The dynamics of the Brazilian manufacturing industry from 2002 to 2024: empirical research in light of the New-Developmentalist Theory

ABSTRACT

This study examines Brazil’s manufacturing industry (2002-2024) through New-Developmentalist Theory. Using a VECM, it shows that real exchange rate appreciation and import penetration are key drivers of manufacturing sector’s decline, amplified by commodity cycles (Dutch disease) and hysteresis. Three major structural breaks are identified: commodity-driven growth (2003-2008), post-2008 decline, and post-2017 stagnation. Findings indicate that exchange rate adjustments alone are insufficient: coordinated industrial, technological, and investment policies are required to restore manufacturing dynamism and strengthen Brazil’s role in the global market.

JEL Classification: F43, F63, O14, F41, L60

KEYWORDS:
New developmentalism; Brazilian manufacturing industry; real exchange rate; import penetration; Dutch disease

INTRODUCTION

This paper investigates the dynamics of Brazil’s manufacturing industry between 2002 and 2024 through the lens of New-Developmentalist Theory. It examines how structural and macroeconomic variables, such as the real exchange rate and manufacturing import penetration, have impacted Brazilian manufacturing output performance and integration within global commerce. Emphasizing empirical findings, the study identifies key trends and structural challenges linked to manufacturing decline, economic stagnation, and external shocks.

The decline of Brazil’s manufacturing industry is evident in its shrinking share of GDP – falling from around 25% in the 1980s to approximately 10% in recent years. Additionally, the manufacturing industry’s real GDP growth rate has deteriorated significantly: it averaged 3.9% between 2002 and 2008 but dropped to -1.9% from 2009 to 2016 and remained stagnant at 0.1% between 2017 and 2024, according to IBGE data presented in Section 4.

The debate on the decline of Brazil’s manufacturing industry has been a central theme in economic discussions over recent decades. This process, which intensified from the 1980s onwards, is attributed to multiple factors, including unregulated economic liberalization and the adoption of adverse macroeconomic policies. During this period, Brazil underwent significant economic transformations driven by the adoption of neoliberal principles promoted by high-income countries, particularly the United States, under the framework of the “Washington Consensus.” Policies advocated by institutions such as the IMF and the World Bank – such as trade liberalization, capital flow deregulation, and restrictive fiscal policies – shaped Brazil’s economic trajectory, accelerating its integration into the global economy.

Although this integration was perceived as a means of fostering economic modernization, it brought about substantial challenges. Globalization and market liberalization facilitated greater interconnectivity among economies but also led to the fragmentation of global production. Transnational corporations began to operate within globally integrated networks, known as Global Value Chains (GVCs), reallocating lower-value-added processes to countries with cheaper and more abundant labor while retaining higher-value-added activities, such as research and development, in their headquarters in high-income countries (Sarti and Laplane, 2021).

In Brazil, integration into GVCs has been characterized by significant asymmetries. The country’s participation in global value chains has been predominantly concentrated in lower-value-added sectors, posing challenges to the competitiveness of its domestic industry (Tregenna and Andreoni, 2020). Disorganized economic liberalization, combined with the absence of a robust industrial policy, has exacerbated these issues, contributing to a deindustrialization process that has dramatically reduced the manufacturing sector’s share in Brazil’s GDP.

Furthermore, the adverse macroeconomic environment, marked by inflation-targeting policies with high interest rates and an overvalued exchange rate policy, has further eroded the dynamism of Brazilian manufacturing sector. These factors, coupled with premature deindustrialization and technological gaps, have prevented Brazil from fully leveraging the potential benefits of its integration into global market, such as productivity gains and technology transfers (Palma, 2019).

Through a Vector Error Correction Model (VECM), this study examines the structural evolution and performance of Brazil’s manufacturing industry from 2002 to 2024. The analysis identifies major phases: an external boom driven by commodity cycles during the 2000s, a sharp decline following the 2008 global financial crisis, and stagnation after 2017. Empirical findings highlight the critical roles of exchange rate dynamics, trade patterns, and import penetration in shaping manufacturing performance. In particular, the results underscore the indirect effects of commodity exports through exchange rate appreciation, consistent with the Dutch disease mechanism, which additionally constrained competitiveness and long-term investment in the manufacturing sector. The study provides evidence supporting the New-Developmentalist Theory but emphasizes that exchange rate management alone is insufficient for recovery and that coordinated industrial policies, demand- and investment-stimulating measures, technological upgrading, and strategic mitigation of Dutch disease effects are necessary to restore manufacturing dynamism and promote sustainable growth.

The remainder of the paper is structured as follows: Section 2 reviews the literature on New-Developmentalist Theory as the theoretical foundation. Section 3 outlines the methodology, emphasizing the application of the Vector Error Correction Model (VECM) and its variables to capture the interplay among exchange rate dynamics, manufacturing import penetration, manufacturing terms of trade, commodity exports, and manufacturing performance. Section 4 discusses the empirical results, highlighting the structural breaks identified, the hysteresis effects, and the indirect role of commodity cycles through exchange rate appreciation, consistent with Dutch disease. Finally, Section 5 draws the conclusions, offering policy recommendations and directions for future research, including strategies to mitigate regressive integration into global markets, reduce import dependency, and restore the long-term dynamism of Brazil’s manufacturing industry.

NEW DEVELOPMENTALISM THEORY REVIEW

New Developmentalism, as articulated by Bresser-Pereira (2016), integrates elements of Classical Political Economy, Keynesianism, and Latin-American Structuralism, addressing the challenges faced by middle-income countries like Brazil. It advocates a development model centered on industrialization and productive sophistication, with the state playing a strategic role in policy formulation and economic coordination. The manufacturing sector is recognized as a cornerstone of economic development, essential for driving structural transformation, adopting advanced technologies, and increasing productivity.

The theory highlights five key roles of manufacturing industry. First, it facilitates innovation and creates high-value jobs, relying on a competitive exchange rate and supportive policies. Second, it promotes competitive international integration through the export of manufactured goods rather than reliance on commodities, supported by an industrial equilibrium exchange rate. Third, it prioritizes neutralizing Dutch disease and chronic exchange rate overvaluation, essential to maintaining global competitiveness and preventing premature deindustrialization. Fourth, industrialization drives sustained economic growth by boosting productivity, raising wages, and reducing inequalities. Fifth, New Developmentalism critiques comparative advantage theory, arguing it confines low- and middle-income economies to low-value-added sectors like commodity exports. Instead, it emphasizes industrial policy and a competitive exchange rate to enable productive diversification and reduce dependency.

New Developmentalism opposes neoliberal practices such as high-interest rates and exchange-rate anchors, which harm industrial competitiveness and long-term growth. Instead, it emphasizes balanced macroeconomic prices – including profit margins, interest rates, exchange rates, wages, and inflation – as prerequisites for sustainable development. The state plays a critical role in addressing market failures, maintaining a competitive exchange rate to counter Dutch disease, and investing in infrastructure, science, technology, and education. It also promotes inclusive development, balancing economic growth with social justice and environmental protection.

In low- and middle-income countries like Brazil, currency appreciation remains a persistent issue, driven by factors such as Dutch disease, exchange rate policies for inflation control, speculative capital inflows from high-interest rates, and current account deficits financed by external savings. These factors lead to increased imports, reduced manufacturing exports, and dependence on commodities, resulting in external imbalances. Within the manufacturing sector, currency appreciation erodes competitiveness, accelerates deindustrialization, and limits investment in innovation and technology. On a broader scale, the shrinking manufacturing base restricts GDP growth and heightens vulnerability to global price fluctuations and volatile capital flows. New Developmentalism underscores the need for strategic macroeconomic policies to restore manufacturing competitiveness and achieve sustainable, inclusive development.

Feijó (2024) underscores the relevance of New Developmentalism as a guiding framework for macroeconomic debates in low- and middle-income economies. This perspective emphasizes the critical role of the real exchange rate as a key variable in explaining the dynamics of growth and structural development over time. The study highlights Bresser-Pereira’s innovative approach to exchange rate determination, focusing on its cyclical and chronic tendency toward overvaluation in low- and middle-income countries. The author further explores the relationship between the exchange rate and interest rate, detailing how financial integration limits policy autonomy in low- and middle-income economies.

Borghi (2024) highlights Bresser-Pereira’s contribution in emphasizing the role of exchange rate overvaluation in the context of Dutch disease and financial liberalization, which leads to manufacturing competitiveness losses and a productive-commercial specialization focused on abundant natural resources, with high dependence on external resources and short-term capital flows. High interest rates further fuel rentier practices and raise financing costs for productive investments, thus negatively affecting income and employment creation.

Additionally, Bresser-Pereira et al. (2016) argue that countries whose catching up process has been interrupted should strive to resume it through national reindustrialization strategies. While well-designed and effectively implemented industrial policies are essential, they will only prove effective if integrated into a consistent macroeconomic regime. Such a regime must ensure price and fiscal stability while enabling average real interest rates lower than the real return on capital, maintaining competitive real exchange rates (at the “manufacturing equilibrium” level), and fostering wage growth aligned with productivity increases. However, Brazil has failed to meet these conditions in recent decades, leading to premature deindustrialization, low productivity, and limited economic growth. They suggest that future work will propose concrete measures to align the macroeconomic regime with industrial policy, aiming to promote reindustrialization and resume sustainable economic growth. This alignment is deemed essential for placing the country back on the path of catching up, which was interrupted in the early 1980s.

On the other hand, the emergence of global value chains (GVCs), closely linked to the restructuring of transnational corporations (TNCs) between the mid-1970s and 1990s, added further challenges to low- and middle-income economies. Advances in communication and transport enabled production fragmentation across countries, with specialization based on comparative advantages (Baldwin, 2012).

As Gereffi (1999) emphasizes, GVCs represent a new organizational framework of production, marked by outsourcing and subcontracting, which intensified asymmetries in international economic relations (Hiratuka and Sarti, 2017). Dicken (2011) shows that this reconfiguration increased market concentration, with a small group of large corporations dominating global trade flows, while UNCTAD (2018) highlights that rather than reducing inequality, GVCs exacerbated income polarization. The distribution of value along GVCs, often illustrated by the “smile curve,” concentrates high-value activities such as R&D, design, and marketing in high income economies, while manufacturing and assembly – typical of low- and middle-income countries – capture only a small fraction of total value (OECD, 2013).

This dynamic has reinforced the productive specialization of low- and middle-income economies in commodities and low-value-added segments, contributing to premature deindustrialization (Palma, 2019). In Brazil, participation in global value chains, marked by specialization in low-value-added activities and reliance on imported inputs, has constrained domestic innovation capacity and increased vulnerability to external shocks.

This situation underscores the challenges central to New Developmentalism. While this theoretical framework does not directly address Global Value Chains (GVCs), we propose an integrated approach to understand the dynamics of Brazilian manufacturing production during the period under analysis, as the two perspectives are often complementary. Together, they elucidate how low- and middle-income economies become trapped in regressive integration patterns, which limit their potential for productive diversification and technological upgrading. The GVC lens, in particular, details how structural asymmetries in global production reinforce premature deindustrialization. This analysis thereby strengthens the New-Developmentalist case for coordinated industrial and macroeconomic policies aimed at restoring manufacturing dynamism and achieving structural transformation.

METHODOLOGY

The Vector Error Correction Model (VECM)

As evidenced in the literature, Brazil’s integration into the global market (and GVCs) over the past three decades has occurred in an asymmetric and regressive manner. Between 1999 and 2024, for instance, the export structure was concentrated in commodities and basic inputs (over 65% of the total), while the relative share of capital goods exports in total exports fell by 43%, and consumer goods exports declined by 38% (Table 1). On the import side, there was an increase of approximately 9% in the relative share of manufacturing inputs, which today account for roughly 60% of total imports, alongside a 35% decline in capital goods imports. This pattern reinforces the regressive integration: Brazil exports lower value-added goods while increasingly relying on the import of higher-technology and more sophisticated inputs and products.

Table 1
Brazil’s Manufacturing Industry Imports and Exports, 1999-2024

Category	Capital Goods Imports (FOB, US$ million)	Consumer Goods Imports (FOB, US$ million)	Intermediate Inputs Imports (FOB, US$ million)	Other Imports	Total	Capital Goods Exports (FOB, US$ million)	Consumer Goods Exports (FOB, US$ million)	Intermediate Products Exports (FOB, US$ million)	Other Exports	Total
Relative Share in Total (monthly average, 2024)	13%	15%	60%	13%	100%	5%	13%	65%	17%	100%
Relative Share in Total (monthly average, 1999)	20%	15%	55%	10%	100%	9%	20%	69%	2%	100%
% Change in Relative Share 2024-1999	-35%	-5%	9%	27%	–	-43%	-38%	-5%	750%	–

Figure 1 shows that the year-on-year cumulative variation of the manufacturing industry’s GDP between March 2011 and March 2024 was -21.6%. Its share in total GDP, according to IBGE data, fell from about 25% in the 1980s to around 10% at present. In sum, except during the commodities boom between 2000 and 2008, the period was largely characterized by the loss of dynamism and decline of the manufacturing industry. While acknowledging the internal heterogeneity of Brazil’s manufacturing industry – comprising segments with distinct technological levels and sector-specific dynamics – this study focuses on the aggregate analysis of the manufacturing industry as a whole. This choice reflects the interest in understanding the sector’s structural and macroeconomic evolution, observing its trajectory at the systemic level rather than detailing sector-specific particularities, which may be the subject of future research.

Figure 1
Manufacturing Industry GDP

This trajectory aligns with the framework of New Developmentalism. According to Bresser-Pereira (2016), the real exchange rate constitutes the strategic macroeconomic price for the competitiveness of the manufacturing industry. Cyclical and chronic over-appreciation of the currency, exacerbated by the Dutch disease and high interest rates, favored commodities and natural resources while reducing the competitiveness of manufactured goods and discouraging long-term investment.

To empirically capture these phenomena, variables were selected as appropriate proxies for the New-Developmentalist Theory and GVCs applied to the Brazilian case. The time-series variables and their respective abbreviations, which will be utilized in programming and analysis using R Studio version 4.3.3, are as follows:

• PIBit: manufacturing industry GDP.

• TxCamRe: real exchange rate.

• ExBasCom: exports of basic inputs and commodities goods (export reprimarization).

• PimExpit: import penetration over manufacturing industry exports.

• Ttit: terms of trade for the manufacturing industry.

These variables capture the relationship among exchange rate, trade, and the manufacturing industry, reflecting the dynamics of regressive international integration in GVCs and allowing the central New-Developmentalist hypothesis to be tested: that the real exchange rate is a necessary, albeit not sufficient, condition for manufacturing GDP growth. Moreover, the Dutch disease hypothesis is central in Brazil: exchange rate appreciation, stimulated by commodity export expansion, reduces manufacturing performance, discourages productive investment, and heightens vulnerability in an increasingly connected and GVC-dominated international environment. The result is a loss of dynamism, global market share, and relative technological positioning for Brazilian manufacturing.

Variables such as interest rate, gross fixed capital formation, innovation (R&D), productivity, employment, export value added (FVA), “Brazil cost,” or financing were not included for the following reasons: (i) as will be discussed, the objective of the VECM model is not to infer causality in the traditional econometric sense, nor to project future trends or estimate regime shifts. Rather, it aims to explore the dynamic interdependence among key macro-structural variables and the Brazil’s manufacturing industry GDP, considering their co-evolution between 2002 and 2024; (ii) the institutional or structural nature of these variables makes them more suitable for qualitative analyses, comparative studies, or traditional unidirectional cause-effect approaches, but less compatible with a VECM framework, which requires statistically testable relationships among time series; (iii) moreover, the effects represented by these variables are already indirectly captured by the selected variables, particularly via exchange rate, foreign trade, and manufacturing industry performance; and (iv) limitations in monthly frequency and long series availability prevent their application in cointegration models. Finally, direct GVC measures, such as Foreign Value Added (FVA), were excluded because the relevant databases (TiVA/OECD-WTO, ICIO/ADB, WIOD) – despite being updated in 2023 – only provide annual data with coverage ending in 2020. This makes them incompatible with a monthly cointegration analysis. Therefore, commercial and macroeconomic proxies were chosen to robustly reflect also Brazil’s integration into the global market. This methodological choice will be further detailed below.

The time-series data for the variables used in the study were obtained from reliable public sources. Box 1 provides a description and sources of the time-series data utilized.

Box 1
Description and Sources of Time-Series Data

Series	Definition and Comments
Manufacturing Industry GDP – Physical Output.	Physical Output – Table 8888 – Index number – CNAE 2.0 – Source: IBGE – Monthly Industrial Survey.
Bilateral Real Exchange Rate – IPA-DI – Brazil/United States.	The bilateral real exchange rate is defined as the product of the nominal exchange rate (in R$/unit of foreign currency) and the Producer Price Index (PPI) of the respective country, divided by the Domestic Availability Producer Price Index (IPA-DI) for Brazil. An increase indicates depreciation, and a decrease indicates appreciation. Source: BCB.
Exports (FOB) – USD (millions) Basic Goods + Commodities.	Exports – (FOB) – USD (millions) Basic Goods + Commodities (basic/processed food and beverages + basic industrial goods + fuels) – SECEX12_XVTOT12. Source: IBGE-SECEX.
Terms of Trade for the Manufacturing Industry.	Export-import price ratio – FUNCEX12_MDPBCNDGCE12. Average Index. Source: FGV-FUNCEX.
Import Penetration Ratio (USD) in the Manufacturing Industry.	Import Penetration (Ratio in USD) of the Manufacturing Industry relative to Exports of the Manufacturing Industry – Prepared by the author using data from SECEX12_MCOBLUBGCE12. Source: IPEADATA-SECEX.

The Vector Error Correction Model (VECM), incorporating structural breaks and seasonally adjusted variables, was used to capture both long-term and short-term relationships among the economic variables described above and their influence on the performance of the manufacturing industry’s GDP. The cointegration equation of the model represents long-term relationships among the variables (Bueno, 2012).

The cointegration equation is expressed as follows:

(equation 1) $\text{PIBit}\_\text{t}={\beta }_0+{\beta }_1\text{TxCamRe}\_\text{t}+{\beta }_2\text{ExBasCom}\_\text{t}+{\beta }_3\text{Ttit}\_\text{t}+{\beta }_4\text{PimExpit}\_\text{t}+ϵ\_\text{t}$

Where:

• PIBit_t: manufacturing industry GDP at time t.

• CamRe_t: real exchange rate at time t.

• ExBasCom_t: exports of basic goods and commodities at time t.

• Ttit_t: terms of trade for manufactured goods at time t.

• PimExpit_t: import penetration ratio for manufactured goods at time t.

• ε_t: stochastic error term at time t.

The error correction equation, which represents short-term adjustments when deviations occur from long-term equilibrium, is expressed as:

(equation 2) $\Delta \text{PIBit}\_\text{t}={\alpha }_1\text{ECM}\_\left\{\text{t}-1\right\}+\Sigma \_\left\{\text{i}=1\right\}ˆ\left\{\text{p}\right\}\gamma \_\text{i}\Delta \text{X}\_\{\text{t}-\}+ϵ\text{\_t}$

Where:

• ΔPIBit_t: variation in the manufacturing industry GDP at time t.

• ECM_{t-1}: error correction term measuring deviation from long-term equilibrium in the previous period.

• α₁: short-term adjustment coefficient.

• ΔX_{t-i}: lagged variations of explanatory variables.

• ε_t: stochastic error time t.

Time Series Diagnostics, Stationarity, and Structural Breaks

The ACF and PACF correlogram analyses reveal significant autocorrelation in key variables – particularly Manufacturing Industry GDP and the Real Exchange Rate –indicating non-stationarity in levels and temporal dependence that diminishes after first differencing. These results are confirmed by unit root tests (ADF, PP, and KPSS), which show that most series are non-stationary in levels, with a few test-specific exceptions, while all variables become stationary in first differences.

The structural break tests based on the Bai-Perron methodology identify multiple breakpoints across the series, corresponding to distinct economic phases: the economic boom period (2002-2008), the Global Financial Crisis (2009-2010), a modest and short-term recovery phase (2011-2013), the recession during Dilma Rousseff’s administration (2014-2016) and Temer’s government, as well as the period from 2017 onward, including the COVID-19 pandemic shock in 2020. Due to their limited individual impact, the analysis focuses on the three major structural breaks identified in this article: the commodity-driven growth phase (2003-2008), the post-2008 decline, and the period of stagnation observed from 2017 onward. The OLS-CUSUM test corroborates these findings, detecting structural instability at dates broadly consistent with those identified by the Bai-Perron procedure.

These breaks were incorporated into the VECM via dummy variables, ensuring robustness without compromising cointegration. This enabled the integration of exogenous shocks and historically significant events into the model, either via dummies or through the endogenous identification of breaks.

Johansen Cointegration Analysis and VECM Model Justification

The Johansen cointegration tests indicated the presence of multiple long-term relationships among the variables, as shown in Table 2.

Table 2
Johansen Cointegration Test Results

Hypothesis	Test Statistic	Critical Value (10%)	Critical Value (5%)	Critical Value (1%)
r ≤ 4	2.45	7.52	9.24	12.97
r ≤ 3	10.09	13.75	15.67	20.20
r ≤ 2	15.74	19.77	22.00	26.81
r ≤ 1	34.99	25.56	28.14	33.24
r = 0	190.90	31.66	34.40	39.79

Methodologically, the VECM is well suited to this analysis because the variables are non-stationary and cointegrated (Johansen, r = 2). The model captures short-term dynamics through lagged adjustments and error-correction terms, while long-run relationships are represented by the cointegration vectors. This framework is appropriate when non-stationary variables – such as manufacturing GDP and the real exchange rate – share a common long-term trajectory despite short-run fluctuations.

Unlike univariate approaches (e.g. ARIMA) or regime-switching models, the VECM treats all variables as endogenous, avoiding a priori causal assumptions and allowing for the identification of interdependence and hysteresis effects. This is particularly relevant in contexts such as deindustrialization, where exchange rates, import penetration, and industrial output evolve jointly and respond to mutual shocks.

The model incorporates temporal lags, enabling shocks to propagate gradually across variables. Error-correction coefficients measure the speed of adjustment toward long-run equilibrium, while impulse-response functions trace the magnitude and persistence of dynamic effects. Variance decomposition further assesses the relative contribution of each variable to fluctuations in manufacturing GDP, supporting the analysis of feedback mechanisms and causal interactions inferred directly from the data (Bueno, 2012).

In summary, the VECM with structural breaks constitutes the most consistent method for this study because it (1) simultaneously captures long-term relationships and short-term adjustments; (2) treats all variables as endogenous, avoiding causal reductionism; (3) incorporates relevant historical shocks without violating theoretical coherence; and (4) empirically reflects Brazil’s regressive international integration in line with New-Developmentalist Theory. This model thus allows for the identification of how structural and cyclical factors co-evolved and contributed to the decline of the manufacturing industry during the analyzed period.

RESULTS AND DISCUSSION

Brazilian Manufacturing Industry Data and Analysis (2002-2024)

Firstly, it is important to note that the 2000s were marked by significant growth in the Brazilian economy, including the manufacturing industry, driven by a favorable external environment. According to Borghi and Sarti (2019), this period was characterized by the largest global liquidity cycle, rising commodity and agricultural product prices, increases in the prices of basic inputs, processed goods, and industrial transport components, and domestic consumption growth fueled by higher real income and credit expansion, despite currency appreciation. Despite the currency’s appreciation, the manufacturing industry’s terms of trade improved (Figure 5), leading to greater import penetration at the expense of local production (Figure 4). It can be concluded that the relative price increases for local manufacturing products outpaced the effects of currency appreciation (Figure 2), resulting in higher import penetration during this period.

Nassif et al. conclude that Brazil’s manufacturing industry displayed signs of structural challenges that align with “early deindustrialization” (2015). Despite experiencing some growth in the early 2000s, driven by a favorable external environment, Brazil’s manufacturing sector faced increasing import dependency and an inability to significantly expand its technological capabilities or international competitiveness.

According to Morceiro and Guilhoto (2023), the global financial crisis of 2008 and the stagnation of global growth marked a turning point in Brazil’s economic trajectory, including a decline in the manufacturing industry GDP (Figure 1). The macroeconomic policies implemented by the government after 2010, though not achieving the desired effects, aimed to strengthen the domestic manufacturing industry. These measures included interest rate cuts, currency devaluation, subsidies, and tax relief. However, deteriorating external conditions, with reduced liquidity, sluggish global growth, and Brazil’s fiscal deficits, prompted the government to adopt stringent fiscal adjustments, culminating in two years of economic recession between 2015 and 2016.

Brazil resumed growth in 2017, albeit modestly. However, the manufacturing industry remained stagnant, achieving an average growth rate of only 0.1% between 2017 and 2023. In 2020, the COVID-19 crisis struck the country. Since then, the external scenario has favored exports of basic goods, commodities, and manufacturing products (Figure 3), especially basic and low-tech processed inputs, due to global supply disruptions caused by the pandemic, climate change, inflation, and rising international demand.

Table 3 shows that Manufacturing Industry GDP grew by 3.9% between 2002 and 2008, broadly in line with services (4.0%) but below agriculture (4.7%). Between 2009 and 2016, manufacturing was the only sector to record negative growth (-1.9%), while all other sectors and global GDP continued to expand. From 2017 to 2023, manufacturing stagnated (0.1%), lagging behind agriculture, services, total GDP, and global growth, leading to a continued decline in its relative weight in the Brazilian economy.

Table 3
Real Growth Rate – %

GDP	Agro GDP	Manufacturing Industry GDP	Real State GDP	Services GDP	Total GDP	Global
	Brazil	Brazil	Brazil	Brazil	Brazil	GDP
2002 to 2008	4.7%	3.9%	2.7%	4.0%	4.0%	3.5%
2009 to 2016	1.8%	-1.9%	1.9%	1.6%	1.2%	2.6%
2017 to 2023	4.9%	0.1%	0.9%	1.7%	1.7%	2.6%

Figure 1 shows manufacturing GDP growth until 2008, a sharp post-crisis decline, and stagnation from 2017 to March 2024, except for the 2020 COVID-19 shock.

Figure 2 shows strong real exchange rate appreciation in the early 2000s, renewed volatility after 2015, and sharp depreciation from 2017 onward, followed by partial stabilization at elevated levels through early 2024.

Figure 2
Real Exchange Rate

Figure 3 shows strong growth in exports of basic goods and commodities from 2002, a decline after 2014, and a sharp post-2020 rebound driven by higher commodity prices, currency depreciation, and global recovery.

Figure 3
Exports of Basic Goods and Commodities – USD Millions (deflated)

Figure 4, shows a rising import penetration ratio in manufacturing since 2002, exceeding 1.0 after the 2008 crisis and reflecting growing external competition and competitiveness challenges.

Figure 4
Import Penetration Relative to Exports

Figure 5 shows that manufacturing terms of trade fluctuated since 2002 but remained mostly below 100, indicating persistently low external competitiveness.

Figure 5
Terms of Trade for the Manufacturing Industry

Empirical VECM Results

The VECM results indicate that the real exchange rate has a significant long-run effect on manufacturing industry GDP, with a beta coefficient of 0.4988. Import penetration also exerts a meaningful influence, with an estimated beta coefficient of 0.266311.

The diagnostic test results for the VECM are summarized in Table 4, confirming the model’s stability and adequacy, although normality of residuals was not observed for all variables.

Table 4
VECM Diagnostic Test Results

Test	Result	Interpretation
Stability Analysis	Stable (roots within the unit circle)	The VECM is stable and suitable for long-term analysis.
Model Residuals (ACF)	No significant autocorrelation	Residuals are random, and the model is well specified.
Portmanteau Test	p-value = 11.57% (fails to reject H0)	No autocorrelation in residuals, confirming model adequacy.
LM Test	p-value = 11.61% (fails to reject H0)	No significant serial autocorrelation, reinforcing adequacy.
Normality Test	p-value = 8.5% (Ttit is normal)	Only Ttit exhibits normality; others do not follow normal distribution.*
Heteroscedasticity Test	p-value = 1 (no heteroscedasticity)	Error variance is constant, suggesting no heteroscedasticity.

• Source: Prepared by the authors using R Studio.
• ^*
  The residuals of most equations are non-normally distributed, except for Ttit; however, this does not invalidate the VECM, as its core assumptions are satisfied. In addition, the large sample size mitigates potential effects of non-normality, and the main inference tools – impulse response functions and forecast error variance decomposition – do not rely on normality.

The diagnostic tests confirm that the VECM is stable, well-adjusted, and robust. The stability of roots and the absence of autocorrelation in residuals reinforce the reliability of the estimates obtained, while the normality and heteroscedasticity tests suggest that the model was correctly specified.

Analysis and Findings of Empirical VECM Results

The estimation of the Vector Error Correction Model (VECM) for the period 2002-2024 allowed for the identification of both long-term equilibrium relationships and short-term dynamics that are crucial for understanding the behavior of Brazil’s manufacturing industry GDP. The econometric results, combined with historical analysis, reveal a complex picture of manufacturing decline, hysteresis, and the limits of corrective policies.

The Johansen test confirmed the existence of two cointegration vectors (r = 2), indicating long-term equilibrium relationships among the variables. The first vector (ECT1), normalized on the manufacturing industry GDP (PIBit), is expressed as follows:

$\text{ECT1t}=\text{PIBitt}-0.4988·\text{TxCamRet}-0.0009·\text{ExBasComt}+0.9175·\text{Ttitt}-0.266311·\text{PimExpitt}+173.98193$

The interpretation of the β coefficients reveals important nuances. The coefficient of the real exchange rate (TxCamRe, β = 0.4988), although positive and seemingly counterintuitive, reflects a historical long-term equilibrium relationship in which phases of higher manufacturing industry activity coincided with appreciation pressures on the exchange rate, driven by commodity cycles and capital inflows. This does not imply direct causality, but rather an observed equilibrium correlation in the sample. The effect of basic and commodity exports (ExBasCom, β = -0.0009) is negligible, confirming that the re-primarization of the export basket has no significant direct relationship with long-term manufacturing activity. Conversely, improvements in the terms of trade (Ttit, β = 0.9175) are associated with higher equilibrium manufacturing industry GDP, while import penetration (PimExpit, β = -0.266311) correlates with a strong reduction, capturing the loss of domestic production space.

The short-term adjustment mechanism is governed by the equation:

$\Delta \text{PIB\_it}=-0.1353·\text{ECT1\_}\left\{\text{t}-1\right\}-0.03859·\text{ECT2\_t-1\text{}}+\dots +\epsilon \text{t,}$

where the adjustment coefficients are α₁ = 0.1353 and α₂ = 0.03859. These negative and statistically significant coefficients indicate that deviations from long-term equilibrium are corrected in the opposite direction, with ECT1-related deviations being corrected more rapidly (≈13.5% per month) and ECT2-related deviations more slowly (≈3.9% per month). Economically, ECT1 captures the long-run equilibrium between manufacturing GDP and key factors such as the real exchange rate, terms of trade, import penetration, and basic commodity exports, whereas ECT2 reflects a secondary long-term relationship, capturing other structural influences on manufacturing GDP.

The Impulse Response Function (IRF), shown in Figure 6, highlights the short-term causal dynamics: an exchange rate appreciation shock generates a negative and persistent impact on manufacturing industry GDP, in line with the New-Developmentalist Theory predicting a loss of competitiveness; an increase in import penetration also negatively affects manufacturing activity, while favorable shocks in the terms of trade of manufactured goods produce a positive short-term effect.

Figure 6
Impulse Response Function Results

Variance decomposition, shown in Table 5, quantifies the relative importance of each variable in the fluctuations of manufacturing industry GDP. In general, the manufacturing industry itself accounts for 34% of total variance, reflecting internal dynamics and structural constraints. The remaining 66% is explained by external factors related to global integration: the real exchange rate (28%), manufacturing import penetration (26%), manufacturing terms of trade (9%), and basic and commodity exports (3%).

Table 5
Variance Decomposition Results

Variable	Share in Variance (%)
Manufacturing GDP	34
Real Exchange Rate	28
Import Penetration	26
Terms of Trade	9
Commodity Exports	3

This highlights that manufacturing GDP fluctuations are largely driven by global dynamics, trade exposure, and exchange rate movements, alongside significant endogenous sectoral factors, consistent with New Developmentalism Theory and GVCs, as described before.

In the short run (10-15 months), manufacturing output still explains about 50% of its own variance, while the real exchange rate and import penetration jointly account for roughly 40%, with terms of trade and commodity exports remaining marginal. In the medium run (around 40 months), the real exchange rate (≈30%) and import penetration (≈28%) become dominant, jointly explaining more than half of the variance, while the industry’s own share declines to about 30%. In the long run (60 months), external factors clearly prevail: the real exchange rate explains about 35% of the variance, import penetration about 30%, and terms of trade about 10%, while the share of manufacturing itself falls below 20%. This trajectory reinforces the evidence that Brazil’s manufacturing industry is structurally vulnerable to exchange rate appreciation, trade asymmetries, and regressive global integration.

The depreciation observed from 2017 onward proved insufficient to reverse the stagnation of manufacturing industry GDP (0.1% annual growth). The VECM provides a robust explanation for this phenomenon, rooted in the previous period: between 2009 and 2016, while total GDP grew at an average of 1.2% per year, manufacturing industry GDP contracted at an annual rate of -1.9%.

This significantly underperforming trajectory created a large negative deviation from the long-term equilibrium (ECT << 0). Beyond a cyclical deviation, this period generated industrial hysteresis, with permanent plant closures, dismantling of production chains and technological erosion, structurally reducing the sector’s potential productive capacity.

On the other hand, the significant currency depreciation that began in 2017 was not sufficient on its own to stimulate manufacturing GDP growth thereafter. This raises the question: why? Post-2017 exchange rate depreciation acted as a positive corrective force, but its effectiveness was limited by the predominance of other negative shocks – import penetration remained high, the manufacturing industry, dependent on imported inputs, faced higher costs due to currency devaluation, while non-exceptional terms of trade and unmodeled factors, such as high interest rates and low confidence, likely suppressed investment and demand – and by the slow adjustment speed, as the α coefficient = –0.135 implies a slow correction (≈13.5% per month) in the face of a large inherited negative deviation from the 2009-2016 period.

The empirical results of the VECM reveal an apparent paradox when compared to the premises of New Developmentalism. The impulse response function (Figure 6) shows that a real exchange rate depreciation generates, in the very short run, a negative impact on manufacturing GDP (≈ -0.5), which deepens over the long run (≈ -1.5). At first glance, this outcome contradicts the theoretical expectation that real exchange rate depreciation should enhance competitiveness and stimulate manufacturing output. The explanation, however, lies in the sector’s structural conditions: a high dependence on imported inputs, which become more expensive with depreciation; the cumulative effects of industrial hysteresis, resulting from the contraction between 2009 and 2016, which permanently reduced productive capacity; and the simultaneity of adverse shocks, such as high import penetration and unfavorable terms of trade. The correction mechanism towards long-run equilibrium (ECT1), with an adjustment coefficient of 13.5% per month, further underscores the slow pace of convergence, limiting the effectiveness of depreciation when acting alone. Therefore, the empirical evidence reinforces that a depreciated real exchange rate is a necessary but not sufficient condition for the recovery of manufacturing, reinforcing the New-Developmentalist emphasis on complementary industrial, technological, and macroeconomic policies.

Although basic commodity exports have a marginal direct impact on manufacturing industry GDP, their indirect effect is crucial and operates through exchange rate appreciation. The model captures the historical relationship in which commodity cycles generated massive foreign exchange inflows and appreciation pressures on the exchange rate, creating an environment favorable to total GDP but detrimental to the competitiveness of the manufacturing industry – the essence of the Dutch disease. Thus, commodities affected the manufacturing industry not through production linkages, but indirectly by fostering exchange rate appreciation that “crowds out” manufacturing production, explaining the divergence in performance between total GDP and manufacturing industry GDP during 2009-2016 and contributing to the formation of the large negative deviation (ECT).

It can be concluded that the post-2017 stagnation is an expected outcome of the model: a positive exchange rate shock acting slowly on a pre-existing large negative deviation, amplified by hysteresis damage, while other negative shocks of equivalent magnitude in the model – particularly manufacturing import penetration – acted in the opposite direction.

CONCLUSION

This study analyzes the evolution of Brazil’s manufacturing industry GDP between 2002 and 2024 within the New-Developmentalist framework, highlighting the structural and macroeconomic determinants of its long-term performance. The VECM results indicate that manufacturing underperformance is not merely cyclical but reflects persistent structural constraints, particularly chronic real exchange rate appreciation, rising import penetration, and the indirect effects of commodity cycles, consistent with Dutch disease dynamics. The interaction of these factors over time generated hysteresis effects that weakened productive capacity, technological capabilities, and long-run growth potential.

The empirical evidence shows that exchange rate depreciation is a necessary but insufficient condition for the recovery of the manufacturing industry. Real exchange rate appreciation and import penetration emerge as the main drivers of manufacturing decline, while commodity exports affect manufacturing primarily through indirect channels by fostering currency appreciation. The identified structural breaks – commodity-driven expansion (2003-2008), post-2009 decline, and post-2017 stagnation – underscore the sector’s vulnerability to external shocks and domestic policy regimes, as well as the slow adjustment process of manufacturing activity.

These findings support the New-Developmentalist view that restoring manufacturing dynamism requires a coordinated policy framework. Beyond a competitive exchange rate, effective recovery depends on selective industrial policies to reduce import penetration, measures to stimulate investment and demand, rebuilding productive chains, promoting technological upgrading, and mitigating the indirect effects of commodity cycles through fiscal and macroeconomic mechanisms. Future research could extend this analysis by incorporating variables related to global value chain participation, sectoral heterogeneity within manufacturing, interest rates, investment, productivity, innovation, and employment, thereby deepening understanding of the interactions among macroeconomic conditions, trade dynamics, and structural factors shaping manufacturing performance.

• Notes
  Paper presented at the 7th International Congress on New Developmentalism, FGV EAESP, São Paulo, December 7, 2024.
  The authors sincerely thank the anonymous reviewers for their valuable comments and constructive suggestions, which greatly contributed to improving the clarity, rigor, and overall quality of this manuscript. The authors are also grateful to Professor Rosângela Ballini (Institute of Economics at Unicamp) for her class discussions and valuable assistance in the development of this model.

Data Availability Statement

The data used in this study were obtained from publicly available official statistical sources (IBGE, IPEA, SECEX/MDIC, and FUNCEX). The dataset was compiled and processed in RStudio using time series econometric techniques, including Vector Error Correction Models (VECM). The processed dataset and replication code are available from the corresponding author upon reasonable request.

REFERENCES

• Baldwin, R. (2016), The great convergence: information technology and the new globalization, Harvard University Press, Harvard University, Cambridge, MA.
• Banco Central do Brasil (BCB) (s.d.), Sistema Gerenciador de Séries Temporais, disponível em: https://www3.bcb.gov.br/sgspub (acesso em: 20 junho 2024).
  » https://www3.bcb.gov.br/sgspub
• Borghi, R.A.Z. (2024), “Crescimento e estagnação do Brasil: contribuições do professor Bresser-Pereira”, Cadernos Gestão Pública e Cidadania, No. 29, disponível em: https://doi.org/10.12660/cgpc.v29.90177, (acesso em: 15 setembro 2024).
  » https://doi.org/10.12660/cgpc.v29.90177
• Borghi, R.A.Z. and Sarti, F. (2019), “Domestic Demand and Foreign Trade: The Brazilian Growth Trajectory”, Grivoyannis, E. (Ed.) International Integration of the Brazilian Economy, Palgrave Macmillan, New York, NY, ch. 2, pp.25-47.
• Bresser-Pereira, L.C. (2008), “The Dutch disease and its neutralization: a Ricardian approach”, Brazilian Journal of Political Economy, Vol. 28No. 1, pp.47-71.
• Bresser-Pereira, L.C. (2016), “Teoria novo-desenvolvimentista: uma síntese”, Cadernos do Desenvolvimento, Vol. 11, No. 19, pp.145-165.
• Bresser-Pereira, L.C.; Nassif, A.; Feijó, C. (2016), “A reconstrução da indústria brasileira: a conexão entre o regime macroeconômico e a política industrial”, Revista de Economia Política, Vol. 36, No. 3, pp.493-513.
• Bueno, R.L.S. (2012) Econometria de séries temporais 2. ed. rev. e atual. Cengage Learning, São Paulo, ISBN 978-85-221-1157-2.
• Dicken, P. (2011), Global shift: mapping the changing contours of the world economy. Sage Publications, London, 6th edition.
• Feijó, C. and Araújo, E. (2024), “O papel da taxa de câmbio na teoria novo-desenvolvimentista”, Cadernos Gestão Pública e Cidadania, No. 29, disponível em: https://doi.org/10.12660/cgpc.v29.90198, (acesso em: 25 setembro 2024).
  » https://doi.org/10.12660/cgpc.v29.90198
• Fundação Getulio Vargas (FGV) (s.d.), FGVDados, disponível em: https://extraibre.fgv.br/IBRE/sitefgvdados/default.aspx, acesso em: 28 junho 2024.
  » https://extraibre.fgv.br/IBRE/sitefgvdados/default.aspx
• Gereffi, G. (1999), “International trade and industrial upgrading in the apparel commodity chain”, Journal of International Economics, Vol. 48 No. 1: pp.37-70.
• Hiratuka, C. and Sarti, F. (2017), “Transformações na estrutura produtiva global, desindustrialização e desenvolvimento industrial no Brasil”, Revista de Economia Política, Vol. 37 No. 1, pp.189-207.
• Instituto Brasileiro de Geografia e Estatística (IBGE) (s.d.) Banco de Dados Sidra, disponível em: https://sidra.ibge.gov.br/home/ipca/brasil, (acesso em: 15 junho 2024).
  » https://sidra.ibge.gov.br/home/ipca/brasil
• Instituto de Pesquisa Econômica Aplicada (IPEA) (s.d.), Ipeadata, disponível em: http://ipeadata.gov.br/Default.aspx, (acesso em: 17 junho 2024).
  » http://ipeadata.gov.br/Default.aspx
• Morceiro, P.C.; Guilhoto, J.J.M. (2023), “Sectoral de-industrialization and long-run stagnation of Brazilian manufacturing”, Revista de Economia Política, Vol. 43 No. 2, pp.418-441, disponível em: http://dx.doi.org/10.1590/0101-31572023-3340
  » http://dx.doi.org/10.1590/0101-31572023-3340
• Nassif, A.; Feijó, C.; Araújo, E. (2014), “Structural change and economic development: is Brazil catching up or falling behind?”, Cambridge Journal of Economics, Vol. 39 No. 4, pp.1307-1332.
• OECD (2013), Interconnected Economies: benefiting from global value chains, OECD Publishing, Paris.
• Palma, J.G. (2019), “Desindustrialización, desindustrialización ‘prematura’ y ‘síndrome holandês’”, El Trimestre Económico, Vol. 86 No. 4, pp.901-966.
• Sarti, F. and Laplane, M.F. (2021), “Internacionalização produtiva, competitividade e desenvolvimento”, Diegues, A.C. and Sarti, F. (Orgs.), Brasil: indústria e desenvolvimento em um cenário de transformação do paradigma tecno-produtivo, Editora CRV, Curitiba, cap. 1, pp.23-47.
• Tregenna, F. and Andreoni, A. (2020), “Deindustrialisation reconsidered: structural shifts and sectoral heterogeneity”, UCL Institute for Innovation and Public Purpose, Working Paper Series (IIPP WP 2020-06).
• UNCTAD (2018), Trade and Development Report 2018 – Power, Platforms and the Free Trade Delusion, United Nations, New York and Geneva.: United Nations.

• Editor responsible for the evaluation process:
  Luiz Carlos Bresser-Pereira

Publication Dates

• Publication in this collection
  10 July 2026
• Date of issue
  2026

History

• Received
  09 Jan 2025
• Accepted
  09 Dec 2025