October 26, 2026
Macroeconomic Indicators and the Performance of the Individual’s Credit Portfolio of a Banking Institution
Pamela Siller de Melo; Marcio Pereira Basilio
DOI: 10.22167/2675-6528-202601292
Article prepared by the ResumeAI tool, an artificial intelligence solution developed by the Pecege Institute focused on synthesis and writing.
Summary
Credit granting plays a strategic role in Brazilian economic growth, being sensitive to macroeconomic fluctuations that impact household consumption and the risk management of financial institutions. The relationship between macroeconomic indicators, such as Gross Domestic Product (GDP), Broad Consumer Price Index (IPCA), Selic rate, and unemployment, and the behavior of the credit portfolio balance of individuals from a banking institution was evaluated, in the period from 2015 to 2025. A quantitative case study was conducted, employing multiple linear regression with analysis of temporal lags of up to four quarters on quarterly data. The results indicated that GDP was the most robust explanatory variable, exerting a direct positive influence on portfolio growth. The Selic rate, with a one-quarter lag, showed a negative relationship with the credit stock, confirming the effectiveness of monetary policy. Contrary to theory, the unemployment rate exhibited a positive coefficient, suggesting the use of credit to smooth consumption during periods of labor market deterioration. The IPCA did not show statistical relevance. It was concluded that the dynamics of the credit portfolio were intrinsically linked to the economic cycle and responded with a lag to variations in the cost of capital. The identification of these dynamic and lagged relationships showed that efficient portfolio management depends on the institution’s ability to adapt its policies to economic fluctuations and the cost of capital.
Keywords: Econometric analysis; Economic activity; Interest; Multiple linear regression.
1. Introduction
Credit, in its various forms, is a fundamental pillar for contemporary economic dynamics, acting as a catalyst for consumption, investment, and consequently, for the growth of a nation’s Gross Domestic Product (GDP). In Brazil, the granting of credit to individuals plays a strategic role, boosting demand for durable goods, real estate financing, and access to essential services, generating multiplier effects on productive activity and the labor market (Assaf Neto, 2015; Mankiw, 2021). The relevance of this segment is evidenced by its significant volume in the National Financial System (SFN), which reflects the banking intermediation capacity in channeling resources from surplus agents to deficit agents, promoting liquidity and monetary circulation.
However, the health and performance of the retail credit portfolio are intrinsically sensitive to fluctuations in the macroeconomic environment. Indicators such as the basic interest rate (Selic), inflation (IPCA), the level of economic activity (GDP), and the unemployment rate influence borrowers’ repayment capacity and financial institutions’ propensity to grant credit. Theoretically, an increase in GDP, which signals economic expansion and higher income, tends to stimulate credit demand and improve households’ repayment capacity. Conversely, increases in the Selic rate, the main monetary policy instrument, raise the cost of capital, discouraging credit and consumption, in line with the monetary transmission mechanism (Mankiw, 2021; Linardi, 2008). Inflation, in turn, can have ambiguous effects: in the short term, it can boost the search for credit to smooth consumption, but in the long term, inflationary uncertainty inhibits new financial commitments (Camargos, 2022; Maia, 2025). Unemployment, by reducing income and financial stability, generally reduces credit demand and increases the risk of nonpayment, driving more restrictive lending policies (Buss, 2024).
Brazilian academic literature has extensively explored these relationships. Silveira (2015) analyzed bank credit and confirmed the correlation between income, interest rates, and default, impacting family financial stability. Buss (2024) identified unemployment and interest rates as determinants of family default, while Murtadha (2025) reinforced the influence of Selic, inflation, and exchange rates on credit portfolio performance, particularly in card operations. Davoglio (2019) complemented this by showing that default follows sectoral economic performance, underscoring the importance of the macroeconomic environment for banking sustainability. Tiryaki et al. (2017) highlighted the procyclical nature of credit, where income expansion stimulates concessions and increases the outstanding balance of operations. However, most studies focus on aggregate data or specific sectors, leaving a gap in the empirical analysis of the simultaneous and lagged impact of these indicators on the retail credit portfolio of a particular banking institution. Understanding these dynamics is vital, as efficient portfolio management and public policy formulation depend on the ability to anticipate and react to economic fluctuations.
In this context, the present research is justified by the need to deepen the understanding of how macroeconomic conditions affect credit for individuals in a specific banking institution, providing practical subsidies for risk management, product pricing, and lending strategies. The study contributes to the literature by empirically investigating, with analysis of temporal lags, the sensitivity of the individual credit portfolio to variations in Gross Domestic Product (GDP), Extended National Consumer Price Index (IPCA), Selic rate, and unemployment, in the period from 2015 to 2025, through the estimation of multiple linear regression models.
2. Material and Methods
The present study was characterized as a research of an applied nature, with a quantitative and descriptive approach, following the methodological classification proposed by Gil (2023). The applied nature of the research aimed to deepen the understanding of the dynamics of credit for individuals in a specific banking context, while the quantitative approach allowed for the statistical analysis of numerical data. The descriptive design sought to identify and analyze the relationships between macroeconomic variables and the performance of the credit portfolio, without manipulating variables, focusing on the description of observed phenomena. This methodological combination was essential to achieve the proposed objectives.
The research was developed as a case study, with the credit portfolio of individuals from a specific banking institution as the unit of analysis. This institution, classified in segment S3, is headquartered in the state of Espírito Santo and operates with credit operations for individuals and legal entities. To preserve the confidentiality of the study, it was decided not to identify the institution by name. The analysis period covered the years 2015 to 2025, with data organized in quarterly frequency, allowing observation of the dynamics of the variables throughout distinct phases of the Brazilian economic cycle and their impacts on credit.
The technical procedures of the research were based on a documentary and bibliographic approach, using exclusively secondary data. The macroeconomic information, which served as explanatory variables, was collected from official and publicly accessible sources, including the Central Bank of Brazil and the Brazilian Institute of Geography and Statistics (IBGE). The data referring to the credit portfolio of individuals of the analyzed banking institution, which constituted the dependent variable, were obtained through public reports made available on the investor relations website of the institution itself. This methodology guaranteed the reliability and availability of the data for the analysis.
The dependent variable of the study was defined as the percentage growth of the commercial credit portfolio balance of individuals of the banking institution. This portfolio includes loan operations, real estate and other asset financing, as well as credit card operations. The selected explanatory variables, based on economic literature, were the Gross Domestic Product (GDP), the basic interest rate (Selic Meta), inflation measured by the Broad Consumer Price Index (IPCA), and the unemployment rate. The choice of these variables was justified by their recognized influence on both credit demand and household payment capacity.
To ensure the consistency, comparability, and suitability of the data for econometric modeling, the variables used were previously treated. The growth of the retail credit portfolio, the dependent variable, was measured in quarterly percentage terms. This measurement was carried out based on the variation in the portfolio balance between consecutive periods, a transformation that aimed to reduce the inherent trend in the time series and make it more appropriate for regression analysis. Such a procedure is common in econometric studies to stabilize variance and meet model assumptions.
The explanatory variables – Selic rate, GDP, IPCA, and unemployment rate – had their series organized on a quarterly basis, making them compatible with the frequency of credit portfolio data. The Selic Meta rate was calculated as the quarterly average of the annualized rate, provided by the Central Bank of Brazil. GDP was used as a quarterly volume index with seasonal adjustment, obtained from table 1621 of IBGE. IPCA represented the accumulated percentage variation over three months, corresponding to quarterly inflation, extracted from table 1737 of IBGE. The unemployment rate was used according to the Continuous National Household Sample Survey (PNAD Contínua), from table 4099 of IBGE, already in quarterly frequency.
Additionally, to capture the dynamic effects of macroeconomic conditions on the credit portfolio, time lags were incorporated into the explanatory variables. This approach considered that the impacts of economic variables do not manifest immediately, but over time. Models with lags of one to four quarters were estimated for all explanatory variables. The choice to incorporate temporal lags is supported by the transmission mechanism of monetary policy, whose effects can materialize with a delay of approximately six to nine months, which corresponds to two to three quarters, as indicated by the Central Bank of Brazil.
For the data analysis, the statistical technique of multiple linear regression was employed. The objective was to evaluate the relationship between the growth of the personal credit portfolio balance, as the dependent variable, and the selected macroeconomic indicators, as explanatory variables. This approach allowed for the identification of the direction, magnitude, and statistical significance of the associations between the variables, contributing to the understanding of the credit portfolio’s behavior in the face of macroeconomic environment variations. The selection of the most relevant explanatory variables in each specification was carried out using the stepwise method, an algorithmic process based on statistical significance criteria.
Statistical analyses were conducted with the aid of electronic spreadsheets, such as Microsoft Excel, and the Stats Studio platform, a statistical tool developed by faculty members of the “Luiz de Queiroz” Higher School of Agriculture of the University of São Paulo (USP/ESALQ). Stats Studio was used for the estimation of multiple linear regression models, the calculation of coefficients, the performance of significance tests, and the verification of model assumptions. This toolkit contributed to the precision in the measurement and comparison of historical series, ensuring transparency and reproducibility. Tests with different specifications, including temporal lags, were performed to evaluate the robustness of the results and capture possible dynamic effects between the variables.
Regarding the ethical aspects, the research adopted the principle of confidentiality, opting not to identify the analyzed banking institution by name. This measure aimed to protect the entity’s data and privacy, ensuring the integrity of the case study. Additionally, the author assumed full responsibility for the consistency, interpretation of results, and conclusions presented, after critical review of all content. Such procedures ensured transparency and ethics in conducting the research, in accordance with best practices and the protection of sensitive information involved.
The study presented some methodological limitations that deserve to be highlighted. The use of aggregated data and the non-consideration of more complex econometric models, capable of capturing the temporal dynamics of the series more sophisticatedly, represent restrictions. The analyzed sample, comprising the period from 2015 to 2025, although relevant, may not cover all the nuances of the economic cycle. Furthermore, the complexity of credit may be influenced by unobserved factors, and the possibility of omitted relevant variables, as well as the limitation of using national macroeconomic variables for an institution with predominantly regional operations, are points to be considered in the interpretation of the results.
3. Results and Discussion
This section details the results obtained in the empirical research, focusing on the descriptive analysis of macroeconomic variables and the growth of the individual credit portfolio, as well as the interpretation of the estimated multiple linear regression models. The analyzed data, with quarterly frequency, cover the period from 2015 to 2025, allowing for an in-depth understanding of credit dynamics in a specific banking context, located in the state of Espírito Santo, whose identity is preserved for confidentiality reasons. The explanatory variables – Selic rate, accumulated IPCA, GDP, and unemployment rate – were carefully treated to ensure consistency and adequacy for econometric modeling, as detailed in the methodology.
Initially, the descriptive analysis of the time series revealed important patterns and trends in the evolution of the credit portfolio balance. Continuous growth in the balance of the individual credit portfolio was observed over the period, although punctuated by moments of deceleration. This cyclical behavior of the credit portfolio is intrinsic to the nature of the financial market and reflects the sensitivity of credit to fluctuations in economic activity. Starting in 2018, a change in the way data was disclosed, which began to be expressed in thousands of reais (R$ thousand), indicated an improvement in the accuracy of the information provided by the institution, which, in turn, contributes to a more robust analysis. The expansion of credit, even with oscillations, is an indicator of the growing demand for financial resources by families, whether for consumption, investment in durable goods, or real estate financing, as pointed out by Assaf Neto (2015) and Mankiw (2021), who highlight the role of credit as an engine of the economy.
The trajectory of the Selic rate, the basic interest rate of the Brazilian economy, exhibited distinct monetary policy cycles, characterized by periods of tightening and easing. The analysis period began with high rates, around 14.25% per year, between the end of 2015 and mid-2016. Subsequently, the Selic experienced a sharp reduction, reaching a historical low of 2.00% per year in 2020, in response to a scenario of economic slowdown and the health crisis. However, starting in 2021, a new upward cycle began, raising the rate to 13.75% per year in 2023 and ending 2025 at 15% per year. These Selic movements acted as a thermostat for the credit portfolio balance. The fall in interest rates until 2020 significantly reduced the cost of capital, stimulating new borrowing by households and, consequently, boosting the growth of the credit portfolio. In contrast, subsequent increases made installments more expensive and led to greater selectivity by banking institutions in granting credit, as well as discouraging demand. Linardi (2008) corroborates this interpretation, stating that the interest rate directly influences the cost of capital, and increases in the Selic tend to curb credit expansion. Murtadha (2025) complements this by observing that monetary policy impacts households’ financial planning, leading to moderation in credit balance growth during periods of high interest rates, especially for immediate consumption modalities. For banking institutions, understanding these cycles is crucial for adjusting their pricing, risk management, and credit product offering strategies, ensuring portfolio sustainability.
The behavior of the Índice Nacional de Preços ao Consumidor Amplo (IPCA), which measures inflation, was marked by considerable fluctuations throughout the period. An inflationary peak was observed in 2015, with 10.67% per year, followed by a new period of pressure in 2021, when the index reached 3.02% in the third quarter. At the end of the period, in 2025, inflation showed new volatility, registering 2.04% in the first quarter, before a slowdown. High inflation, by eroding families’ real purchasing power, impacts the credit portfolio balance in two main ways. In the short term, there may be an increase in the credit balance due to families’ need to finance basic items to maintain their consumption level, which can momentarily increase indebtedness, as highlighted by Camargos (2022). However, in the long term, persistent inflation inhibits demand for new, larger credit contracts, leading to more cautious behavior and, eventually, to the stabilization of the total credit stock, as suggested by Maia (2025). For financial institutions, inflationary volatility represents a challenge in portfolio management, requiring pricing and provisioning models that consider the erosion of purchasing power and the impact on borrowers’ payment capacity.
Brazil’s Gross Domestic Product (GDP), a fundamental indicator of economic activity, faced a sharp contraction in the 2015-2016 biennium, with drops of 3.5% and 3.3%, respectively, reflecting a period of recession. The subsequent recovery was slow but consistent, culminating in steady growth at the end of 2024 and 2025, with the quarterly volume index evolving from 189.63 in the second quarter of 2024 to 194.92 by the end of 2025. This pattern of economic activity recovery translates positively into the credit portfolio balance. The increase in national production and the improvement of economic expectations strengthen the confidence of both borrowers, who feel more secure in taking on debt, and lenders, who perceive a lower risk of default. Tiryaki et al. (2017) highlight the procyclical nature of credit, where periods of income expansion stimulate lending and increase the balance of operations. Davoglio (2019) reinforces this link, showing the agility with which the credit portfolio’s performance responds to national wealth movements. For the institution, a growing GDP signals a favorable environment for credit expansion, justifying more flexible lending policies and investments in new products.
The unemployment rate, another crucial macroeconomic indicator, presented alarming levels at the beginning of the analyzed period, rising from 8% in the first quarter of 2015 to 13.9% in the first quarter of 2017, and reaching a peak of 14.9% in the third quarter of 2020, amidst the health crisis. However, the most recent data indicate a consistent downward trend, reducing from 7.0% in the first quarter of 2025 to 5.1% in the last quarter of the same year. The improvement in the labor market in 2025 favored families’ ability to access the credit market, sustaining the portfolio balance. Theoretically, unemployment acts as a risk signal for banking institutions, which tend to restrict credit supply during periods of low employment, negatively impacting the total volume of operations (Buss, 2024). Although Camargos (2022) observes that unemployment can, paradoxically, increase the demand for subsistence credit, the general trend is a contraction of the portfolio due to the compromise of the primary income source. The recovery of employment, therefore, is a positive factor for the health of the credit portfolio, reducing the risk of default and expanding the base of potential borrowers.
The graphical analysis of the series allowed a preliminary assessment of the relationships between macroeconomic variables and the growth of the credit portfolio for individuals, complementing the econometric results and offering a visual perspective of the observed dynamics.
Source: Original research results from data from the Central Bank of Brazil and public data from the financial institution of the S3 segment
Figure 1. Quarterly growth of the individual loan portfolio balance and the Selic Rate Quarterly annualized target
As shown in Figure 1, which relates the evolution of the Selic rate with the quarterly growth of the credit portfolio, a clearly defined contemporary relationship is not observed. Periods of interest rate hikes do not immediately coincide with reductions in portfolio growth. This behavior suggests that the effects of monetary policy on credit do not manifest instantaneously, but rather with a time lag, which is consistent with the transmission mechanism of monetary policy. Monetary policy, by altering interest rates, affects the cost of credit and the expectations of economic agents, but these impacts take time to materialize in the real economy and, consequently, in the demand and supply of credit. This preliminary observation reinforces the need to incorporate time lags into econometric models to capture the complexity of these interactions.
The relationship between the quarterly IPCA and credit portfolio growth, although not illustrated by a specific figure in this excerpt, demonstrated high volatility of the inflationary indicator, without a consistent visual pattern of direct and contemporaneous association with portfolio growth. This preliminary result indicates that inflation, in the short term, may have low explanatory power over credit dynamics, an evidence that was later confirmed in the econometric models, where the IPCA did not prove to be statistically significant. This does not mean that inflation has no impact, but that its effects may be indirect, lagged, or already incorporated into other variables, such as the interest rate, which is often adjusted in response to inflationary pressures.
Source: Original research results from data from the Central Bank of Brazil and public data from the financial institution of the S3 segment
Figure 2. Quarterly growth of the commercial credit portfolio for individuals and the quarterly unemployment rate
In Figure 2, which relates the unemployment rate to portfolio growth, a less intuitive association than expected by economic theory was observed. There were periods when the increase in unemployment coincided with credit growth, notably between 2020 and 2021, during the COVID-19 pandemic. This phenomenon can be explained by the reduction in the Selic rate during the same period, which made credit cheaper and may have encouraged families to resort to loans as a consumption smoothing mechanism in a time of economic uncertainty and income loss. This atypical dynamic suggests that, in crisis contexts, credit can act as a “buffer” for families, although this may lead to increased indebtedness in a scenario of financial fragility. The interpretation of this finding is crucial for financial institutions, which need to consider not only credit risk in unemployment scenarios but also the potential demand for subsistence credit.
Source: Original research results from data from the Central Bank of Brazil and public data from the financial institution of the S3 segment
Figure 3. Quarterly growth of the commercial credit portfolio of individuals and quarterly GDP
Finally, Figure 3 presents the relationship between Gross Domestic Product (GDP) and credit portfolio growth. Here, a more consistent pattern was observed, with periods of economic activity expansion associated with increased credit portfolio growth. This evidence reinforces the procyclical nature of credit, where the health of the economy and families’ income generation capacity translate directly into higher demand and supply of credit. This result anticipates the central role of GDP in the regression models, confirming its robustness as the main explanatory variable. For bank managers, Figure 3 underscores the importance of closely monitoring economic activity indicators, as they offer a clear signal about the prospects for credit portfolio expansion or contraction.
In general, graphical evidence suggests that the relationship between macroeconomic variables and credit is not fully captured by simple contemporaneous relationships. This indicates the need to incorporate temporal lags in econometric models, an approach that allows for the analysis of how the impacts of economic conditions propagate over time, a fundamental aspect for understanding credit dynamics. The complexity of these interactions requires a more sophisticated analysis, which transcends the observation of immediate correlations.
Multiple linear regression results
In order to analyze the relationship between macroeconomic variables and the growth of the credit portfolio of individuals, multiple linear regression models were estimated, both with and without the inclusion of temporal lags. The stepwise procedure was applied in all specifications to select the most relevant explanatory variables, ensuring the parsimony and statistical significance of the models.
a) Model without lag:
The first estimated model, without the use of temporal lags, presented results that offer an initial view of the contemporary relationships between the variables.
Table 1. Model output without time lag
|
Quarter |
Individual Commercial Credit Card Balance (R$ thousand) |
Portfolio Growth by Quarter (R$ thousand) |
Portfolio Growth by Quarter (%) |
|
1T15 |
2340000.00 |
40000.00 |
1.74 |
|
2T15 |
2350000.00 |
10000.00 |
0.43 |
|
3T15 |
2350000.00 |
0.00 |
|
|
4T15 |
2290000.00 |
60000.00 |
-2.55 |
|
1Term Paper Stages |
2320000.00 |
30000.00 |
1.31 |
|
2T16 |
2330000.00 |
10000.00 |
0.43 |
|
3T16 |
2290000.00 |
40000.00 |
-1.72 |
|
4T16 |
2230000.00 |
60000.00 |
-2.62 |
|
1T17 |
2250000.00 |
20000.00 |
0.90 |
|
2T17 |
2270000.00 |
20000.00 |
0.89 |
|
3T17 |
2280000.00 |
10000.00 |
0.44 |
|
4T17 |
2270000.00 |
10000.00 |
-0.44 |
|
1T18 |
2363680.29 |
93680.29 |
4.13 |
|
2T18 |
2421545.21 |
57864.91 |
2.45 |
|
3T18 |
2438329.17 |
16783.97 |
0.69 |
|
4T18 |
2483729.28 |
45400.10 |
1.86 |
|
1T19 |
2544798.72 |
61069.44 |
2.46 |
|
2T19 |
2622957.60 |
78158.88 |
3.07 |
|
3T19 |
2655954.63 |
32997.03 |
1.26 |
|
4T19 |
2694406.39 |
38451.76 |
1.45 |
|
1T20 |
2876676.49 |
182270.10 |
6.76 |
|
2T20 |
2912390.15 |
35713.66 |
1.24 |
|
3Q20 |
3041734.63 |
129344.48 |
4.44 |
|
4Q20 |
3210636.72 |
168902.09 |
5.55 |
|
1T21 |
3401280.71 |
190643.99 |
5.94 |
|
2T21 |
3643911.18 |
242630.47 |
7.13 |
|
3T21 |
3836068.64 |
192157.46 |
5.27 |
|
4T21 |
4027600.31 |
191531.67 |
4.99 |
|
1T22 |
4244872.41 |
217272.10 |
5.39 |
|
2T22 |
4521202.96 |
276330.55 |
6.51 |
|
3Q22 |
4787240.77 |
266037.81 |
5.88 |
|
4T22 |
5148001.83 |
360761.07 |
7.54 |
|
1T23 |
5562775.06 |
414773.23 |
8.06 |
|
2T23 |
5970974.25 |
408199.19 |
7.34 |
|
3T23 |
6315490.45 |
344516.20 |
5.77 |
|
4T23 |
6567490.18 |
251999.73 |
3.99 |
|
1T24 |
6910329.97 |
342839.79 |
5.22 |
|
2T24 |
7309712.34 |
399382.38 |
5.78 |
|
3T24 |
7692898.74 |
383186.39 |
5.24 |
|
4T24 |
8043955.98 |
351057.25 |
4.56 |
|
1T25 |
8372414.00 |
328458.02 |
4.08 |
|
2T25 |
8535345.30 |
162931.30 |
1.95 |
|
3T25 |
8653657.42 |
118312.11 |
1.39 |
|
4T25 |
8736579.00 |
82921.59 |
0.96 |
Source: Own elaboration based on Stats Studio results
As shown in Table 1, the model estimation resulted in a parsimonious specification, in which only the GDP and unemployment rate variables were kept as explanatory of the growth in the credit portfolio of individuals. The coefficient of determination (R²) of 0.4168 indicates that approximately 41.68% of the variance in credit portfolio growth is explained by the selected variables. Although this R² is moderate, the F-test presented a p-value of 0.0000, allowing the null hypothesis that the coefficients are jointly equal to zero to be rejected, which demonstrates the overall significance of the model at the 5% level.
The coefficient associated with GDP was positive and statistically significant, indicating that an increase in economic activity is directly related to credit portfolio expansion. This result is in full agreement with economic theory, which postulates that product growth raises households’ disposable income and, consequently, stimulates both consumption and credit demand (Mankiw, 2021; Assaf Neto, 2015). For the banking institution, this finding reinforces the importance of aligning credit granting strategies with macroeconomic projections of GDP growth, taking advantage of expansion periods to increase credit supply and mitigate risks during slowdown phases.
On the other hand, the unemployment rate also showed statistical significance at the 5% level, but with a positive coefficient. This result diverges from the theoretical expectation, as an increase in unemployment tends to reduce individuals’ payment capacity and, consequently, credit demand, leading to portfolio contraction (Buss, 2024). This observed positive relationship should be interpreted in light of the context of the analyzed period, especially during the COVID-19 crisis. In this interval, the rise in unemployment occurred simultaneously with a significant reduction in the Selic rate, which made credit cheaper and may have encouraged its use as a mechanism to smooth consumption and maintain families’ living standards. Therefore, this result reflects an atypical context and should be analyzed with caution, suggesting that credit may have acted as a financial “cushion” in times of adversity. Alternatively, this relationship may reflect specific characteristics of the analyzed portfolio, such as the predominance of lower-risk credit modalities or the institution’s ability to manage risk effectively even in adverse scenarios.
The Selic rate and inflation (IPCA) variables were excluded by the stepwise procedure in this non-lagged model, indicating that, in the contemporary specification, they did not present a statistically significant contribution to explaining the variation in the credit portfolio. This result does not imply the absence of an economic effect of these variables, but suggests that their impacts may occur indirectly or with a time lag, not being captured by this more simplified model. For managers, the contemporary non-significance of Selic and IPCA in this initial model highlights the complexity of macroeconomic shock transmission and the need to consider the temporal dimension in the analysis.
In summary, the results of the model without lag indicate that the dynamics of the credit portfolio for individuals are more strongly associated with the economic activity levels and labor market conditions in the short term. However, the presence of a coefficient with an unexpected sign for unemployment and the absence of temporal lags highlight limitations in the model specification, suggesting that a more in-depth analysis, with the inclusion of lags, would be more appropriate to capture the complexity of the relationships.
b) Model with a 1-quarter lag:
The introduction of time lags in the explanatory variables is crucial to capture the dynamic effects of macroeconomic conditions on the credit portfolio, as such impacts do not manifest immediately. The model with a one-quarter lag sought to address this issue.
Table 2. Model output with a 1-quarter lag
|
Quarter |
Selic Target (% p.a.) |
Selic Target Rate (% p.a.) |
t-2 Selic Target Rate (% p.a.) |
t-3 Selic Target Rate (% p.a.) |
Selic Meta Rate (% p.a.) |
|
1T15 |
12.75 |
11.32 |
11 |
10.99 |
10.51 |
|
2T15 |
13.24 |
12.75 |
11.32 |
11 |
10.99 |
|
3T15 |
14.09 |
13.24 |
12.75 |
11.32 |
11 |
|
4T15 |
14.25 |
14.09 |
13.24 |
12.75 |
11.32 |
|
1Term Paper Stages |
14.25 |
14.25 |
14.09 |
13.24 |
12.75 |
|
2T16 |
14.25 |
14.25 |
14.25 |
14.09 |
13.24 |
|
3T16 |
14.25 |
14.25 |
14.25 |
14.25 |
14.09 |
|
4T16 |
13.97 |
14.25 |
14.25 |
14.25 |
14.25 |
|
1T17 |
12.78 |
13.97 |
14.25 |
14.25 |
14.25 |
|
2T17 |
11.05 |
12.78 |
13.97 |
14.25 |
14.25 |
|
3T17 |
9.28 |
11.05 |
12.78 |
13.97 |
14.25 |
|
4T17 |
7.57 |
9.28 |
11.05 |
12.78 |
13.97 |
|
1T18 |
6.83 |
7.57 |
9.28 |
11.05 |
12.78 |
|
2T18 |
6.50 |
6.83 |
7.57 |
9.28 |
11.05 |
|
3T18 |
6.50 |
6.50 |
6.83 |
7.57 |
9.28 |
|
4T18 |
6.50 |
6.50 |
6.50 |
6.83 |
7.57 |
|
1T19 |
6.50 |
6.50 |
6.50 |
6.50 |
6.83 |
|
2T19 |
6.50 |
6.50 |
6.50 |
6.50 |
6.50 |
|
3T19 |
6.10 |
6.50 |
6.50 |
6.50 |
6.50 |
|
4T19 |
5.05 |
6.10 |
6.50 |
6.50 |
6.50 |
|
1T20 |
4.28 |
5.05 |
6.10 |
6.50 |
6.50 |
|
2T20 |
3.19 |
4.28 |
5.05 |
6.10 |
6.50 |
|
3Q20 |
2.10 |
3.19 |
4.28 |
5.05 |
6.10 |
|
4Q20 |
2.00 |
2.10 |
3.19 |
4.28 |
5.05 |
|
1T21 |
2.12 |
2.00 |
2.10 |
3.19 |
4.28 |
|
2T21 |
3.33 |
2.12 |
2.00 |
2.10 |
3.19 |
|
3T21 |
4.96 |
3.33 |
2.12 |
2.00 |
2.10 |
|
4T21 |
7.68 |
4.96 |
3.33 |
2.12 |
2.00 |
|
1T22 |
10.37 |
7.68 |
4.96 |
3.33 |
2.12 |
|
2T22 |
12.46 |
10.37 |
7.68 |
4.96 |
3.33 |
|
3Q22 |
13.57 |
12.46 |
10.37 |
7.68 |
4.96 |
|
4T22 |
13.75 |
13.57 |
12.46 |
10.37 |
7.68 |
|
1T23 |
13.75 |
13.75 |
13.57 |
12.46 |
10.37 |
|
2T23 |
13.75 |
13.75 |
13.75 |
13.57 |
12.46 |
|
3T23 |
13.38 |
13.75 |
13.75 |
13.75 |
13.57 |
|
4T23 |
12.33 |
13.38 |
13.75 |
13.75 |
13.75 |
|
1T24 |
11.36 |
12.33 |
13.38 |
13.75 |
13.75 |
|
2T24 |
10.60 |
11.36 |
12.33 |
13.38 |
13.75 |
|
3T24 |
10.53 |
10.60 |
11.36 |
12.33 |
13.38 |
|
4T24 |
11.27 |
10.53 |
10.60 |
11.36 |
12.33 |
|
1T25 |
13.06 |
11.27 |
10.53 |
10.60 |
11.36 |
|
2T25 |
14.58 |
14.58 |
11.27 |
10.53 |
10.60 |
|
3T25 |
15.00 |
15.00 |
15.00 |
11.27 |
10.53 |
|
4T25 |
15.00 |
15.00 |
15.00 |
15.00 |
11.27 |
Source: Original results from the Stats Studio software survey
As shown in Table 2, this model with a one-quarter lag presented a coefficient of determination (R²) of 0.3588, indicating a moderate capacity to explain the variation in the credit portfolio of individuals. Although slightly lower than the model without lag, the F-test presented a p-value of 0.0001, evidencing that the model is statistically significant as a whole. The variables selected by the stepwise procedure were the Selic rate and GDP, both with statistically significant coefficients at the 5% level.
The coefficient associated with the Selic rate, with a one-quarter lag, showed a negative sign, indicating that the increase in the benchmark interest rate in the previous quarter is associated with a reduction in credit portfolio growth in the subsequent period. This result is highly consistent with the transmission mechanism of monetary policy, according to which an increase in interest rates makes credit more expensive, discourages demand for financing, and consequently, contracts the portfolio balance (Linardi, 2008; Murtadha, 2025). The one-quarter lag (approximately three months) for this effect to materialize is plausible, considering the time required for monetary policy decisions to translate into changes in market conditions and the behavior of borrowers and lenders. For the banking institution, this evidence reinforces the need to monitor COPOM’s decisions and anticipate their impacts on credit demand, adjusting the interest rates of its products and lending policies based on expectations of Selic movements.
In turn, the GDP lagged by one quarter also presented a positive and statistically significant coefficient, indicating that the increase in economic activity in the previous period contributed to the expansion of the credit portfolio. This result reinforces the importance of the economic cycle as a determinant of credit dynamics, showing that the improvement in economic conditions translates, with some lag, into greater demand for credit from households. Economic confidence and employment stability, driven by GDP growth, encourage households to take on long-term financial commitments. For banks, the positive GDP lag suggests that credit portfolio expansion planning can be based on leading economic indicators, allowing for proactivity in credit supply and risk management.
The inclusion of the temporal lag allowed for a more adequate capture of the effects of macroeconomic variables on credit, as such impacts do not occur immediately. In this sense, the results obtained for the model with a one-quarter lag show greater coherence with economic theory when compared to the model without a lag, especially due to the presence of the Selic rate with the expected sign. This model offers a more robust basis for decision-making, allowing the banking institution to develop credit forecasting models that incorporate the temporal dynamics of monetary policy and the economic cycle, optimizing product pricing and risk portfolio management.
c) Model with a 2-quarter lag:
The analysis proceeded with the estimation of a model that incorporates two-quarter lags for the explanatory variables, seeking to verify if the effects of macroeconomic conditions manifest over a more extended time horizon.
Table 3. Model output with a 2-quarter lag
|
Variable |
Coefficient (β) |
p-value |
|
Constant α |
-0,16085 |
0,02448 |
|
β1 |
-0,00341 |
0,00105 |
|
B2 |
0,00130 |
0,00295 |
|
Model Indicators |
Value |
|
|
R2 |
0,2928 |
|
|
Adjusted R² |
0,2583 |
|
|
F (p-value) |
0,0008 |
Source: Original results from the Stats Studio software survey
The model with a two-quarter lag, whose results are presented in Table 3, exhibited a coefficient of determination (R²) of 0.2928, indicating a lower explanatory power compared to the model with a one-quarter lag. Despite the reduction in R², the F-test indicated global significance of the model (p-value = 0.0008), confirming that the selected variables are jointly relevant. The variables selected by the stepwise procedure were the Selic rate and GDP, both statistically significant at the 5% level.
The Selic rate coefficient maintained its negative sign, indicating that interest rate hikes with a two-quarter lag continue to be associated with a reduction in credit portfolio growth. This result reinforces the persistence of the monetary policy effect, albeit with less explanatory intensity. The transmission of monetary policy, which can take six to nine months to fully materialize, as indicated by the Central Bank of Brazil, is supported by the observation of lagged effects. For risk management, this implies that today’s monetary policy decisions can impact the credit portfolio for an extended period, requiring a long-term view in strategy formulation.
Similarly, GDP lagged by two quarters showed a positive coefficient, evidencing that economic activity continues to influence credit expansion, even in a more distant temporal horizon. Economic growth, by generating income and employment, creates a favorable environment for credit expansion, and these effects can be prolonged. However, the reduction in the explanatory power of this model compared to that of one quarter suggests that the effects of macroeconomic variables on the credit portfolio tend to manifest with greater intensity in the short and medium terms. For public policymakers, the persistence of the GDP effect on credit, even with a lag, underscores the importance of policies to stimulate economic growth for the stability and expansion of the financial system.
In summary, although the model with a two-quarter lag shows coherence in the estimated signals, its lower explanatory power compared to the one-quarter model indicates that this specification may not be the most suitable for capturing credit dynamics in the analyzed period. Bank managers should consider that, although long-term effects exist, the most immediate responses of the credit portfolio are more sensitive to recent macroeconomic conditions.
d) Model with a 3-quarter lag:
The investigation of lagged effects continued with the estimation of a model with a three-quarter lag, seeking to identify patterns of influence over an even more extended time horizon.
Table 4. Model output with a 3-quarter lag
|
Quarter |
GDP (index-number) |
t-1 GDP (index-number) |
GDP (index number) |
t-3 GDP (index-number) |
t-4 GDP (index-number) |
|
1T15 |
173.86 |
175.23 |
174.32 |
174.42 |
177.05 |
|
2T15 |
169.75 |
173.86 |
175.23 |
174.32 |
174.42 |
|
3T15 |
166.95 |
169.75 |
173.86 |
175.23 |
174.32 |
|
4T15 |
165.67 |
166.95 |
169.75 |
173.86 |
175.23 |
|
1Term Paper Stages |
163.3 |
165.67 |
166.95 |
169.75 |
173.86 |
|
2T16 |
163.75 |
163.3 |
165.67 |
166.95 |
169.75 |
|
3T16 |
162.86 |
163.75 |
163.3 |
165.67 |
166.95 |
|
4T16 |
162.7 |
162.86 |
163.75 |
163.3 |
165.67 |
|
1T17 |
164.35 |
162.7 |
162.86 |
163.75 |
163.3 |
|
2T17 |
165.73 |
164.35 |
162.7 |
162.86 |
163.75 |
|
3T17 |
166.12 |
165.73 |
164.35 |
162.7 |
162.86 |
|
4T17 |
167.02 |
166.12 |
165.73 |
164.35 |
162.7 |
|
1T18 |
168.01 |
167.02 |
166.12 |
165.73 |
164.35 |
|
2T18 |
167.85 |
168.01 |
167.02 |
166.12 |
165.73 |
|
3T18 |
169.46 |
167.85 |
168.01 |
167.02 |
166.12 |
|
4T18 |
169.01 |
169.46 |
167.85 |
168.01 |
167.02 |
|
1T19 |
169.54 |
169.01 |
169.46 |
167.85 |
168.01 |
|
2T19 |
170.49 |
169.54 |
169.01 |
169.46 |
167.85 |
|
3T19 |
170.53 |
170.49 |
169.54 |
169.01 |
169.46 |
|
4T19 |
172.02 |
170.53 |
170.49 |
169.54 |
169.01 |
|
1T20 |
168.1 |
172.02 |
170.53 |
170.49 |
169.54 |
|
2T20 |
153.16 |
168.1 |
172.02 |
170.53 |
170.49 |
|
3Q20 |
165.3 |
153.16 |
168.1 |
172.02 |
170.53 |
|
4Q20 |
171.63 |
165.3 |
153.16 |
168.1 |
172.02 |
|
1T21 |
173.16 |
171.63 |
165.3 |
153.16 |
168.1 |
|
2T21 |
171.96 |
173.16 |
171.63 |
165.3 |
153.16 |
|
3T21 |
172.18 |
171.96 |
173.16 |
171.63 |
165.3 |
|
4T21 |
174.46 |
172.18 |
171.96 |
173.16 |
171.63 |
|
1T22 |
175.72 |
174.46 |
172.18 |
171.96 |
173.16 |
|
2T22 |
177.74 |
175.72 |
174.46 |
172.18 |
171.96 |
|
3Q22 |
179.56 |
177.74 |
175.72 |
174.46 |
172.18 |
|
4T22 |
180.3 |
179.56 |
177.74 |
175.72 |
174.46 |
|
1T23 |
182.79 |
180.3 |
179.56 |
177.74 |
175.72 |
|
2T23 |
184.25 |
182.79 |
180.3 |
179.56 |
177.74 |
|
3T23 |
184.44 |
184.25 |
182.79 |
180.3 |
179.56 |
|
4T23 |
185.03 |
184.44 |
184.25 |
182.79 |
180.3 |
|
1T24 |
186.47 |
185.03 |
184.44 |
184.25 |
182.79 |
|
2T24 |
189.63 |
186.47 |
185.03 |
184.44 |
184.25 |
|
3T24 |
191.19 |
189.63 |
186.47 |
185.03 |
184.44 |
|
4T24 |
191.19 |
191.19 |
189.63 |
186.47 |
185.03 |
|
1T25 |
194.02 |
191.19 |
191.19 |
189.63 |
186.47 |
|
2T25 |
194.59 |
194.02 |
191.19 |
191.19 |
189.63 |
|
3T25 |
194.64 |
194.59 |
194.02 |
191.19 |
191.19 |
|
4T25 |
194.92 |
194.64 |
194.59 |
194.02 |
191.19 |
Source: Original results from the Stats Studio software survey
The model with a three-quarter lag, according to Table 4, presented a coefficient of determination (R²) of 0.2954, indicating low explanatory power in relation to the other estimated specifications. Despite this, the F-test showed global significance of the model (p-value = 0.0008). The variables selected for this model were GDP and the unemployment rate, both statistically significant at the 5% level.
The GDP lagged by three quarters maintained its positive coefficient, reinforcing the relevance of economic activity in explaining credit dynamics, even with a longer lag. This finding suggests that the effects of the economic cycle can be prolonged, persistently impacting the credit portfolio’s behavior. The resilience of GDP as an explanatory variable at different lags underscores its fundamental importance for the health of the credit market.
However, the unemployment rate again showed a positive coefficient, a result that diverges from theoretical expectation. This inconsistency, already observed in the model without lag, suggests that, over a three-quarter horizon, the relationship between unemployment and credit may be influenced by factors not captured by the model or by atypical dynamics, such as the use of credit to smooth consumption during crises. The persistence of this unexpected signal raises questions about the robustness of the economic interpretation for this variable in longer lags, especially if the low interest rate environment that encouraged subsistence credit does not continue.
In general, the model with a three-quarter lag did not show significant gains in terms of explanatory power and maintained inconsistencies in the economic interpretation of the unemployment coefficient. This makes it a less suitable specification for representing the dynamics of the credit portfolio in the analyzed period, suggesting that the most relevant effects are concentrated in shorter lags.
e) Model with a 4-quarter lag:
The last estimated specification incorporated four-quarter lags, seeking to capture long-term effects of macroeconomic variables on the credit portfolio.
Table 5. Model output with a 4-quarter lag
|
Variable |
Coefficient (β) |
p-value |
|
Constant α |
-0,52479 |
0,00003 |
|
β1 |
0,00263 |
0,00149 |
|
B2 |
0,00968 |
0,00000 |
|
R2 |
0,4258 |
|
|
Adjusted R² |
0,3978 |
|
|
F (p-value) |
0,0000 |
Source: Original results from the Stats Studio software survey
As shown in Table 5, the model with a four-quarter lag presented the highest coefficient of determination (R²=0.4258) among all the specifications analyzed. This high R² indicates a greater explanatory capacity of the variation in the credit portfolio of individuals. The F test presented a p-value of 0.0000, evidencing the high statistical significance of the model as a whole. The variables selected by the stepwise procedure were GDP and the unemployment rate, both statistically significant at the 5% level.
The GDP lagged by four quarters maintained its positive coefficient, reinforcing the evidence that economic activity exerts relevant influence on credit expansion, albeit with a longer time lag. This result suggests that the effects of the economic cycle can be prolonged and impact the behavior of the credit portfolio persistently, for up to a year. For the institution’s strategic planners, this implies that investment and portfolio expansion decisions must consider medium to long-term economic trends, using GDP as a key indicator to anticipate future scenarios.
On the other hand, the unemployment rate again showed a positive coefficient, maintaining the divergence from theoretical expectation. The persistence of this unexpected signal, even with the model’s greater explanatory power, indicates that the economic interpretation for this variable should be made with caution. The absence of the Selic rate in this model, despite the better statistical performance, and the maintenance of the unexpected signal for unemployment, suggest that, although the model with a four-quarter lag presents a superior statistical fit, it may not constitute the most theoretically appropriate specification for a direct causal interpretation. The complexity of macroeconomic interactions and the possibility of omitted variables or nonlinear transmission channels may explain these inconsistencies.
Comparison between models and definition of the final specification
The comparative synthesis of the estimated models, with and without temporal lag, is fundamental to jointly evaluate the statistical performance and economic coherence of the different specifications, allowing the definition of the most adequate final specification for the interpretation of the results.
Table 6. Comparative synthesis of regression models
|
Variable |
No lag |
t-1 |
t-2 |
t-3 |
t-4 |
|
Constant α |
-0,58149 |
-0,20524 |
-0,16085 |
-0,45087 |
-0,52479 |
|
ẞ SELIC TARGET |
-0,00347 |
-0,00341 |
|||
|
ẞ GDP |
0,00302 |
0,00155 |
0,00130 |
0,00231 |
0,00263 |
|
ẞ Unemployment |
0,00827 |
0,00772 |
0,00968 |
||
|
R² |
0,4168 |
0,3588 |
0,2928 |
0,2954 |
0,4258 |
|
Adjusted R² |
0,3883 |
0,3275 |
0,2583 |
0,2610 |
0,3978 |
|
F (p-value) |
0,0000 |
0,0001 |
0,0008 |
0,0008 |
0,0000 |
Source: Original results from the Stats Studio software survey
As shown in Table 6, Gross Domestic Product (GDP) proved to be the most robust explanatory variable, presenting a positive and statistically significant coefficient in all models where it was selected by the stepwise procedure. This result reinforces the central role of economic activity in determining credit dynamics, confirming the literature that points to the procyclical nature of credit (Tiryaki et al., 2017; Davoglio, 2019). The consistency of GDP across different lags and specifications underscores its importance as a key indicator for credit portfolio management and for the formulation of economic policies.
It is noteworthy that the inflation variable, represented by the IPCA, was not selected in any of the specifications estimated by the stepwise procedure, regardless of the lag considered. This result suggests that, for this model and in the analyzed period, inflation did not present a statistically significant contribution to the explanation of the variation in the credit portfolio of individuals. Such evidence does not imply the absence of an economic effect of inflation on credit, but it may indicate that its impacts occur indirectly, possibly already incorporated into other variables in the model, such as interest rates, or that they manifest with a greater lag or through channels not captured by the adopted specification. For the institution, this may mean that, although inflation affects purchasing power and payment capacity, its direct effects on credit volume are less noticeable than those of other macroeconomic variables.
In the non-lagged model, the GDP and unemployment rate variables were selected, with a relatively high coefficient of determination (R²=0.4168). However, the presence of a positive coefficient for unemployment indicates an inconsistency with economic theory, suggesting limitations in the contemporary specification, especially due to the atypical context of the COVID-19 pandemic, where credit may have been used to smooth consumption in a low-interest rate scenario.
The introduction of time lags allowed for the capture of dynamic effects of macroeconomic variables, which are crucial for understanding the transmission of economic shocks. The model with a one-quarter lag stood out for presenting coefficients with signs consistent with economic theory, especially in the case of the Selic rate, which exhibited a negative relationship with the growth of the credit portfolio. This result is compatible with the transmission mechanism of monetary policy, in which increases in the interest rate tend to restrict credit (Linardi, 2008). The theoretical coherence of this model, combined with its statistical significance, makes it a valuable specification for interpretation.
However, the model with a four-quarter lag showed the highest coefficient of determination (R²=0.4258), indicating the best statistical fit. Despite this, the absence of the Selic rate and the maintenance of the positive coefficient for the unemployment rate indicate limitations in economic interpretation, despite the greater explanatory power. This suggests a trade-off between statistical fit and theoretical coherence. The models with two and three-quarter lags, in turn, showed lower explanatory capacity and did not bring relevant gains in terms of economic interpretation, not standing out as adequate specifications.
Thus, there is a trade-off between the quality of statistical fit and the theoretical coherence of the models. Considering the consistency of the estimated signals, the presence of economically relevant variables, and alignment with the literature, the model with a one-quarter lag proved to be the most suitable for interpreting the results. This model offers the best basis of estimates for the banking institution to develop portfolio management and product pricing strategies that consider the dynamics of monetary policy and the economic cycle more realistically.
Finally, the model with a four-quarter lag is used as a complementary reference for robustness, evidencing that the effects of macroeconomic variables on the credit portfolio can also manifest in the long term, albeit with some interpretative limitations. The ability of a model with a longer lag to explain a larger proportion of the credit portfolio’s variance suggests that, for certain long-term strategic decisions, the institution may consider these more prolonged effects, even with interpretative caveats. The combination of analyzing models with different lags offers a comprehensive and multifaceted view of the complex relationship between the macroeconomic environment and the performance of the retail credit portfolio.
4. Conclusion
It is concluded that the objective was achieved by evaluating the relationship between macroeconomic indicators such as GDP, IPCA, Selic, and unemployment rate and the behavior of the credit portfolio of individuals from a banking institution, in the period from 2015 to 2025. The research showed that Gross Domestic Product (GDP) was the most robust variable, exerting a positive and consistent influence on the growth of the credit portfolio. The Selic rate, when considered with a one-quarter lag, presented the expected negative relationship, confirming the transmission mechanism of monetary policy. However, the unemployment rate exhibited a positive coefficient, suggesting an atypical use of credit to smooth consumption in periods of crisis, while the IPCA did not prove to be statistically significant. The results reinforce that the dynamics of the credit portfolio are intrinsically linked to the economic cycle and respond with a lag to variations in the cost of capital, underscoring the importance of temporal lags for a more precise understanding.
As limitations of the study, the use of national macroeconomic data to explain the behavior of a specific banking institution stands out, as well as the complexity inherent in credit, which can be influenced by unobserved factors or omitted variables. The temporal sample, from 2015 to 2025, although comprehensive, may not capture all the nuances of longer economic cycles. For future studies, it is suggested to explore more complex econometric models capable of capturing more sophisticated temporal dynamics and multiple simultaneous lags. The inclusion of new explanatory variables, such as consumer confidence indicators or microeconomic data from the institution itself, and the application of advanced time series techniques could deepen the understanding of the relationship between the macroeconomic environment and the performance of the credit portfolio, offering even more precise subsidies for risk management and the formulation of banking strategies.
Bibliographic References
Assaf Neto, A. 2015. Mercado financeiro. 16. ed. Atlas, São Paulo, SP, Brasil.
Buss, L. S. 2024. Determinantes macroeconômicos da inadimplência da pessoa física no Sistema Financeiro Nacional Brasileiro. Trabalho de Conclusão de Curso Bacharel em Ciências Econômicas. Universidade Federal de Mato Grosso, Cuiabá, MT, Brasil.
Camargos, M.A.R. 2022. Análise do Comportamento do Nível de Endividamento dos Brasileiros. Trabalho de Conclusão de Curso Bacharel em Ciências Contábeis. Universidade Federal de Uberlândia, Uberlândia, MG, Brasil.
Davoglio, T. R. 2019. O impacto do desempenho setorial na inadimplência de carteiras de crédito. Trabalho de Conclusão de Curso Bacharel em Administração. Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brasil.
Gil, A. C. 2023. Como elaborar projetos de pesquisa. 7. ed. Atlas, São Paulo, SP, Brasil.
Linardi, F.M. 2008. Avaliação dos Determinantes Macroeconômicos da Inadimplência Bancária no Brasil. Dissertação de Mestrado em Economia. Universidade Federal de Minas Gerais, Belo Horizonte, MG, Brasil.
Maia, A.C. 2025. Influência dos indicadores macroeconômicos na inadimplência de pessoas físicas no Brasil. Revista Delos 18(67):01-29.
Mankiw, N. G. 2021. Macroeconomia. 10. ed. LTC, Rio de Janeiro, RJ, Brasil.
Murtadha, N. A. 2025. Cartão de crédito: prevendo a taxa de inadimplência através dos indicadores macroeconômicos. Dissertação de Mestrado em Economia. Universidade Católica de Brasília, Brasília, DF, Brasil.
Silveira, J. R. 2015. Crédito e endividamento no Brasil: fatores macroeconômicos e impactos sociais. Trabalho de Conclusão de Curso Bacharel em Ciências Econômicas. Universidade Federal de Pelotas, Pelotas, RS, Brasil.
Tiryakia, G.F.; Gavazzab, I.O.; Andradec, C.M.; Motad, A.L. 2017. Ciclos de crédito, inadimplência e as flutuações econômicas no Brasil. Revista de Economia Contemporânea 21(1): 1-33.
Article originating from the Final Course Work of the Specialization in Economics, Investments and Banking of the MBA USP/Esalq
To learn more about the course, click here and access the MBX Academy platform