Democracy, economic growth and volatility in Brazil: what is the causality? ¹

Democracia, crescimento econômico e volatilidade no Brasil: qual a causalidade?

Abstract:

The impact of democracy on economic growth has been studied a great deal in recent years. This study is intended to add to this empirical literature by analyzing its relationship specifically in Brazil, by means of time series cointegration techniques. With a long time series, that goes from 1900 to 2022, we identified a robust positive, causal relationship between democracy and growth using a new Brazilian GDP database. In addition, a negative causal relationship was identified between democracy and growth volatility, but with results that were not as sturdy as those in the previous relationship.

Keywords:
democracy; growth; cointegration; volatility

Resumo:

O impacto da democracia no crescimento econômico tem sido muito estudado nos últimos anos. Este estudo pretende agregar a essa literatura empírica, analisando sua relação especificamente no Brasil através de técnicas de séries de tempo por cointegração. Com séries longas, de 1900 a 2022, identificamos uma relação positiva, causal e robusta entre democracia e crescimento, utilizando uma nova base de dados de PIB brasileiro. Além disso, também se identificou uma relação causal e negativa entre democracia e volatilidade de crescimento, mas com resultados menos robustos que o anterior.

Palavras-chave:
democracia; crescimento; cointegração; volatilidade

Introduction

The impact of democracy on economic growth has been one of the most heavily studied topics over the last few decades. A number of studies have been published that are aimed at trying to identify the relationship between these two variables, but that also include an assessment of causality, in other words, whether democracy causes economic growth or the other way around.

An initial assessment may indicate that this relationship is a two-way street. If it is true that democracy can cause growth, the opposite/reverse could also make sense. Societies in which democratic regimes develop could result in more stable growth, a decrease in income inequality on account of the voting power given to the population and greater social spending on health and education, which has a positive effect on growth over the long term. More authoritarian regimes would lead to increased growth volatility, have less room for social spending due to the fact that it would have a negative impact on military and defense spending, and would not attract risky investments that could generate growth on account of the more visible presence of the state in these regimes.

On the other hand, increased continuous growth in a country could generate greater social demands from the population in the future, which would be better met under democratic regimes than under authoritarian ones. A higher level of wealth would generate greater demand on the part of the elites for freedom, particularly from a business point of view, and this could result in a heightened level of demand for democratic freedom. There are examples of countries such as South Korea and Singapore that were initially more authoritarian but which evolved over time into fuller democracies.

But what is the causal relationship? Various studies have shown that the causality seems to flow from democracy to growth. In a recent study, Acemoglu et al. (2019) point out that the democratization of a country can result in an increase in GDP per capita growth of up to 20% over the long term (Acemoglu et al., 2019). This study has become a benchmark due to its robust econometrics, in which various methods and data were used to examine this relationship.

However, the purpose of this article is less ambitious. It retains the aim of corroborating the work of Acemoglu and his colleagues, but analyzes the Brazilian economy through a longer time series than those used by the authors, who took into account a panel of 175 countries between 1960 and 2010. For this article, we will use the V-Dem database (Coppedge et al., 2023) from 1900 to 2022 and the GDP data calculated by Haddad (1978) and by the IBGE - The Brazilian Institute of Geography and Statistics (2023a). Furthermore, we will make use of a new economic growth database developed by Bacha et al. (2023a, 2023b). which demonstrates, in a convincing way, that Brazilian growth in the 20^th century was lower than is usually taken into account. What impact would this lower GDP growth have on the relationship between democracy and growth?

Moreover, our study will also assess the impact of democracy on the volatility of growth. Less volatile growth makes an economy more predictable and, consequently, increases the potential for growth. Therefore, the question is whether democracy not only increases economic growth, but also reduces growth volatility. The answer to both questions is a resounding yes, in other words, democracy resulted in greater growth in Brazil and lower volatility of growth over the course of the 20^th century and during the early part of the 21^st century. This result alone is significant in that it reinforces the idea that anti-democratic events have important welfare consequences for society as a whole over time.

The next section gives a brief review of the literature, after which the data is presented in order for the estimates to be discussed in the following section. Finally, we present our conclusions.

A brief review of the literature

The economic and political literature has made progress in identifying the causal relationships between democracy and growth. Although some early econometric articles identified a weak causality (Barro, 1996, Tavares; Wacziarg, 2001), in addition to the classic study by Lipset (1959), which placed the causality of growth on democracy, more recent studies have bolstered the positive relationship between democracy and economic growth (Acemoglu; Johnson; Robinson, 2001, 2005; Acemoglu et al., 2019; Gründler; Krieger, 2016; Persson; Tabellini, 2006; Rodrik; Subramanian; Trebbi, 2004). As Zakaria (2024, p. 120) reminds us, “economic advances do not magically produce political advances”.

In fact, the results suggest that democracy enables an accumulation of human capital, which generates an increase in productivity and, consequently, in growth, which can also help reduce income inequality. Furthermore, the demand for social spending increases and, therefore, one of the findings of these studies is that democracy increases public spending, particularly on education and health (Tavares; Wacziarg, 2001), but leads to less physical capital. This would follow from a traditional economics result that greater public spending leads to higher interest rates, with a crowding out effect on investments. So, in theory, democracy would lead to more human capital and less physical capital and in general those studies that have identified a negative relationship between democracy and growth encourage us to believe that this negative result for physical capital outweighs the positive impact of human capital. To a certain extent, this was also the idea of Huntington (1968), who was of the opinion that the increased pursuit on the part of society for spending on education and health would cause a decrease in private returns for companies, thus slowing down the pace of growth.

Another interesting study is that of Persson and Tabellini (2009). The research finds that “democratic capital” - measured by a country’s own democratic history and the presence of democracy in neighboring nations - plays a crucial role in maintaining democratic systems while also encouraging transitions from autocratic regimes. The study reveals that higher levels of democratic capital contribute to economic growth by creating more stable democratic institutions. The researchers observe that different countries respond variably to democracy, leading to a natural sorting of nations into distinct political systems, which explains observed differences between democratic and autocratic states. The findings point to a potential positive feedback loop where the growth of both physical and democratic capital strengthens each other, simultaneously fostering economic development and democratic stability.

Even though the most recent results corroborate the causal relationship between democracy and growth, this conclusion is not yet fully consolidated in the literature. Doucouliagos and Ulubaşoğlu (2008) undertook a meta-analysis of 483 articles that had been published up to that point and robustly found that there is no direct relationship between democracy and economic growth, but rather a solid, indirect relationship between democracy and human capital, low inflation, low political instability and high levels of economic freedom. However, it should be pointed out that these four components have a significant impact on economic growth and even if it is not direct, it is clear that democracy builds up gains that lead to greater growth in the future. In this sense, perhaps the recent article by Acemoglu et al. (2019) will become the analysis reference on the subject, on account of its exhaustive technical analysis of this issue, driving home the key fact: namely, that the short-term impact of democracy on growth is likely to be unclear, but the long-term impact is increasingly consolidated. In fact, in a more recent meta-analysis, Colagrossi, Rossignoli and Maggioni (2020) analyzed 188 studies from the last 36 years and identified a positive relationship between democracy and growth, especially in the more recent studies that have been carried out since Doucouliagos and Ulubaşoğlu (2008), which is connected to the more modern techniques applied, but also with the disclaimer that the results are not necessarily homogeneous in spatial and temporal terms, which reinforces the need for studies like our one.

At the very least, it has become clearer that the opposite political system, namely that of authoritarianism, cannot coexist with long periods of growth. On the contrary, very few authoritarian regimes have managed to keep up sustained growth, and it is a commonplace occurrence that the economic crises produced by these regimes lead to their demise (Levitsky; Way, 2022). But in advanced, historically more democratic countries, the consensus in relation to the importance of democracy’s impact on growth is very strong (Soskice; Iversen, 2019). This point reinforces the argument put forward by Przeworski et al. (1997), that democracies would consolidate above a certain per capita income threshold (in the case of this study, US$6,000 per year) and with it being more likely that economic development does not generate democracy in a later study by the same authors (Przeworski et al., 2000).

Other more localized studies in relation to Latin America have also corroborated this positive relationship, with some nuances and contrary results. Sloan and Tedin (1987) analyzed 20 Latin American countries between 1960 and 1979 and identified authoritarian regimes as being more growth stimulating. This old theory of modernity, in which dictatorships would be prone to economic growth and, consequently, be able to generate democracy in the future, is not backed up by results published later, such as in the case of the aforementioned Przeworski et al. (1997), and Przeworski and Limongi (1997).

Grier and Tullock (1989), on the other hand, obtained different results with Latin American countries with democracy showing greater economic expansion. Remmer (1990) also corroborates these results, although with fewer countries and over a shorter period of time. Meanwhile Fittipaldi et al. (2017) pointed out that, for Latin America, institutional stability is of greater importance than democracy for economic growth, whether it is democratic or not. In other words, the fewer regime changes in Latin America, the more economic growth would benefit.

Time series cointegration techniques have provided particularly valuable insights by addressing the long-run relationship between democratic institutions and economic performance while accounting for their complex temporal dynamics. At the local level, using cointegration techniques which, as is well known, can be used for analyzing a country, in this case we identified some studies that corroborate the causal relationship between democracy and growth. Qasim (2021) identified a sturdy relationship between these variables for Pakistan. Similarly, Narayan and Smyth (2006) identified a positive relationship between these variables for China, with the additional point that human and physical capital also generate growth. Sakyi and Adams (2012) identified the same long-term relationship for Ghana, taking into account the positive effect of public spending. Narayan, Narayan and Smyth (2011) performed a number of cointegration tests for sub-Saharan Africa and identified different results, with some countries showing a causal relationship between democracy and growth while for others the result is the opposite, without a theoretical discussion of differences in the results for a set of countries with certain similarities.

Other recent studies utilizing the comprehensive democracy measures from PolityV, Freedom House, and V-Dem have shed new light on this relationship. These indices capture different aspects of democratic governance: PolityV focuses on institutional characteristics; Freedom House emphasizes political rights and civil liberties; while V-Dem provides highly disaggregated data on multiple democracy dimensions. This measurement diversity has strengthened our understanding of democracy’s multifaceted effects on growth.

The aforementioned article by Acemoglu et al. (2019) conducted groundbreaking research using cointegration techniques with PolityV data across 175 countries from 1960-2010. Their findings reveal a robust positive long-run relationship between democracy and GDP per capita, estimating that democratization increases GDP per capita by about 20% over the long run. The cointegration framework was crucial in distinguishing between short-term adjustment effects and the permanent impact of democratic transitions. Their error correction models showed that while the immediate growth effects of democratization may be modest, the cumulative impact builds substantially over time through institutional improvements and investment in human capital.

A reverse relationship between growth and democracy can be found in Slater and Wong (2022). These authors studied 12 countries in Asia and identified clusters in which the development process eventually led to democracy, but only in those where two conditions were present: confidence in victory and confidence in stability, as they call it. The idea is that the transition to democracy only took place in those countries where the authoritarian regime felt confident that its political group would remain in power and with stability. Where this was not the case, democracy failed to take hold. In Japan, South Korea and Taiwan, democracy has been maintained on account of these two factors. Indonesia, Thailand and Myanmar failed to make a stable transition because they were unable to maintain the authoritarian regime’s structures within the structures of democratic power.

The Brazilian case is certainly different, and perhaps on account of the degree: when the authoritarian regime was aggressive and posted poor economic results, such as during the 1964-85 period, the authoritarian group did not remain in the circle of power. But when Vargas left power, his presence remained strong, whether due to the fact that his former defense minister became president, or because he became president again in 1951. Not to mention Vargas’ presence in the PTB (Brazilian Labor Party) during the democratic period between 1945 and 1964. The Brazilian case is even more tenuous, as it involves a very strong construction of state capitalism starting from the time of the Vargas regime, which has not been lost throughout history. The continued strong state presence at the time of the democratic transition in 1985 was unable to eliminate rent seeking (Lisboa; Latif, 2013), which can still be seen today, to a certain extent, in the presence of state-run companies in which the public/private sector relationship is, at the very least, blurred (Musacchio; Lazzarini, 2015).

The difference between these studies and ours is that the period of data used here is much longer than in the traditional studies, which makes the results the results more reliable.

But looking at the effect of democracy as a stock whose effects accumulate over time (Gerring et al., 2005), the overall impact could be positive, as identified by Acemoglu et al. (2019). One possibility of this long-term effect is that not only does human capital grow, but economic stability also increases, in other words, growth volatility decreases. Less variance in growth is an important factor for companies to invest, as it enables a more predictable growth horizon, which allows them to increase their investments. The more volatile the economy, the smaller the increase in investment. For this reason, the volatility channel may be important in signaling that even though physical capital may decrease with the growth in human capital, as a result of the increase in public spending, the reduction in volatility may itself become an attraction for companies to invest. Some studies have indicated that the volatility of growth decreases with democracy (Klomp; De Haan, 2009; Mobarak, 2003). One reason for this effect is that democratic countries have fewer barriers to entry for companies and this industrial diversity reduces the volatility of growth over time (Cuberes; Jerzmanowski, 2009).

Database and methodology

This study is not intended to add another panel study to the already wide range of studies on the topic using this method. The aim here is to do something more specific, namely to identify this causal relationship for Brazil, using very long and recently revised time series.

The economic series usually used for Brazil in these studies start in the 1950s at the latest. Bearing in mind that the democratic process is a lengthy one, this study opted for a longer period, from 1900 to 2022. In this case, two per capita GDP growth variables will be used. Firstly, the data compiled by Haddad (1978), from 1900 to 1947 and thereafter by the FGV (Getulio Vargas Foundation) between 1947 and 1980 and from 1980 onwards by the IBGE (IBGE, 2023a). This is the traditional GDP per capita series used to study growth in Brazil.

In addition, we incorporated the recent revision of population growth made by the IBGE (IBGE, 2023b), which is important due to the significant revision made to population growth between 2010 and 2022, with a decrease of roughly 10 million people vis-à-vis the known estimates. As a result, the GDP per capita for this period will be higher than is usually taken into account in the estimates made prior to the Census.

But a second assessment will be made with the data that was recently constructed by Bacha et al. (2023a, 2023b) which changes the view regarding Brazilian growth over the last two centuries. According to the authors, Brazil grew more in the 19^th century than has previously been estimated, confirming Furtados’s (1968) view that the country had grown more than was stated during that period. At the same time, Brazil’s GDP growth in the 20^th century was overestimated due to the fact that sectors that usually have lower growth, particularly services, were left out. Incorporating the sector’s growth over the last century, Bacha et al. (2023a, 2023b) and Bacha (2023) came up with surprising results, such as the fact that the Brazilian economy’s average growth during the democratic period between 1945 and 1963, which was one of 7.1%, was actually higher than that observed during the dictatorship, from 1964 to 1984, when the comparable figure was one of 6.3% (Bacha, 2023). The differences are significant when looking at the period as a whole: traditional GDP per capita grew by 1,723% in this period, while adjusted GDP grew by “a mere” 905% (Graph 1). It is important to reiterate here that the annual data was provided by one of the authors via email, Guilherme Tombolo.

It is worth detailing the authors’ methodology and results. The paper presents a critical reanalysis of Brazil’s GDP growth estimates from 1900 to 1980, challenging the widely accepted narrative of extremely slow growth in the 19th century followed by remarkable growth until 1980. The authors argue that these dramatic structural breaks are partially statistical illusions, particularly focusing on how the exclusion of slower-growing service sectors from GDP calculations led to overestimation of growth rates.

The methodology analysis reveals significant issues with how Brazil’s national accounts were calculated during different periods. From 1947 to 1985, the Getúlio Vargas Foundation (FGV) was responsible for these calculations, followed by IBGE (Brazilian Institute of Geography and Statistics) from 1986 onwards. The authors identify a crucial methodological change in 1969 when FGV introduced a new series that excluded government, financial intermediaries, and rentals from real output calculations due to “conceptual problems and unavailability of statistical sources.” This exclusion is particularly problematic as these sectors typically grew more slowly than the rest of the economy.

The technical analysis of this methodological shift is compelling. The authors demonstrate that the exclusion of slower-growing sectors artificially inflated growth rates. For instance, in the 1947-1966 period, the “old” methodology (including all sectors) showed a 5.3% annual growth rate, while the “new” methodology (excluding certain services) yielded 6.1%. This 0.8 percentage point difference highlights how selective sector inclusion can significantly distort growth figures.

A critical weakness in the national accounts methodology becomes apparent in how service sector outputs were measured. Initially, service sector output was calculated based on the number of employees, but later methods involved deflating nominal income by cost-of-living indices, leading to questionable results. The authors note that this change caused services to show surprisingly higher growth rates than aggregate output, contrary to previous patterns and economic logic.

The paper’s mathematical approach to correcting these biases is particularly noteworthy. Using a weighted average formula (Yo = 0.7 * Yn + 0.3 * S), where Yo represents growth under the old methodology, Yn under the new methodology, and S the excluded services growth rate, the authors provide a more realistic picture of Brazil’s economic growth. This recalculation suggests that the famous “Brazilian economic miracle” may have been less miraculous than previously thought, with growth rates in the 1966-1980 period being closer to 7.1% annually rather than the reported 8.6%.

The analysis of the 1900-1947 period reveals additional methodological concerns. Haddad’s (1978) widely accepted estimates for this period suffer from similar selection biases, potentially overestimating growth by excluding informal and handicraft activities. The authors’ adjustment of these figures, accounting for excluded service sectors, suggests a more modest growth rate of 4% compared to Haddad’s 4.4%.

A particularly compelling criticism concerns the double-counting of intermediate goods in the national accounts. The practice of computing real output by aggregating production volumes rather than real value added led to overestimation, especially during periods of rapid industrialization when intermediate inputs grew faster than final outputs. The authors cite a striking example from the chemical industry, where production volume jumped 134% from 1954 to 1955, while real value added increased by only 13%.

The paper’s findings have significant implications for understanding Brazil’s economic history. The revised estimates suggest a more gradual and less dramatic growth pattern than previously accepted, challenging the narrative of extreme structural breaks in the country’s economic development. This reanalysis demonstrates how methodological choices in national accounting can significantly affect our understanding of historical economic performance. Considering the novelty of the published results, we believe it is important to make the estimates in this article considering both methodologies.

Graph 1.
Brazil: traditional and adjusted GDP (1900=100)

In the case of democracy indicators, we will use both V-Dem (Coppedge et al., 2023), which goes from 1900 to 2022, and PolityV (Marshall; Gurr, 2020), which goes from 1900 to 2018. The data is similar for the Brazilian case (Graph 2), with some marked differences, particularly during the democratic period between 1945 and 1963, in which PolityV regards the period as being one of high democracy in the country, while indicating that it considers the last ten years of the dictatorship as being similar to the period of the Old Republic. V-Dem tends to be more conservative and only shows democratic progress in the country in the wake of re-democratization, with a drop in recent years, especially during the Bolsonaro period, during which PolityV did not identify any change. To a certain extent, one could say that the V-Dem data is more cautious regarding the Brazilian political regime than PolityV.

Despite the recent criticisms of these methodologies, especially because they indicate a more pronounced democratic backsliding in recent years (Little; Meng, 2023), these longer-term figures are virtually the only ones that exist, which allow the analysis required in this article. All of the variables used here will be in logarithm in order to be able to identify the elasticities. It should be pointed out that we will only analyze the two variables of GDP and democracy because there are few long series of more than 100 years and cointegration demands time series of greater length in order to obtain more reliable results.

Graph 2.
Brazil: V-Dem and PolityV indices

In order to estimate the relationship between democracy and growth, we opted to use time series via the cointegration technique (Johansen, 1988; Johansen; Juselius, 1990). This method enables us to identify long-term relationships when there is non-stationarity in the series and also allows us to identify exogeneity. In our case, it is important to identify the causal relationship and specific econometric tests will be carried out for the purpose of identifying the aforesaid relationship.

We will use cointegration in this case because the variables identified are non-stationary, as we will show below, in other words, they are considered a Random Walk in which the variable in t is equal to the variable in t-1 plus an error term. Non-stationary variables regressed against each other can result in what is referred to as spurious regression, i.e. variables that may have a similar trend, but because they are non-stationary end up distorting the R2 and the t-statistic.

As Kennedy (2008, p. 302) explains, “a nonstationary variable (I(1)) tends to wander extensively (that is what makes it nonstationary), but some pairs of nonstationary variables can be expected to wander in such a way that they do not drift too far apart, thanks to disequilibrium forces that tend to keep them together. Some examples are short- and long-term interest rates, prices and wages, household income and expenditures, imports and exports, spot and future prices of a commodity, and exchange rates determined in different markets. Such variables are said to be cointegrated: although individually they are I(1), a particular linear combination of them is I(0)”.

Under Johansen’s (1988) and Johansen’s and Juselius’ (1990) method, the variables are initially considered endogenous and a VAR (Vector Autoregression) model is constructed for the initial econometric estimate. We will test for the existence of cointegration using the trace and maximum likelihood tests along with specific tests to pinpoint the best cointegration model. Based on this we will build the vector error correction model, which will enable us to identify the elasticity between the variables, and the LR test, which will allow us to estimate which variable is exogenous in the model.

To estimate growth volatility, we will use the GARCH methodology (Bollerslev, 1986), which is simply a joint estimate of the mean and variance of a series, with the variance being precisely the desired volatility. The GARCH model allows for autoregressive and moving average components in the heteroscedastic variance, making it a more complete model than the traditional ARCH model, which only takes into account autoregressive components. With this model it is possible to econometrically estimate the volatility of economic growth in order to estimate democracy’s impact on economic volatility.

As expected, the volatility of adjusted GDP per capita is lower than the traditional metric (Graph 3) and there is an increase in the volatility of growth between the restoration of democracy in 1945, but which accelerated throughout the period of the military dictatorship. Volatility is uncertainty and the figures show that the military period increased this degree of economic uncertainty while the long democratic period at the start of the century that includes the Vargas period was one of decreasing uncertainty. Despite the fact that Vargas was also a dictator for a while, economic growth in this period was more stable, with the only clearer recessionary period being in 1930, while the military period faced more recessions and external instabilities, such as the oil crises. Volatility rose again until 2015, when there was a marked recession, since when it has been decreasing.

Periods of greater volatility are normally associated with lower growth, particularly if this volatility is caused by growth resulting from increased public spending (Norrbin; Yigit, 2005; Ramey; Ramey, 1995). To some extent, the growth recorded from the 1940s to the 1980s and during the Lula/Dilma period can be characterized by periods of a marked lack of control over public spending. This may have played a role in increasing economic volatility during the period, which was the result found in an analysis of data from a number of countries (Carmignani; Colombo; Tirelli, 2007). GDP volatility exhibits a similar pattern to the volatility of GDP per capita (Graph 4).

Graph 3.
Volatility of GDP per capita and adjusted GDP per capita

Graph 4.
Volatility of GDP and adjusted GDP

Results

The Dickey-Fuller tests of unit roots by GLS (DF-GLS (Elliott; Rothenberg; Stock, 1996) identify non-stationarity of the series, which justifies not using OLS and instead opting for cointegration⁴ (Table 1).

Table 1.
DF-GLS unit root tests

V-Dem	-1.29***
Polity5	-2.28*
GDP	1.63***
GDP Adjusted	1.77***
GDP PC	2.40***
GDP PC Adjusted	2.51***
GDP volatility	-0.34***
GDP Adjusted volatility	-0.66***
GDP PC volatility	-0.01***
GDP PC Adjusted volatility	-0.44***

• Source: author´s estimation. Note: ***p<0.001, **p<0.05, *p<0.1. Test based on Elliot et al. (1996).

A first assessment will be to identify a weak Granger causality relationship (Granger, 1969), which can be considered more as a test of whether variable x can be used to project variable y rather than a causality test in the strict sense. Table 2 shows the results in accordance with the idea that democracy comes before GDP. In fact, the first two tests for both V-Dem and PolityV corroborate the hypothesis that democracy comes before GDP. In the case of adjusted GDP, the result is corroborated in the case of PolityV, but not in the case of V-Dem. In the case of both traditional and adjusted GDP per capita, PolityV predicts this variable, while the results are dubious in the case of V-Dem. Either way, there is a sign that for both GDP and GDP per capita there is some impact from democracy that helps to predict GDP.

Table 2.
Granger Causality Test

Null Hypothesis:	Obs	F-Statistic	Prob.
GDP does not Granger Cause V-Dem	120	1.71	0.19
V-Dem does not Granger Cause GDP		3.18	0.05
GDP does not Granger Cause Polity5	117	1.32	0.27
Polity5 does not Granger Cause GDP		3.39	0.04
GDP Adjusted does not Granger Cause V-Dem	120	2.76	0.07
V-Dem does not Granger Cause GDP Adjusted		2.32	0.10
GDP Adjusted does not Granger Cause Polity5	118	0.99	0.38
Polity5 does not Granger Cause GDP Adjusted		3.07	0.05
GDP PC does not Granger Cause V-Dem	120	2.75	0.07
V-Dem does not Granger Cause GDP PC		2.09	0.13
GDP PC Adjusted does not Granger Cause V-Dem	120	2.85	0.06
V-Dem does not Granger Cause GDP PC Adjusted		1.54	0.22
GDP PC does not Granger Cause Polity5	118	1.08	0.34
Polty5 does not Granger Cause GDP PC		3.57	0.03
GDP PC Adjusted does not Granger Cause Polity5	118	1.10	0.34
Polity5 does not Granger Cause GDP PC Adjusted		2.63	0.08

• Source: author´s estimation.

The cointegration results between GDP and GDP per capita in the two estimates and the V-Dem corroborate the hypothesis that more democracy gives rise to growth (table 3). The elasticities are lower in the case of GDP per capita, as expected, and adjusted GDP per capita is also lower than traditional GDP, which is consistent with the fact that adjusted growth is lower than traditional growth. An increase of 1% in the democracy indicator would lead to a 1.46% long-term increase in GDP per capita in the case of adjusted GDP. The LR test of exogeneity indicates at 1% that the democracy variable is exogenous in columns 2 to 4, while the GDP variables appear more consistently as endogenous, corroborating much of the literature mentioned in the literature review.

Table 3.
Cointegration results with V-Dem

	(1)	(2)	(3)	(4)
	GDP	GDP Adjusted	GDP Per Capita	GDP Per Capita Adjusted
V-Dem	3.629***	7.259***	1.886***	1.465***
	0.748	2.022	0.522	0.484
Constant	7.670***		6.813***	6.026***
	1.282		0.894	0.848
Dummies	1932, 1947, 1965, 1966	1932, 1947, 1965, 1966	1932, 1947, 1965, 1966	1932, 1947, 1965, 1966
Assumptions	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant	Deterministic assumptions: Case 1: No deterministic terms.	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant
LR Exogeneity Test	GDP 10.45025 0.001226***	GDP Adjusted 24.03689 0.000001***	GDP Per Capita 11.54413 0.00068***	GDP Per Capita Adjusted 11.03553 0.000894***
	V-Dem 9.152059 0.002484***	V-Dem 1.101841 0.293862	V-Dem 5.169589 0.022986**	V-Dem 3.891691 0.048526**
Obs	117	119	119	119

• Source: author´s estimation. Note: VAR identified through lag-lenght criteria by Hannan-Quinn and Schwartz. VAR Residual Portmanteau Tests for Autocorrelations and VAR Residual Serial Correlation LM Tests do not identify autocorrelation. Normality tests do not identify the democracy variables as normal, but Osterwald-Lenum critical values are used. Trace tests and Maximum Eigenvalue tests identify 1 cointegration relation for all the equations. LR Exogeneity Test: first numer is the Chi Square result and the second is the probability *** p < 0.001, ** p < 0.05, * p < 0.1

The results using Polity5 as the alternative democracy variable present similar effects, with the elasticity of the adjusted GDP per capita being lower than the traditional one and indicating that a 1% increase in the democracy indicator would lead to a 1.5% long-term increase in GDP per capita. Here, in an even stronger way, the LR exogeneity tests confirm that the democracy variable is exogenous to GDP in all specifications.

Table 4.
Cointegration results with Polity5

	(1)	(2)	(3)	(4)
	GDP	GDP Adjusted	GDP Per Capita	GDP Per Capita Adjusted
Polity5	1.446***	1.265***	1.735***	1.566***
	0.380	0.368	0.266	0.265
Dummies	1966, 1975, 1946, 1986, 1965, 1947, 1932, 1935, 1933	1966, 1975, 1946, 1986, 1965, 1947, 1932, 1935, 1933	1966, 1975, 1946, 1986, 1965, 1947, 1932, 1935, 1933	1966, 1975, 1946, 1986, 1965, 1947, 1932, 1935, 1933
Assumptions	Deterministic assumptions: Case 1: No deterministic terms.	Deterministic assumptions: Case 1: No deterministic terms.	Deterministic assumptions: Case 1: No deterministic terms.	Deterministic assumptions: Case 1: No deterministic terms.
LR Exogeneity Test	GDP 24.12095 0.000001***	GDP Adjusted 24.11217 0.000001***	GDP Per Capita 18.27499 0.000019***	GDP Per Capita Adjusted 18.17889 0.00002***
	Polity5 2.097088 0.14758	Polity5 0.035266 0.851039	Polity5 0.033226 0.855364	Polity5 0.082349 0.774139
Obs	117	118	118	118

• Source: author´s estimation. Note: VAR identified through lag-lenght criteria by Hannan-Quinn and Schwartz. VAR Residual Portmanteau Tests for Autocorrelations and VAR Residual Serial Correlation LM Tests do not identify autocorrelation. Normality tests do not identify the democracy variables as normal, but Osterwald-Lenum critical values are used. Trace tests and Maximum Eigenvalue tests identify 1 cointegration relation for all the equations. *** p < 0.001, ** p < 0.05, * p < 0.1

The results between growth volatility and democracy are less conclusive (table 5). Overall, we corroborate the idea that more democracy reduces the volatility of adjusted GDP per capita, in particular (column 4 of Table 5). This is the most interesting result to watch, as it reinforces the idea that the relationship is as expected in the case of the revised GDP, while the result is the opposite in the case of traditional GDP (column 3 of Table 5), with greater volatility reducing democracy over time. Since it is assumed that the revised GDP data is the most correct one and should be used from now on, we reaffirm the expected result.

Table 5.
Results of Cointegration with V-Dem and volatility

	(1)	(2)	(3)	(4)
	GDP volatility	GDP Adjusted volatility	GDP Per Capita volatility	GDP Per Capita Adjusted volatility
V-Dem	-0.434***	0.155	-0.979***	-0.432***
	0.107	0.132	0.145	0.164
Constant	7.071***	6.577***	7.908***	7.316***
	0.186	0.235	0.254	0.286
Dummies	1947, 1965, 1966, 1932, 1938, 1986, 1989, 1991, 1992, 1983	1947, 1965, 1966, 1932, 1938, 1986, 1989, 1991, 1992, 1983, 2021	1947, 1965, 1966, 1932, 1938, 1986, 1989, 1991, 1982	1947, 1965, 1966, 1932, 1938, 1986, 1989, 1991, 1982, 1937, 1943, 2021
Assumptions	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant	Deterministic assumptions: Case 2: Cointegrating relationship includes a constant
LR Exogeneity Test	GDP 3.978 0.046**	GDP Adjusted 9.062 0.002***	GDP Per Capita 0.907 0.754	GDP Per Capita Adjusted 3.270 0.070*
	V-Dem 4.676 0.030**	V-Dem 0.392 0.53	V-Dem 11.53 0.00006***	V-Dem 1.427 0.232
Obs	116	118	118	118

• Source: author´s estimation. Note: VAR identified through lag-lenght criteria by Hannan-Quinn and Schwartz. VAR Residual Portmanteau Tests for Autocorrelations and VAR Residual Serial Correlation LM Tests do not identify autocorrelation. Normality tests do not identify the democracy variables as normal, but Osterwald-Lenum critical values are used. Trace tests and Maximum Eigenvalue tests identify 1 cointegration relation for all the equations. LR Exogeneity Test: first numer is the Chi Square result and the second is the probability *** p < 0.001, ** p < 0.05, * p < 0.1

The implications of these estimates are very clear. In order for the country to be able to keep up and create more growth, an essential, but obviously not sufficient condition, is to sustain its democracy. The recent attacks on democracy by the Bolsonaro administration, which culminated in the attempted coup on January 8, 2023, are examples of what to avoid if the country really wants to maintain a minimum level of growth over the next few years. The solidity of institutions is the basis of stability on the horizon for companies so that they can invest and promote growth.

The latest V-Dem report (Nord et al., 2024) reveals progress in terms of the quality of Brazilian democracy in 2023, in the wake of the unsuccessful attacks on democracy. Last year Brazil was the country that registered the greatest positive growth in the institute’s democracy indicator, out of only 9 countries that demonstrated a positive evolution.

But the report is negative in relation to the general state of global democracy, with signs of a marginal deterioration in recent years together with additional risks that could result from a possible Trump victory in the US election. Taking into account the literature discussed here, world growth could feel the consequences of this deterioration, just as Brazil has over the course of its history.

Conclusion

The question of democracy’s effect on growth is not yet fully consolidated in the literature, despite the fact that the most recent studies point to a positive causal relationship between democracy and growth. This study corroborates this view by carrying out a time series analysis for Brazil. The progress we made in this case was due to the fact that we used a very long period of data, beginning in 1900 and ending in 2022. In addition, a new Brazilian GDP database has recently been developed and merited our attention in relation to identifying whether or not there is any difference vis-à-vis the official figures already known. Furthermore, we also identified the impact of growth volatility and democracy, which corroborated in a fairly robust way that more democracy can lead to less growth volatility, although in this case the results will require further study in the future.

The results identified here give an indication of the importance of democracy to increase Brazil’s GDP per capita. To some extent, the recent study by Bacha (2023) gives an example of this, when it shows that Brazil’s GDP grew more during the democratic period between 1946 and 1963, than it did in the dictatorship period between 1964 and 1985. This new GDP data reveals that the dictatorship delivered less growth than imagined in relation to the immediately preceding democratic period, and the question we are left with is whether or not the trend that began in 1946 would have continued. Based on the results presented here it is our assumption that this could well be the case. If Brazil had maintained its democracy and become a liberal democracy, GDP per capita would have been much higher. Assuming that our V-Dem index was 0.9, which would suggest that we were a liberal democracy, instead of the 0.523 figure that the country currently exhibits, Brazil’s GDP per capita growth between 1900 and 1922 could have been 108% higher than it actually was, which is a significant difference.

References

• ACEMOGLU, Daron; JOHNSON, Simon; ROBINSON, James A. The colonial origins of comparative development: an empirical investigation. American Economic Review, v. 91, n. 5, p. 1369-1401, 2001.
• ACEMOGLU, Daron; JOHNSON, Simon; ROBINSON, James. Institutions as a fundamental cause of long-run growth. Handbook of Economic Growth, v. 1A, p. 385-472, 2005.
• ACEMOGLU, Daron; NAIDU, Suresh; RESTREPO, Pascual; ROBINSON, James A. Democracy does cause growth. Journal of Political Economy, v. 127, n. 1, p. 47-100, 2019.
• BACHA, Edmar. Democracia e economia. Revista Brasileira, Rio de Janeiro, fase X, ano II, n. 114. p. 37-43, 2023. Available at: Available at: https://www.academia.org.br/sites/default/files/publicacoes/arquivos/revista_brasileira_114_internet.pdf Accessed on: December 1, 2024.
  » https://www.academia.org.br/sites/default/files/publicacoes/arquivos/revista_brasileira_114_internet.pdf
• BACHA, Edmar; TOMBOLO, Guilherme; VERSIANI, Flávio R. Reestimating Brazil’s GDP growth from 1900 to 1980. Revista Brasileira de Economia, Rio de Janeiro, v. 77, p. 1-13, 2023a. [e132023].
• BACHA, Edmar; TOMBOLO, Guilherme; VERSIANI, Flávio R. Secular stagnation? A new view on Brazil’s growth in the 19th century. Rio de Janeiro, Texto para discussão IEPE/ Casa das Garças n. 74, 2023b. Available at: Available at: https://iepecdg.com.br/wp-content/uploads/2022/11/20230109Revisiting19centuryBrazilGDPpc.pdf Accessed on: December 1, 2024.
  » https://iepecdg.com.br/wp-content/uploads/2022/11/20230109Revisiting19centuryBrazilGDPpc.pdf
• BARRO, Robert. Democracy and growth. Journal of Economic Growth, v. 1, n. 1, p. 1-27, 1996. Available at: Available at: https://www.jstor.org/stable/40215879 Accessed on: December 1, 2024.
  » https://www.jstor.org/stable/40215879
• BOLLERSLEV, Tim. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, v. 31, n. 3, p. 307-327, 1986.
• CARMIGNANI, Fabrizio; COLOMBO, Emilio; TIRELLI, Patrizio. Public expenditure and growth volatility: do “globalisation” and institutions matter? Working Papers, n. 116, University of Milano-Bicocca, Department of Economics, 2007.
• COLAGROSSI, Marco; ROSSIGNOLI, Domenico; MAGGIONI, Mario A. Does democracy cause growth? A meta-analysis (of 2000 regressions). European Journal of Political Economy, v. 61, 2020.
• COPPEDGE, Michael; GERRING, John; KNUTSEN, Carl Henrik; LINDBERG, Staffan I.; TEORELL, Jan; ALTMAN, David; BERNHARD, Michael; CORNELL, Agnes; FISH, M. Steven; GJERLØW, Haakon; GLYNN, Adam; GRAHN, Sandra; HICKEN, Allen; KINZELBACH, Katrin; MARQUARDT, Kyle L.; MCMANN, Kelly; MECHKOVA, Valeriya; NEUNDORF, Anja; PAXTON, Pamela; PEMSTEIN, Daniel; RYDÉN, Oskar; VON RÖMER, Johannes; SEIM, Brigitte; SIGMAN, Rachel; SKAANING, Svend-Erik; STATON, Jeffrey; SUNDSTRÖM, Aksel; TZELGOV, Eitan; UBERTI, Luca; WANG, Yi-ting; WIG, Tore; ZIBLATT, Daniel. V-Dem [Country-Year/Country-Date] Dataset v13. Varieties of Democracy (V-Dem) Project, 2023. Available at: https://doi.org/10.23696/vdemds23. Accessed on: December 1, 2024.
  » https://doi.org/10.23696/vdemds23
• CUBERES, David; JERZMANOWSKI, Michał. Democracy, diversification and growth reversals. The Economic Journal, v. 119, n. 540, p. 1270-1302, 2009. Available at: Available at: https://www.jstor.org/stable/40271391 Accessed on: December 1, 2024.
  » https://www.jstor.org/stable/40271391
• DOUCOULIAGOS, Hristos; ULUBAŞOĞLU, Mehmet Ali. Democracy and economic growth: a meta-analysis. American Journal of Political Science, v. 52, n. 1, p. 61-83, 2008.
• ELLIOTT, Graham; ROTHENBERG, Thomas J.; STOCK, James H. Efficient tests for an autoregressive unit root. Econometrica, v. 64, n. 4, p. 813-836, 1996.
• ENGLE, R. F.; GRANGER, C. W. J. Cointegration and error-correction: representation, estimation, and testing. Econometrica, v. 55, n. 2, p. 251-276, 1987.
• FITTIPALDI, Ítalo; GAMA NETO, Ricardo Borges; ARAÚJO, Cletiane Medeiros; COSTA, Saulo Felipe. Crescimento econômico, democracia e instituições: quais as evidências dessas relações causais na América Latina? Revista de Sociologia e Política, Curitiba, v. 25, n. 62, p. 115-129, 2017.
• FURTADO, Celso. The economic growth of Brazil: a survey from colonial to modern times. Translated by R. W. Aguiar and E. C. Drysdale. Berkeley: University of California Press, 1968.
• GERRING, John; BOND, Philip; BARNDT, William T; MORENO, Carola. Democracy and economic growth: a historical perspective. World Politics, v. 57, n. 3, p. 323-364, 2005. Available at: Available at: https://www.cambridge.org/core/journals/world-politics/article/abs/democracy-and-economic-growth-a-historical-perspective/1BD38458A835E7F34F9A25191B68ECC7 Accessed on: December 1, 2024.
  » https://www.cambridge.org/core/journals/world-politics/article/abs/democracy-and-economic-growth-a-historical-perspective/1BD38458A835E7F34F9A25191B68ECC7
• GRANGER, Clive. W. J. Investigating causal relations by econometric models and cross-spectral methods. Econometrica, v. 37, n. 3, p. 424-438, 1969.
• GRIER, K.B.; TULLOCK, G. An empirical analysis of cross-national economic growth, 1951-1980. Journal of Monetary Economics, v. 24, n. 2, p. 259-276, 1989.
• GRÜNDLER, Klaus; KRIEGER, Tommy. Democracy and growth: evidence from a machine learning indicator. European Journal of Political Economy, v. 45, p. 85-107, 2016. Available at: Available at: https://www.sciencedirect.com/science/article/pii/S0176268016300222 Accessed on: December 1, 2024.
  » https://www.sciencedirect.com/science/article/pii/S0176268016300222
• HADDAD, Claudio. O crescimento do produto real do Brasil, 1900-1947. Rio de Janeiro: Fundação Getúlio Vargas, 1978.
• HUNTINGTON, S. P. Political order in changing societies. Yale University Press, 1968.
• IBGE. Sistemas de Contas Nacionais, 2023a. https://www.ibge.gov.br/estatisticas/economicas/contas-nacionais/9052-sistema-de-contas-nacionais-brasil.html Accessed on: December 1, 2024.
  » https://www.ibge.gov.br/estatisticas/economicas/contas-nacionais/9052-sistema-de-contas-nacionais-brasil.html
• IBGE. Censo 2022, 2023b. https://censo2022.ibge.gov.br/panorama/?utm_source=ibge&utm_medium=home&utm_campaign=portal Accessed on: December 1, 2024.
  » https://censo2022.ibge.gov.br/panorama/?utm_source=ibge&utm_medium=home&utm_campaign=portal
• JOHANSEN, S. Statistical analysis of cointegrating vectors. Journal of Economic Dynamics and Control, v. 12, n. 2-3, p. 231-254, 1988.
• JOHANSEN, S.; JUSELIUS, K. Maximum likelihood estimation and inference on cointegration - with applications to the demand for money. Oxford Bulletin of Economics and Statistics, v. 52, n. 2, p. 169-210, 1990.
• KENNEDY, Paul. A guide to econometrics. 6. ed. Blackwell Publishing, 2008.
• KLOMP, Jeroen; DE HAAN, Jakob. Political institutions and economic volatility. European Journal of Political Economy, v. 25, n. 3, p. 311-326, 2009. Available at: Available at: https://www.sciencedirect.com/science/article/abs/pii/S0176268009000135 Accessed on: December 1, 2024.
  » https://www.sciencedirect.com/science/article/abs/pii/S0176268009000135
• LEVITSKY, S.; WAY, L. Revolution & dictatorship: the violent origins of durable authoritarianism. Princeton University Press, 2022.
• LIPSET, S. M. Some social requisites of democracy: economic development and political legitimacy. The American Political Science Review, v. 53, n. 1, p. 69-105, 1959.
• LISBOA, Marcos; LATIF, Zeina. Democracy and growth in Brazil. Insper Working Paper WPE: 311/2013, 2013. https://inctpped.ie.ufrj.br/spiderweb/dymsk_5/5.3-10S%20Lisboa%20Latif.pdf Accessed on: December 1, 2024.
  » https://inctpped.ie.ufrj.br/spiderweb/dymsk_5/5.3-10S%20Lisboa%20Latif.pdf
• LITTLE, Andrew; MENG, Anne. Subjective and objective measurement of democratic backsliding. Working Paper, 2023.
• MARSHALL, Monty G.; GURR, Ted Robert. POLITY5 political regime characteristics and transitions, 1800-2018 dataset users’ manual. Center for Systemic Peace and Societal-Systems Research Inc., 2020.
• MOBARAK, Ahmed M. Democracy, volatility, and economic development. The Review of Economics and Statistics, v. 87, n. 2, p. 348-361, 2003. Available at: Available at: http://spinup-000d1a-wp-offload-media.s3.amazonaws.com/faculty/wp-content/uploads/sites/45/2019/06/democracy-volitality.pdf Accessed on: December 1, 2024.
  » http://spinup-000d1a-wp-offload-media.s3.amazonaws.com/faculty/wp-content/uploads/sites/45/2019/06/democracy-volitality.pdf
• MUSACCHIO, Aldo; LAZZARINI, Sergio G. Reinventando o capitalismo de estado: o leviatã nos negócios: Brasil e outros países. Penguim-Portfolio, 2015.
• NARAYAN, Paresh Kumar; SMYTH, Russell. Democracy and economic growth in China: evidence from cointegration and causality testing. Review of Applied Economics, v. 2, n. 1, p. 81-98, 2006.
• NARAYAN, Paresh Kumar; NARAYAN, Seema; SMYTH, Russell. Does democracy facilitate economic growth or does economic growth facilitate democracy? An empirical study of Sub-Saharan Africa. Economic Modelling, v. 28, n. 3, p. 900-910, 2011.
• NORRBIN, Stefan C.; YIGIT, F. Pinar. The robustness of the link between volatility and growth of output. Review of World Economics/Weltwirtschaftliches Archiv, v. 141, n. 2, p. 343-356, 2005.
• NORD, Marina; LUNDSTEDT, Martin; ALTMAN, David; ANGIOLILLO, Fábio; BORELLA, Cecilia; FERNANDES, Tiago; GASTALDI, Lisa; GOD, Ana Good; NATSIKA, Natalia; LINDBERG, Staffan I. Democracy report 2024: democracy winning and losing at the ballot. Gothenburg, Sweden: V-Dem Institute, 2024.
• PERSSON, Torsten; TABELLINI, Guido. Democracy and development: the devil in the details. American Economic Review, v. 96, n. 2, p. 319-324, 2006. Available at: https://www.aeaweb.org/articles?id=10.1257/000282806777212396 Accessed on: December 1, 2024.
  » https://doi.org/10.1257/000282806777212396
• PERSSON, Torsten; TABELLINI, Guido. Democratic capital: the nexus of political and economic change. American Economic Journal: Macroeconomics, v. 1, n. 2, p. 88-126, 2009.
• PRZEWORSKI, Adam; ALVAREZ, Michael; CHEIBUB, José Antônio; LIMONGI, Fernando. O que mantém as democracias? Lua Nova, São Paulo, n. 40-41, p. 113-135, 1997.
• PRZEWORSKI, Adam; LIMONGI, Fernando. Modernization: theories and facts. World Politics, v. 49, n. 2, p. 155-183, 1997.
• PRZEWORSKI, Adam; ALVAREZ, Michael; CHEIBUB, José Antônio; LIMONGI, Fernando. Democracy and development: political institutions and well-being in the world, 1950-1990. Cambridge University Press, 2000.
• QASIM, Hafiz Muhammad. The relationship between democracy and economic growth of Pakistan: a cointegration analysis. South Asian Studies: A Research Journal of South Asian Studies. v. 36, n. 2, p. 399-418, 2021.
• RAMEY, Garey; RAMEY, Valerie A. Cross-country evidence on the link between volatility and growth. The American Economic Review, v. 85, n. 5, p. 1138-1151, 1995.
• REMMER, K. Democracy and economic crisis: the Latin American experience. World Politics, v. 42, n. 3, p. 315-335, 1990.
• RODRIK, Dani; SUBRAMANIAN, Arvind; TREBBI, Francesco. Institutions rule: the primacy of institutions over geography and integration in economic development. Journal of Economic Growth, v. 9, p. 131-165, 2004.
• SAKYI, Daniel; ADAMS, Samuel. Democracy, government spending and economic growth: the case of Ghana, 1960-2008. The Journal of Applied Economic Research, v. 6, n. 3, p. 361-383, 2012.
• SLATER, D.; WONG, J. From development to democracy: the transformations of modern Asia. Princeton University Press, 2022.
• SLOAN, J.; TEDIN, K. L. The consequences of regimes type for public-policy outputs. Comparative Political Studies, v. 20, n. 1, p. 98-124, 1987.
• SOSKICE, David; IVERSEN, Torben. Democracy and prosperity: reinventing capitalism through a turbulent century. Princeton University Press, 2019.
• TAVARES, Jose; WACZIARG, Romain (2001). How democracy affect growth. European Economic Review, v. 45, p. 1341-1378, 2001. Available at: Available at: https://www2.novasbe.unl.pt/Portals/0/KnowledgeCenters/Economics%20for%20Policy/documents/How-democracy-affects-growth.pdf Accessed on: December 1, 2024.
  » https://www2.novasbe.unl.pt/Portals/0/KnowledgeCenters/Economics%20for%20Policy/documents/How-democracy-affects-growth.pdf
• ZAKARIA, Fareed. Age of revolutions: progress and backlash from 1600 to the present. W.W. Norton & Company, 2024.

Data availability

The datasets generated and analyzed during the current study are available in the Harvard Dataverse repository: https://doi.org/10.7910/DVN/P5JWLJ

• ¹
  Fapesp project N. 19/16970-5.
• ³
  Obviously, if the two series are integrated of order 1, i.e. they are non-stationary, we could still do regression by OLS if the residuals were stationary (Engle; Granger, 1987). However, in this case, the residuals of the regressions between democracy and GDP are not stationary.

• Editors
  Debora Rezende de Almeida
  Rebecca Neaera Abers

Publication Dates

• Publication in this collection
  13 Oct 2025
• Date of issue
  2025

History

• Received
  29 Apr 2024
• Accepted
  12 May 2025