LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 608.
DOI: https://doi.org/10.56712/latam.v6i5.4619
An estimation of the nominal exchange rate level that would
eliminate the mexican currency misalignment: is it moving in
the right direction in 2025?
Una estimación del tipo de cambio nominal que eliminaría del
desalineamiento cambiario del peso mexicano: ¿se está moviendo en la
dirección correcta en 2025?
Adrián Jiménez Gómez
adrian.jimenez@correo.buap.mx
https://orcid.org/0000-0002-8909-9056
Benemérita Universidad Autónoma de Puebla
Puebla – México
Héctor Flores Márquez
hector.flores@correo.buap.mx
https://orcid.org/0000-0002-1766-5266
Benemérita Universidad Autónoma de Puebla
Puebla – México
Carlos Absalón Copete
carlos.absalon@correo.buap.mx
https://orcid.org/0000-0001-7233-354X
Benemérita Universidad Autónoma de Puebla
Puebla – México
Artículo recibido: 15 de junio de 2025. Aceptado para publicación: 03 de octubre de 2025.
Conflictos de Interés: Ninguno que declarar.
Abstract
We estimate an approximation of the equilibrium real exchange rate to measure the Mexican
currency overvaluation, which is approximately of 25.3% in the first quarter of 2024. The nominal
exchange rate level that would return the real exchange rate to its equilibrium level would be $22.77
Mexican pesos per US dollar, in contrast to an average nominal exchange rate of $17.00 observed in
first quarter of 2024 and an average nominal exchange rate of $19.51 in the second quarter of 2025.
The National Institute of Statistics and Geography (INEGI) changed in 2023 the base year to measure
real variables as GDP, private consumption, public consumption and exports from 2013 to 2018
chained pesos. It is essential to estimate the VAR and a VEC models for period 1995Q1-2024Q1
using the new time series provided by INEGI to obtain updated and subsequent estimates of the
magnitude and how it changes the Mexican peso overvaluation. The fitted values of the
cointegration equation provide us with the approximation of the equilibrium real exchange rate.
Although the observed real exchange rate has risen from the local minimum reached in the first
quarter of 2024, it is appreciating again starting in the second quarter of 2025. If this appreciation
continues, there is a risk of a return to significant exchange rate overvaluation.
Keywords: nominal exchange rate, real exchange rate, VAR-cointegration, Mexican currency
misalignment, Mexican peso overvaluation
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 609.
Resumen
Estimamos una aproximación del tipo de cambio real de equilibrio para medir la sobrevaluación de la
moneda mexicana, la cual es de aproximadamente 25.3% en el primer trimestre de 2024. El nivel de
tipo de cambio nominal que regresaría al tipo de cambio real a su nivel de equilibrio sería de $22.77
mexicanos pesos por dólar estadounidense, en contraste con un tipo de cambio nominal promedio
de $17.00 observado en el primer trimestre de 2024 y con un tipo de cambio promedio de $19.51 en
el segundo trimestre de 2025. El Instituto Nacional de Estadística y Geografía (INEGI) cambió en
2023 el año base para medir variables reales como el PIB, el consumo privado, el consumo público y
las exportaciones de precios constantes de 2013 a precios constantes de 2018. Es esencial estimar
los modelos VAR y VEC para el periodo 1995T1-2024T1 utilizando las nuevas series de tiempo
publicadas por el INEGI para obtener estimaciones actualizadas y posteriores de la magnitud y de
cómo cambia la sobrevaluación del peso mexicano. Los pronósticos por la ecuación de
cointegración proporcionan la aproximación del tipo de cambio real de equilibrio. Aunque el tipo de
cambio real observado ha subido desde el mínimo local alcanzado en el primer trimestre de 20024, a
partir del segundo trimestre de 2025 se está apreciando nuevamente. De mantenerse dicha
apreciación se corre el riesgo de regresar a una sobrevaluación cambiaria considerable.
Palabras clave: tipo de cambio nominal, tipo de cambio real, VAR-cointegración,
desalineamiento de la moneda mexicana, sobrevaluación del peso mexicano
Todo el contenido de LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades,
publicado en este sitio está disponibles bajo Licencia Creative Commons.
Cómo citar: iménez Gómez, A., Flores Márquez, H., & Absalón Copete, C. (2025). An estimation of the
nominal exchange rate level that would eliminate the mexican currency misalignment: is it moving in
the right direction in 2025?. LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades 6
(5), 608 – 625. https://doi.org/10.56712/latam.v6i5.4619
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 610.
INTRODUCTION
The Mexican GDP fell 8.4% in 2020 with respect the previous year, while the inflation rate remained
low, reaching a minimum of 2.1% (annual rate) in April in that year. At the beginning of 2021, Banco de
México reacted decreasing its objective nominal interest rate to 4.0% and kept it in that level from the
12 of February till the 24 of June of 2021. However, inflation picked up in Mexico -as in many other
countries- in the second half of 20211. Banco de México started increasing its objective nominal
interest rate to curb inflation by the end of June 2021, around three quarters earlier than the FED did.
The lack of synchronization between the monetary policies between those two central banks
provoked by the delay of the FED in recognizing that inflation was becoming a serious problem
caused that the interest rates differential widens. The lowest interest rate differential was around 3.9
percentage points, from February 18 to June 22, 2021. Since this last date, only the Bank of Mexico
had uniquely increased its objective nominal interest rate until the 24 of February 2022. The interest
rate differential had reached 5.9 percent at this date. Starting on March 19, 2022, the FED tightened its
monetary policy and then both central banks alternated increases in their respective interest rates
until May 10, 2023, when the Bank of Mexico increased its objective interest rate for the last time that
year. In this period of alternating increases, the interest rate differential reached a maximum of 6.4
percentage points. In the second half of 2023, inflation rates began to slow in both countries: in July
and October for the United States and Mexico, respectively. The increased in the interest rate
differential from 3.9 to 6.4 per cent points caused an accelerated appreciation of the Mexican peso:
from around MXN $20.0 per dollar when Banco de México began to tighten its monetary policy, to
below MXN $17.0 by mid-April 2024. This caused the appreciation of the real exchange rate (RER).
However, appreciation of the RER does not necessarily mean overvaluation because the latter is
determined by the difference between the observed RER and the equilibrium real exchange rate
(ERER). The problem with the ERER is that it is a “moving target”, as it recognized by Gil Díaz y
Carstens (1996: 11). Another reference that the ERER is a moving target is provided by Harberger
(1996: 44-45), who pointed out that the ERER for Mexico changed from its previous level in
September to a new level in November 1993, when the U. S. Congress approved NAFTA. Again, the
ERER changed twice because of the capital outflow registered in April and in December 1994. This
exemplifies how the ERER can change quickly. Both the ERER and RER are moving in time, and the
overvaluation (or undervaluation) estimates result from the comparison between them. For this
reason, we must support how the ERER is obtained. We decided to choose the cointegration approach
to identify the ERER through the cointegration equation, as it will be discussed in the review of the
literature.
As both the nominal and the real exchange rates appreciates over time, the latter diverts from its
long-run equilibrium value. In addition, INEGI change the base year and now it measures Mexican real
variables in 2018 chained pesos instead of 2013 chained pesos, as it will be discussed later. The
hypothesis of this research is that there is one cointegration equation for the new sample 1995Q1-
2024Q1, despite the change of the base year for the Mexican variables.
We consider this estimate very valuable to achieve a triple objective: i) it will allow us to estimate an
updated Mexican peso overvaluation; ii) it will allow us to carry out a sensitivity analysis to find out
how the base change from 2013 to 2018 chained pesos affected the overvaluation estimates, and iii)
it will allow us to identify which is the level of nominal exchange rate consistent with the elimination
of the Mexican currency misalignment in the first quarter of 2024. The justification of this new
estimate is that Mexico faces a currency overvaluation problem, and it is very valuable to get results
that shed light about its current magnitude and if it is getting bigger or smaller. We organize the rest
1 It was mainly caused by a second quantitative easing program instrumented by the U. S. Federal Reserve Bank (FED). See
Figure 1 in Jimenez-Gomez et al. (2023:4898) for more details.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 611.
of the paper as follows: a survey is presented in section 2. In section 3, we re-estimate the model of
Jiménez-Gómez (2023) with the most recent information available by the time we write this paper. In
section 4, we present the currency overvaluation estimations and discuss which would be the nominal
exchange level necessary to correct the Mexican peso misalignment. In addition, we discussed the
effect of the U. S. dollar weakness that arose since the announcement of the tariffs impositions to
many countries on “liberation day”. Some concluding remarks are discussed at the end of this
chapter.
DEVELOPMENT
Review of the Literature on How Currency Overvaluation Has Been Estimated
Theoretical Model
The difference between the ERER and the observed RER can be considered as a measure of currency
overvaluation or undervaluation. First, we must select a theoretical approach that allows us to identify
the ERER. Then we must estimate this “moving target” using statistical techniques. There are already
two very good surveys on the determination of the ERER: Hinkle and Montiel (1999) and Lee et al.
(2008). These authors propose a classification of the different approaches. For example, Montiel and
Hinkle (1999: 4) classify the models according to the methodologies that are used to obtain an
estimation of the ERER: i) models of trade, ii) computable general equilibrium models, iii) Purchasing
Parity Power (PPP), and iv) econometric estimates, that commonly use time series with unit roots and
try to estimate cointegration vectors. In this chapter, we follow the time series approach to estimate
the ERER according to the theoretical model of Stockman (1987) and Krugman et al. (2012)2.
According to these authors, the ERER is identified where the relative supplies and demands of two
countries intersect. In present case, the first curve is defined as the supply of Mexico relative to the
supply of the U. S., while the second curve is the demand of Mexico relative to the demand of the U. S.
2 Stockman expresses his original model in terms of indifference curves, while Krugman et al. (2012) expresses the
Stockman original model in terms of relative supplies and demands.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 612.
Graphic 1
Equilibrium real Exchange rate: An increase of the domestic country relative supply
Graphic 2
Equilibrium real exchange rate: An increase of the domestic country relative demand
In Figure 1, we represent the case of an increase in the ratio Mexican supply / U. S. supply. The
economy is initially at ERER0 and that the supply of Mexico grows more than the supply of the U. S.
keeping the relative demand without change, then the relative supply shifts to the right and the new
equilibrium level will be ERER1. The ERER depreciates because the price level in Mexico increases
less than the corresponding variable in the U. S. because of a larger supply expansion in the former
country´s supply than of the latter. In figure 2, we represent an increase in the ratio Mexican demand /
U. S. demand. Initially, the economy is at ERER0 and if the demand in Mexico increases more than the
demand in the U. S., keeping the relative supply constant, then the relative demand moves to the right,
being ERER1 the new equilibrium level. As the Mexico´s price level increases more than the one in the
U. S. because of a larger expansion in the demand of the domestic country than of the foreign
country.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 613.
Estimates of Currency Misalignment for Mexico
These estimates can be classified if the prevailing exchange rate regime was fixed and currency band
or flexible. In the first group we find a lot of currency misalignment estimations for several years:
Zedillo (1992: 36), Solís (1996: 87), Elbadawi and Soto (1997, 102), Dornbusch and Werner (1994:
286) and Montiel (1999: 259). For 1982 and 1994, some authors estimated the largest currency
overvaluations. In the second group there are few currency misalignment estimations: Carrera et al.
(2021: 71), Galindo and Guerrero de Lizardi (2001: 8), González-García et al. (2022, 148), Jiménez-
Gómez (2023: 187), and Jiménez-Gómez et al. (2023, 4914).
For this period the largest currency undervaluation was estimated for 1995 and no serious persistent
overvaluation was detected. We must emphasize that the 2023 information was the latest for the
most recent studies.
Estimates for Other Countries
As we will review, the concern of estimating currency misalignments is widespread among
researchers, mainly in developing countries. Nassif, et al. (2011, 8-11) consider the RER as a function
of long-term variables. These authors find statistically significant the estimated coefficients
associated to the level of international assets, the Brazil´s risk premium, the difference between
domestic and foreign interest rates, the current account and the gross domestic product per capita.
The authors identify a currency overvaluation around 80% in April 2011. García-Solanes and Torrejón-
Flores (2013, 10) consider a non-tradable and tradable goods model, highlighting total factor
productivity. According to the Balassa-Samuelson approach, countries that register more accelerated
productivity growth rates in the tradable sector will have an appreciating real exchange rate over time.
García-Solanes and Torrejón-Flores (2013, 14) estimate using a panel data model that real exchange
rate for Spain was overvalued around 28% in 2008. Villegas et al. (2013) estimate a cointegration
vector and an error correction model for Venezuela for the period 1999-2010 and they find a stable
relationship between the RER, and the degree of openness, terms of trade, aggregate productivity,
government expenditure and capital flows in the long run. Based on their work, these authors find
periods of overvaluation and undervaluation in Venezuela. Such exchange rate misalignment tends to
persist. Cheung and Fujii (2014, 94) regress the real exchange rate in terms of the real per capita
income of a country relative to the corresponding U. S. variable and a stochastic error. The estimated
residual (��̂��) indicated overvaluation if it is positive or undervaluation if it was negative. They find the
currencies of BRIC countries were undervalued in 2005 with respect to the U. S. dollar. On the other
hand, Tashu (2018) studies the ERER in Peru following the cointegration approach. Based on his
results, this author argues that the ERER of Peru is mainly driven by government expenditure and
productivity. He concludes that the Peru´s currency -the sol- is not a commodity currency because the
prices of the raw materials for export do not have a statistically significant impact on its real effective
exchange rate. González-Sánchez (2020) estimate an ARDL model following the Behavioral
Equilibrium Exchange Rate (BEER) methodology for Dominican Republic for the period 1996-2020.
The author uses the gross capital formation, terms of trade and public consumption.
The author concludes that: “For 2019 the average total misalignment was 0.8%, maintaining the same
trend during the first quarter of 2020, this suggests the need for a slight real appreciation and is
indicative that the RER is practically in equilibrium”, González-Sánchez (2020: 14). These references
represent examples of how the ERER can be estimated as a previous step to identify a currency
misalignment and using time series with unit roots is a very popular method among researchers.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 614.
Estimating of the Equilibrium Real Exchange Rate (ERER) with Updated Information: Mexican
Variables in 2018 Chained pesos
Jiménez-Gómez (2023) estimated the theoretical model of Stockman (1987) and Krugman et al.
(2012) for the period between the first quarter of 1995 and the first quarter of 2021 and no currency
misalignment problem was detected. Jiménez-Gómez et al. (2023) re-estimated the previous model
using updated information till the first quarter of 2023 and a currency overvaluation of 9.8% was
estimated. In both cases, the series for the Mexican variables were in 2013 chained pesos. But the
National Institute of Statistics and Geography (INEGI) decided to change the base year used to
measure the GDP and the aggregate demand variables. The base years used previously were: 1993,
2003 and 2013, which suggests that the base year is changed every ten years. This indicates that
INEGI would have changed the base year when they had all the necessary information for 2023.
However, the INEGI went ahead to change the base year, choosing 2018 instead of 2023, which is
inconsistent with what has been done in the last 30 years. For this reason, the models and estimates
quoted at the beginning of this section became obsolete. Therefore, any update to the model and its
estimates from the second quarter of 2023 onwards must be done using a completely new series for
the Mexican GDP and for the sum of private and public consumptions, and exports, in chained pesos
of 2018.
The approximate variables for Mexico and the U. S. supplies are GDP (Y: chained pesos of 2018) and
the Industrial Production Index (IPI: 2017=100), respectively. The approximate variables for Mexico
and the U. S. demands are private consumption, public consumption and exports for the first country
(CGX: chained pesos of 2018) and real fixed private investment (RFPI) for the second one. The RER is
calculated multiplying the nominal exchange rate (Mexican pesos for one U.S. Dollar) times the U. S.
Consumer Price Index divided by the Mexican National Consumer Price Index. We use the natural logs
of these variables to estimate the new model. The data was obtained from the Banco de México,
INEGI and St. Louis FED3. E-views is the statistical software we use to estimate the VAR and the VEC
models and to perform the statistical tests.
3 www.banxico.org.mx, www.inegi.org.mx and www.stlouisfed.org are the web pages, respectively.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 615.
Unit Root Tests
Table 2
Unit root test for RER, Y, IPI, CGX and RFPI. Sample: 1995 1st quarter – 2024 1st quarter
Source: own elaboration.
The augmented Dickey-Fuller and the Phillips-Perron tests have as a null hypothesis that the variable
has a unit root. If a variable is tested in levels and the null hypothesis is not rejected, and
subsequently the test is repeated on the same variable but now in differences and the null hypothesis
is rejected, then we can say that the variable is integrated of order one [I (1)]. The obtained results
allow us to identify the five variables as integrated of order one, therefore, we can continue with the
VAR model estimation. The VAR is estimated with 6 lags as in Jiménez Gómez et al. (2023). We also
incorporate three dummy variables in 2008Q4, 2009Q1 and 2020Q2 because of the Great Recession
and the SAR-COV-2 pandemic seriously affected macroeconomic variables of both countries. The
statistical results are reported in Table 3.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 616.
Johansen Cointegraton Test
According to Juselius (2006: 132), Johansen (1991, 1995) cointegration test discriminates between
those eigenvalues which correspond to stationary relations and those which correspond to non-
stationary relations.
Table 3
Johansen integration test. Sample: 1995 1st quarter – 2024 1st quarter
i) Cointegration test
Eigenvalues 0.35 0.18 0.15 0.03 0.01
Null hypothesis rank = 0 rank ≤ 1 rank ≤ 2 rank ≤ 3 rank ≤ 4
λ trace statistic 92.95 46.11* 23.66 5.23 1.52
Critical values (95%) 69.82 47.86 29.80 15.49 3.84
P-value 0.00 0.07 0.22 0.78 0.22
λ maximun eigenvalue statistic 46.84 22.45* 18.44 3.70 1.52
Critical values (95%) 33.88 27.58 21.13 14.26 3.84
P-value 0.00 0.20 0.11 0.89 0.22
ii) Cointegration vector and adjustment coefficients (Johansen)
Variables RER Y IPI CGX FPI
Normalized cointegration coefficients 1.00 55.20 -114.81 -16.62 9.48
(standard error) 52.89 23.22 34.42 8.68
Adjustment coefficients 0.00 0.00 0.00 0.00 0.00
(standard error) 0.00 0.00 0.00 0.00 0.00
iii) Specification tests
0.58
p-value
Jarque-Bera 2.71 0.99
Skewness (Chi-sq) 2.26 0.81
Kurtosis (Chi-sq) 0.45 0.99
p-value
White (no crossed terms, 945 d. of f.) 999.19 0.11
p-value
lag "h" 3 4 5 6 7 8
Rao F Statistic (p-value) 1.2 ( 0.186) 1.1 ( 0.270) 1.0 ( 0.475) 1.1 ( 0.220) 1.1 (0.182) 1.3 ( 0.066)
* Trace and Max-eigenvalue test indicate 1 cointegration equation at the 0.05 level.
Source: Own elaboration.
Heteroscedasticity Test statistic
Serial correlación LM Test statistic
Null hypothesis: no serial correlation from lags 1 to h
Normality Test statistic
Table 2
Johansen cointegration test
Sample: 1995 1st quarter - 2024 1st quarter
Trace correlation Test statistic
Source: own elaboration.
The first null hypothesis is that there is not a cointegration vector and in this case, it is rejected
because both the trace and the maximum eigenvalue statistics are larger than their critical values.
Then the subsequent null hypothesis is that there is just one cointegration vector which is not rejected
at the 95% confidence because both the trace and the maximum eigenvalue statistics are smaller
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 617.
than their critical values. The estimated coefficients of the cointegration vector have the signs
predicted by Stockman (1987) and Krugman et al. (2012). The sign of the adjustment coefficient
corresponding to the RER is negative, which means that if this variable is out of its long run
equilibrium, then the RER will tend to return to it. We use the trace correlation as a measure of
goodness of fit, which is 0.58 in this case. Finally, the diagnostic tests reveal that the assumptions
about errors are fulfilled.
Estimated Parameters Constancy Tests
We use the likelihood ratio logarithm calculated recursively to test the constancy of the estimated
parameters following a bias-corrected test statistic (Juselius, 2006: 152). The test statistic formula is:
����
��������(��1) =
��1
��
√
��
2��
[{������|Ω̂��1| − ������|Ω̂��|} +
1
��
{(
1
2
��(1 − ��) + �� + ��(�� − 1) + 1) (1 −
��1
��
)}]
Where:
Ω̂�� is the covariance matrix of the errors obtained by the estimation using the complete subsample, in
this case from 1995q1 to 2023q3.
Ω̂��1 is the covariance matrix of the errors obtained by the estimation using a just a part of the
subsample, which changes as ��1 runs.
��1 is the time index that runs to enlarge the part of the subsample.
�� is the full subsample size.
�� is the number of variables.
�� is the number of cointegration vectors.
�� is the number of lags in the variables in levels.
If the sample increases forward, we get the statistic estimating the R-model for the periods 1996q3 –
2004q1, then 1996q3 – 2004q2, and so on until we use the full sample: 1996q3-2024q1. If the sample
increases backwards, we get the test statistic estimating the R-model for the periods 2004q1-2024q1,
2003q4-2024q1, 2003q3-2024q1 and so on until we use the full sample: 1996q3-2023q3. We can plot
the test statistic as the size of the sample increases forward (or backward) in Figure 3. The values of
this test statistic are lower than 1.0, which provides some evidence that there was no structural
change within the sample4. The next step is to estimate the vector error correction model (VEC),
whose results are in Table 35.
4 See Juselius (2006: 152) for details
5 The dummy variables already incorporated in the VAR model are also in the VEC model but in differences. In addition, we
incorporate dummy variables in differences for 2009Q3, 2023Q3 and 2023Q4. We also include a dummy variable 2024Q1
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 618.
Graphic 3
Recursive tests of constancy of the estimated coefficients in the Co-integration vector QT
Corr (t1),
forward and backward. 1995Q1 – 2024Q1
Source: own elaboration.
The cointegration equation is identified in the first row. We use this equation to obtain �������� = ������̂.
From the tests in the second section of Table 3, we can see that all the estimated coefficients in the
cointegration equation are statistically significant. From the tests reported in the third section, the
RER can be considered as a “reaction” variable because its estimated adjustment coefficient is
statistically significant. On the other hand, Y, IPI, CGX and FPI are “push” variables because their
adjustment coefficients are not statistically significant. The VEC model fulfills the assumptions of
normality, homoscedasticity and no serial correlation of errors as we can see in the last section of
Table 4.
Vector Error Correction Model and Diagnostic Tests
We can estimate ΔRERt as a function of the differences of RER, Y, IPI, CGX and RFPI including 1 to 5
lags and the error correction term using Ordinary Least Squares because all the variables are
stationary. The results are presented in Table 4. We start with the same dummy variables in
differences incorporated in the VEC model. We follow “from the general to particular approach” to
estimate the specific VEC model for our variable of interest. We perform the redundant variable test to
find out if a group of coefficients are not statistically significant. If the null hypothesis that the group
of coefficients are statistically equal to zero is not rejected, the group of variables can be removed
from the model. We repeat this procedure until no further group of variables can be removed. The
results are presented in Table 4.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 619.
Table 4
Vector error correction model, cointegration equation, and weak exogeneity and diagnostic tests.
Sample: 1995 1st quarter – 2024 1st quarter
Source: own elaboration.
Vector Error Correction Model Specific to ΔRER and Diagnostic Tests
The ECM specific to the RER has few variables. Only ΔRERt-1, ΔYt-1 and ECTt-1 remain in the final
model. This implies that the influence of the other three variables (IPI, CGX and RFPI) on ΔRERt is
through the error correction term. The number of dummy variables is reduced from seven to five. The
dummy variables in differences for 2008Q4, 2009q1 and 2002q2 remain from the VAR model. We add
a dummy variable in differences for 1995Q4 and a dummy variable for 2024Q1. The specification
tests reveal that the classical assumptions about errors are fulfilled.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 620.
Table 5
Error correction model specific for ΔRERt, 1995 1st quarter – 2023 3rd quarter
Source: own elaboration.
Estimate of the Overvaluation of the Mexican Peso
We use the estimated cointegration equation reported in Table 3 to obtain the fitted value of the RER
that we consider as the ERER and it is represented in equation (1).
���������� = ������̂�� = 3.94���� − 3.98�������� − 1.70�������� + 0.97���������� − 23.49 (1)
Mexican Peso Overvaluation Estimation and the Level on the Nominal Exchange Rate that eliminates
the Currency Misalignment
The percentage deviation of the RER with respect to ERER6 is our calculation of the overvaluation of
the Mexican peso. In this way, we use equation (1) to calculate an overvaluation of the Mexican
currency of 25.3% in first quarter of 2024 (see Figure 4). The current Mexican peso misalignment
started in the last quarter of 2022 (-8.9%) and the first quarter of 2023 (-7.8%), however the
overvaluation more than doubled in the next two quarters: second and third quarters of 2023 with -
6 Already in levels and not in natural logarithms.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 621.
21.3 and -22.4%, respectively. In the last quarter of 2023, the currency misalignment reached a
maximum estimated overvaluation (-29.1%) since the flexible exchange rate regime was adopted in
1995, according to this estimation. The Mexican currency misalignment observed in the first quarter
of 2024 is very similar in magnitude (but with an opposite sign) to the one observed in the second
quarter of 1995, when the RER reached its maximum level because of the “overshooting”, after the
abandonment of the currency band for the Mexican peso in December 1994.
Graphic 4
Estimated percentage of undervaluation (+)/overvaluation (-) of the Mexican currency. Sample 1995Q1
– 2024 Q1
Source: own elaboration.
According to Nassif, et al. (2011, 5), the increase in the interest rates differential between Mexico and
the U. S. and the expectations of the nominal exchange rate appreciation caused flows of foreign
short-term capital. However, these capital flows can stop and even begin to leave Mexico suddenly.
Sooner or later, the nominal exchange rate will depreciate to undo the overvaluation. Under the
assumption that the nominal exchange rate is the only variable that adjusts so that the RER returns to
its equilibrium level, we calculate that the exchange rate would depreciate to MXN $22.77 per dollar.
In addition to this depreciation, we must add the “overshooting” as well. Capistrán et al. (2017, 22)
found evidence of “… the existence of an overshooting in the Mexican exchange market...”. In this way,
when the nominal exchange rate adjusts, we expect it will situate temporarily above our estimation.
An adjustment of such magnitude will represent an important shock to the Mexican economy.
Comparison to other authors´ estimations
This overvaluation estimation is not directly comparable with those in Jiménez-Gómez (2023, 187)
and Jiménez-Gómez et al. (2023, 4914) mentioned in the survey. The reason is that the supply and
demand variables for Mexico are in 2013 chained pesos in the former studies, while in the present
research those series are in 2018 chained pesos. For this reason, it is important to identify to what
extent the estimates for the overvaluation differ between each other depending on if the Mexican real
variables in the model are in 2013 or in 2018 chained pesos.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 622.
The last four observations of the model estimated in Jiménez-Gómez et al. (2023) (from the 2nd
quarter of 2022 to the 1st quarter of 2023) are compared with the same observations of the present
model. This comparison is reported in Table 5.
Both models point out that the overvaluation problem began in the fourth quarter of 2022. The
estimates of the ERER and those of the nominal exchange rate of the two models do not differ much.
The difference in the nominal exchange rate which is consistent with the ERER is no larger than 40
Mexican cents. This comparison suggests that the Mexican peso misalignment estimation is not
seriously affected by using series for Mexican variables in 2013 or in 2018 chained pesos. The
estimates of the latter model are more conservative, calculating lower nominal exchange rates.
Table 6
Estimated equilibrium exchange rates using Mexican real series in 2013 and 2018 chained pesos
Source: own elaboration.
Evolution of the Observed Real Exchange Rate and Recent Events
In December 2024, Banco de México´s objective interest rate was 10.0% and since then it has
decreased to 7.75% by the end of August 2025, despite the FED stop decreasing its interest rate
because of the inflation risks that emerge as a result of the tariffs impositions by the Trump
administration. One possible explanation is that the annual inflation reached 3.5% in July 2025, hence
Bank of Mexico decreased its interest rate. An alternative explanation is that Banco de Mexico made
this decision to reduce the interest payments of the internal debt, which grew considerably to finance
the public deficit mentioned above. The reduction in the interest rate differential is not risky because
the dollar has weakened as a result of the U. S. government´s tariff policy. The U. S. dollar
depreciation has made possible to decrease the Mexican interest rate and the interest rate
differential, however, this depreciation has provoked that the observed RER started appreciating again
in the second quarter of 2005, which makes it even more difficult to eliminate the Mexican peso´s
overvaluation.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 623.
Graphic 5
Observed real exchange rate
Source: own elaboration with data of Banxico, INEGI and FED Saint Louis.
CONCLUSION
We have estimated a new empirical model with updated information until the first quarter of 2024 and
incorporated the series for the Mexican GDP, private consumption, government consumption an
exports expressed in 2018 chained prices. We estimated that the Mexican real exchange rate is
overvalued 25.3% in the first quarter of 2024. We also estimate that the nominal exchange rate
consistent with the equilibrium real exchange rate in the first quarter of 2024 is approximately MXN
$22.77 per dollar. The Mexican peso overvaluation reached a local maximum at the first quarter of
2024, and since then the observed RER has been depreciating which is consistent with the elimination
of the currency misalignment, but the U. S. dollar weakness has made that the observed RER started
appreciating again in the second quarter of 2025. If the U. S. dollar weakness persists, then the
Mexican peso´s overvaluation will tend to persist too, becoming another obstacle to Mexican exports
that are sent to the U. S. economy.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 624.
REFERENCES
Capistrán, C., D. Chiquiar and J. R. Hernández (2017). Identifying Dornbusch's Exchange Rate
Overshooting with Structural VECs: Evidence from Mexico. Documentos de Investigación del 2017-11,
Banco de México.
Carrera, J., B. Gnimassoun, V. Mignon and R. Restout. (2021). Currency Misalignments and Exchange
Rate Regimes in Latina American Countries: a Tred-Off Issue. Annals of Economics and Statistics,
Volume 141, 71-102.
Cheung, Y.‐W. and E. Fujii. (2014). Exchange rate misalignment estimates: sources of differences.
International Journal of Finance & Economics, Volume 19 (2), 91-121.
Dornbusch, R. (1976). Expectations and exchange rate dynamics. Journal of Political Economy,
Dornbusch, R. and M. Werner. (1994). Stabilization, Reform, and No Growth. Brookings Papers on
Economic Activity, Volume 1994 (1), 253-315.
Elbadawi, I. A. and R. Soto. (1997). Real Exchange Rates and Macroeconomic Adjustment in Sub-
Saharan Africa and other Developing Countries. Journal of African Economics, Volume 6 (3), 74-120.
Galindo, L. M. y C. Guerrero de Lizardi (2001). Los Efectos de la Recesión Estadounidense y del Tipo
de Cambio Real sobre el Crecimiento Económico de México en 2001. Momento Económico, 116, pp.
2-9.
García-Solanes, J. and F. Torrejón-Flores. (2013). La sobrevaloración del tipo de cambio real de la
economía española. Cuadernos de Economía, Volume 36 (10), 5-16. Doi:10.1016/s0210-
0266(13)70037-2.
Gil-Díaz, F. and A. Carstens. (1996). Some Pilgrim Tales about Mexico’s 1994-1995 Crisis. American
Economic Review, Volume 82 (2), 164-169.
González-García, J., García-Contreras, J. J., and Osorio-Novela, G. (2022). La política cambiaria de
México y China en una economía Global. Investigación Económica, Volume 81(320), 131–155.
https://doi.org/10.22201/fe.01851667p.2022.320.81521
González-Sánchez, O. (2020). Comportamiento del tipo de cambio real de equilibrio en la República
Dominicana: Estimación del desalineamiento 1996 – 2020. Documento de Investigación 2020-02,
Ministerio de Hacienda. https://www.hacienda.gob.do/wp-content/uploads/2020/08/2020-02-
TCR_compressed.pdf
Harberger, Arnold, C. (1996). “Mexico´s exchange rate crisis”, mimeo prepared for Grupo Financiero
Bancomer.
Hinkle, L. E., and P. J. Montiel (Eds.). (1999). Exchange rate misalignment: concepts and
measurement for developing countries. Oxford University Press, USA.
Jiménez-Gómez, A. (2023). Determinación del tipo de cambio real de equilibrio para el caso de
México: un enfoque de cointegración. Paradigma económico, Volume 15 (1), 159-192.
Jiménez-Gómez, A., B. Martínez-Carreño and A. C. Cabrera-García (2023). An Estimation of the
Mexican Peso´s Overvaluation for 2023. Ciencia Latina, Volume 7 (4), 4896-4918.
Johansen, S. (1991). Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector
Autoregressive Models. Econometrica, Volume 59 (6), 1551-1580.
LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, Asunción, Paraguay.
ISSN en línea: 2789-3855, octubre, 2025, Volumen VI, Número 5 p 625.
Johansen, S. (1995). Likelihood-based inference in cointegrated vector auto-regressive models.
Oxford University Press.
Juselius, K (2006). The cointegrated VAR model. Oxford University Press, United Kingdom.
Krugman, P., M. Obstfeld and M. J. Melitz (2012). Economía Internacional: teoría y política. Pearson,
España.
Lee, J., G. M. Milesi-Ferreti, J. Ostry, A. Prati and L. A. Ricci. (2008). Exchange Rate Assessments:
CGER Methodologies. Occasional paper 261, IMF.
Montiel, P. J. and L. E. Hinkle (1999). Exchange Rate Misalignment: An Overview, in Hinkle, L. E. and P.
J. Montiel (Eds.) Exchange rate misalignment: concepts and measurement for developing countries
(1-32), Oxford University Press.
Montiel. P. J. (1999). The Long-Run Equilibrium Real Exchange Rate: Conceptual Issues and Empirical
Research, in Hinkle, L. E. and P. J. Montiel (Eds.), Exchange rate misalignment: concepts and
measurement for developing countries (219-263), Oxford University Press.
Nassif, A., C. Feijó and E. Araújo (2011). The long-term optimal exchange rate and the currency
overvaluation trend in open emerging economies: the case of Brazil. Discussion Paper 206, United
Nations Conference on Trade and Development.
Noureldin, D. (2018). Much Ado about the Egyptian Pound: Exchange Rate Misalignment and the Path
Towards Equilibrium. Review of Middle East Economics and Finance, Volume 14(2), 1-19.
Solís, L. (1996). Crisis económico-financiera 1997-1995. El Colegio Nacional and Fondo de Cultura
Económica.
Stockman, A. C. (1987), “The equilibrium approach to exchange rates”, Economic Review, Volume 73
(2), 12-30.
Tashu, M. (2018). Determinantes del Tipo de Cambio Real de Equilibrio en Perú: ¿Es el sol una
moneda commodity? Revista Estudios Económicos, Volume 36, 9 – 30.
Villegas, E., N. Labarca and B. Pozo (2013). Un análisis de comportamiento del tipo de cambio real en
Venezuela. Tendencias, Volume XIV (1), 122-145. Volume 84(6), 1161-1176.
Zedillo, E. (1992). La experiencia entre 1973 y 1983 de la balanza de pagos y las perspectivas de
crecimiento de México, en Bazdesch C., N. Bucay, S. Loaeza and N. Lustig (Eds.), México, auge, crisis
y ajuste (17-66), Fondo de Cultura Económica.
Todo el contenido de LATAM Revista Latinoamericana de Ciencias Sociales y Humanidades, publicados en
este sitio está disponibles bajo Licencia Creative Commons .