. As can also be seen in Fig. 3, the fiscal expenditure that Mexico City has received is an insignificant proportion relative to its GDP. At the same time, it represents close to 30% of the Chiapas product. The same holds if GDP per capita is used instead of total GDP. This behavior seems to suggest that, in the largest states, fiscal transfers are less significant in relation to their productive capacity and, therefore, contribute less to the reduction of inequality. In contrast, for the most lagging states, such transfers are very relevant and tend to have a positive impact on reducing inequality.
Fig. 2
Regional relative GDP per capita trend, 1989–2021.
Source: Authors with different data sources from INEGI. Note: The states of Campeche and Tabasco are not included in the data because their oil activity causes biases in regional trends
Fig. 3
Regional fiscal spending and relative GDP per capita, 1989–2021.
Source: Authors with different data sources from INEGI. Note: The states of Campeche and Tabasco are not included in the data because their oil activity causes biases in regional trends
×
×
It is important to note that in Mexico, the tax system is highly centralized, which is why the revenues of the country's states are meager compared to the transfers received from the federal government. This results in a very heterogeneous fiscal structure in which large states, such as Mexico City, have a greater capacity to generate revenues. In contrast, poorer states such as Oaxaca, Chiapas, and Guerrero depend more on federal fiscal transfers. Stossberg et al. (2016) have argued that greater decentralization reduces inequality, especially regarding social spending.
4 Spatial dynamic panel models
To understand the analytical contribution of the spatial lag dynamic panel model is essential to consider the dynamic panel model (Eq. 3), which includes the temporal lag of the endogenous variable (time lag), this makes it possible to analyze the short- and long-term effects of fiscal policy over time; in the spatial Durbin panel model (Eq. 4), the matrix of spatial weights (W) will be incorporated into the endogenous variable and the fiscal policy variables that are used as exogenous variables (spatial lag). With the spatial lag of the endogenous variable, it is possible to analyze spatial externalities of regional inequality toward close neighbors. In contrast, with the spatial lag in exogenous variables, the neighborhood effects of fiscal policy variables are captured. The spatial lag panel model is considered a particular case of the spatial Durbin panel model because it does not consider the spatial lag of fiscal policy variables.
In all specifications, we use lowercase letters to represent natural logarithms; \(y_{it}\) is the relative per capita GDP \(({\text{rgdp}}_{it} )\), and \(x_{it}\) is the vector of fiscal spending variables such as fiscal current spending \(({\text{gc}}_{it} )\), public investment \(({\text{inv}}_{it} )\), and their respective ratios to GDP \({\text{gc}}\_{\text{gdp}}_{it}\) and \({\text{inv}}\_{\text{gdp}}_{it}\).
In the analysis of convergence conditional on fiscal policy, the econometric models are based on the fundamental growth equation of Solow (1956) and Swan (1956), but in its modified form to incorporate fiscal policy variables (Tanchev and Mose 2023), it could be expressed as follows:
\(y_{i,t - 1}\) is the endogenous variable lagged in time.
\(x_{i,t - 1}\) is the fiscal policy variables lagged in time, to avoid endogeneity problems.
\(\mu_i\) are the fixed effects in the spatial units.
\(\alpha\) is a constant.
\(u_{i,t}\) is the random disturbance term.
\(i,t\) are the i spatial units at time t.
To analyze the role of fiscal policy in explaining relative inequality per capita (GDP) by state in Mexico from 1989 to 2021 (\(y_{it}\)), a modified version of Eq. (3) is used through the spatial dynamic panel models methodology with fixed effects in their spatial lag and spatial Durbin versions. These spatial models offer the advantage of dealing with the serial dependence of spatial units in time, the spatial dependence between observations in time, unobservable spatial and temporal effects, and problems of endogeneity of the regressors (Elhorst 2012).
Returning to Elhorst (2012), the general non-dynamic in time but dynamic in the space form of model (3) is the spatial Durbin panel (SPDM) and is expressed as follows:
We will use the dynamic specification of the SPDM that incorporates a term of spatial lag for the endogenous variable \((y_{it} )\) with the parameter \(\rho\) and a matrix of spatial weights \(W\) based on the five K close neighbors to measure spatial externalities to neighbors; the time lag for the endogenous variable \((y_t )\) with the parameter \(\tau\); to measure short- and long-term effects over time; the time lag vector of fiscal spending variables \((x_{t - 1} )\) to avoid problems of endogeneity of fiscal policy and its parameters \(\beta_k\) for each of the four fiscal variables to measure its effects on the relative inequality; the spatial lag of the fiscal spending variables \((Wx_{it - 1} )\) and its \(\theta_k\) shows the importance of the fiscal policy applied to the neighbors and their effects on the relative inequality; the general constant \((a)\) and the individual \(\mu_i\) estimated with the fixed effects procedure. This specification is represented in Eq. (5).
With the spatial Durbin dynamic panel model, we get the spatial effects (direct, indirect, and total) of the fiscal expenditure variables \((x_{it} )\) about the regional per capita income inequality \(({\text{rgdp}}_{it} )\) in the short term and the long term, with average of the row sums of the non-diagonal elements of the matrix (LeSage 2014):
For the spatial lag dynamic panel model, the short and long effects are obtained by eliminating \(\theta_k W\) of Eqs. 7 and 8.
The results of the spatial lag dynamic panel model (Eq. 6) and spatial Durbin dynamic panel model (Eq. 5) are presented in Table 2. Results show that all the fiscal policy variables are significant in both model versions and only differ in signs. In general, the weight of fiscal spending in GDP and the weight of investment in GDP tend to reduce inequality between the states of the country in the long term, a situation that confirms the graphic pattern that we saw before since the weight of these components in GDP increases, the gap between the poorest and wealthiest states is reduced.
Table 2
Results of spatial dynamic panel models, 1989–2021
Source: Models estimated with programming in RStudio using the library SDPDmod (Spatial Dynamic Panel Data Modeling).
Name
Variable
Spatial lag dynamic panel
Spatial Durbin dynamic panel
coefficients
Std. Error
Pr( >|t|)
coefficients
Std. Error
Pr( >|t|)
Relative per capita GDP: spatial lag
Rho-W*log(rgdp)(t)
− 0.017
0.013
0.19
0.204
0.014
0.00
Relative per capita GDP: time lag
log(rgdp)(t − 1)
2.277
0.014
0.00
1.869
0.015
0.01
Fiscal current spending
log(gc)(t − 1)
− 0.073
0.003
0.00
0.006
0.006
0.01
Public investment
log(inv)(t − 1)
− 0.030
0.003
0.00
− 0.046
0.003
0.00
Ratio fiscal current spending to GDP
log(gc_gdp)(t − 1)
0.130
0.004
0.00
0.012
0.007
0.01
Ratio public investment to GDP
log(inv_gdp)(t − 1)
0.042
0.005
0.00
0.069
0.005
0.00
Spatial lag
Fiscal current spending
W*log(gc)(t − 1)
− 0.048
0.008
0.01
Public investment
W*log(inv)(t − 1)
− 0.031
0.007
0.01
Ratio fiscal current spending to GDP
W*log(gc_gdp)(t − 1)
0.084
0.010
0.01
Ratio public investment to GDP
W*log(inv_gdp)(t − 1)
0.024
0.010
0.01
R-square
0.85
0.85
Adjust-R-square
0.78
0.81
Impact (short term)
Direct
Indirect
Total
Direct
Indirect
Total
Fiscal current spending
log(gc)
− 0.074
0.001
− 0.072
0.004
− 0.057
− 0.053
Pr( >|t|)
0.00
0.18
0.00
0.52
0.00
0.00
Public investment
log(inv)
− 0.030
0.000
− 0.029
− 0.047
− 0.050
− 0.097
Pr( >|t|)
0.00
0.18
0.00
0.00
0.00
0.00
Ratio fiscal current spending to GDP
log(gc_gdp)
0.130
− 0.002
0.128
0.015
0.105
0.121
Pr( >|t|)
0.00
0.18
0.00
0.02
0.00
0.00
Ratio public investment to GDP
log(inv_gdp)
0.042
− 0.001
0.042
0.070
0.047
0.118
Pr( >|t|)
0.00
0.18
0.00
0.00
0.00
0.00
Impact (Long term)
Direct
Indirect
Total
Direct
Indirect
Total
Fiscal current spending
log(gc)
0.058
0.001
0.058
− 0.008
0.047
0.039
Pr( >|t|)
0.00
0.18
0.00
0.31
0.00
0.00
Public investment
log(inv)
0.023
0.000
0.023
0.052
0.020
0.072
Pr( >|t|)
0.00
0.18
0.00
0.00
0.00
0.00
Ratio fiscal current spending to GDP
log(gc_gdp)
− 0.102
− 0.001
− 0.103
− 0.011
− 0.078
− 0.090
Pr( >|t|)
0.00
0.18
0.00
0.15
0.00
0.00
Ratio public investment to GDP
log(inv_gdp)
− 0.033
0.000
− 0.033
− 0.079
− 0.008
− 0.087
Pr( >|t|)
0.00
0.18
0.00
0.00
0.39
0.00
The states of Campeche and Tabasco are not included in the data because their oil activity causes biases in regional trends.
On the other hand, current spending and public investment negatively affect inequality in the short term and increase it in the long term, which suggests that investment policy can be an effective instrument to reduce gaps between states only in the short term.
The tau coefficient \((\tau )\) of the dependent variable lagged in time could be interpreted similarly to a convergence coefficient; said indicator is positive and significant, which is indicative of the permanence of the divergence over time and in the long term.
The spatial lag is not significant in the panel model with spatial lag, but it is in the Durbin and is positive and accounts for spillover effects. The latter means that when a state tends to deviate positively (negatively) from the mean, neighboring states deviate in the same (opposite) direction. It is the same situation for the relative coefficients of fiscal spending and investment; inequality between its neighbors increases when their weight grows in a state. This problem is different when only the absolute value of public spending and public investment is considered, which, when increased in a state, tends to generate decreases in inequality in neighboring states.
Using the same data in Table 2, it is possible to analyze the fiscal policy variables' short- and long-term impacts. The first relevant aspect is that in the short term, increases in fiscal spending and investment tend to reduce inequality, but in the long term, they lose effectiveness and increase it. For their part, relative fiscal spending and relative public investment indicate that when the weight of these components in GDP increases, there is an increase in inequality in the short term, but it tends to reduce in the long term.
The results show a heterogeneous behavior in the impact of fiscal policy on regional inequality. On the one hand, more fiscal transfers and more significant amounts of investment have a positive impact on reducing inequality. However, their effect is lost in the long term to the extent that poorer states receive more resources but tend to have poorer productive performance over time. However, inequality tends to reduce in states where fiscal transfers and investment account for a larger share of their income in the long run. The total impacts fail to reduce inequality between states because, in the end, the long-term effects, which could drive a process of convergence, are less than the divergence effects that operate in the short term, which weakens the impact of an active fiscal policy.
4.1 Final considerations
With trade liberalization, the long-term process of regional convergence since the 1940s that characterized the Mexican economy reverted to a process of accelerated regional divergence determined by more significant growth in GDP per capita in the northern and central states and stagnation in the southern region (Acevedo and Medina 2001; Díaz-Bautista 2003; Diaz et al. 2017; German-Soto et al 2020; Rodríguez et al. 2021). Although some research has suggested that fiscal policy positively affects income redistribution (Capella-Ramos et al. 2020) and can significantly accelerate the convergence process when specific fiscal mechanisms are identified (Capella-Ramos et al. 2020), in reality, there is a debate, because the relationship between fiscal policy and growth, as in convergence processes, is not conclusive (Cashin 1995; Devaraja et al. 1996; Bleaney et al. 2001). In the case of Mexico, Asuad et al. (2007) find a negative relationship between the GDP per capita of the country's states and federal transfers, which confirms the effect of productive inefficiency (Barro 1991). Meanwhile, Andrés-Rosales et al. (2021) show that public spending does not influence regional convergence processes due to the sharp reduction in public investment from 1999 to 2019.
This research analyzes the role of fiscal policy, not during the regional convergence identified from 1940 to 1986 but in the process of greater inequality in regional per capita income in Mexico from 1989 to 2021. In general, we assumed that the fiscal policy followed in each state is the main condition of the divergence processes in the country and has also been the explanatory factor of the greater state resilience in the face of the two most recent crises, the financial crisis in 2009 and the Covid-19 pandemic in 2020. To analyze fiscal policy's effects on regional divergence in Mexico, we used the dynamic spatial panel method that incorporates space and time (Billé et al. 2023). To identify the relative economic dynamic, we applied a double ratio of GDP per capita, first about the average for the year 2019 to take the Covid-19 pandemic as a comparison threshold and, second, to the average of each year from 1989 to 2021.
From the results of the trends in the performance of GDP per capita and the application of dynamic spatial panel models, the impact mechanisms of fiscal policy on regional per capita income inequality in Mexico are different when we analyze the fiscal indicators—in amounts or ratios—of GDP for each state. More significant amounts of current expenditure and public investment tend to favor states with a higher relative per capita income, which is why an increase in regional inequality is observed. But suppose the ratios of current spending and investment to GDP tend to be higher in the poorest states. In that case, an active fiscal policy tends to affect low-income regions more, so a reduction in regional inequality is observed. However, these convergence effects are less than those that operate at divergence in the short run, neutralizing them. Therefore, an active fiscal policy would have to be accompanied by a fiscal decentralization policy that would strengthen the endogenous capacity of the poorest states to increase their revenue generation.
This research did not analyze two analytical and methodological aspects that are future research lines. The first concerns the lack of endogeneity of public spending and investment, which are resources for the states that are transferred by the federation. The second relevant aspect concerns the aggregate focus of this article and that in future research we will address in terms of the components of public spending aimed at health, education, and combating poverty, which are relevant to reducing regional inequalities.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.