Brazil’s economy (Part 4): Government spending

“The Tower of Babel” Pieter Bruegel the Elder (c. 1563)

Much has been said in the financial economic press about Brazilian government spending in recent years.  In many cases the articles criticize excessive government spending, and more recently–and from different ideological quarters–articles tend towards criticism of excessive government spending cuts.  Perhaps most provocative are comments by journalists similar to those of Felipe Moura Brasil, who in his video article “How Socialism Ruined My Country” says …

The socialists [Partido Trabalhista] increased government spending, deficits, and debt … [and] from 2008 to 2015, government spending grew nearly four times as fast as tax revenue.

Stunning, if true; so this needs to be examined more closely.  In this connection, this article presents evidence on Brazilian government spending between 1997 and 2016 suggesting among other things–apparently contrary to popular belief–that …

  • Tax revenue growth substantially exceeded spending growth; 1997-2016.
  • Spending growth exceeded tax revenue growth during the Cardoso regime.
  • Tax revenue growth exceeded spending growth during the Lula regime.
  • Spending growth exceeded tax revenue growth during the Rousseff regime.

These results suggest that the apparently common notion that Brazil’s fiscal problems result from political ideologies in a simple way is not correct.  As we all know but often forget, in a world of globally-connected capital markets it’s really quite difficult to implement fiscal policies based unilaterally on a political ideology.  To understand these things better, please read on …

Similar to my previous articles, the analyses presented in this article, by design, do not  fully explore cause-and-effect relationships related to Brazilian government spending.  Rather, these are preliminary analyses designed to develop a preliminary understanding from which I will develop more complete analyses in future articles.

1. Brazilian government spending, 1997-2016

Consider the following graph of Brazilian federal government spending between 1997 and 2016–measured in constant 1995 Brazilian Real currency, and indexed to the 1997.1 quarter–along with government tax revenues, total employment level, and private sector consumption:

Note that government spending does not include investment expenditures (e.g., infrastructure spending); it only includes what can be termed government consumption expenditures.  As in my previous articles on Brazil’s economy, I’ve included a vertical line in mid-2014 to show the time at which most people recognized Brazil was entering an economic downturn.

I selected the different data series in the graph because economic theory suggests government spending is positively associated with–and generally constrained by–tax revenues; but such tax revenues generally depend on employment levels and private sector consumption because these factors are positively associated with economic output and result in employment tax revenues, taxable profits, and therefore tax revenues (including VAT).  As can be seen, the basic predictions of economic theory seem to hold fairly well in the above graph with the exception that Brazilian government spending seems somewhat less sensitive to tax revenues after mid-2014.

As suggested by criticisms of Brazilian government spending referenced in the introduction, government spending depends not only on tax revenues but also on government fiscal / tax policy, which generally depends on political regime.  Based on this idea, the basic causal relationships between the following factors need to be estimated:

  • Brazilian government spending
  • Brazilian tax revenues
  • Brazilian political regime

As I’ve mentioned in previous articles, econometric methods are the only reasonable way to estimate and test such relationships. So, on to the econometric analysis …

2. An econometric model of Brazilian government spending

Estimated model.  Consider the following estimated econometric model of Brazilian government spending growth rate:

The model is stated in terms of growth rates for technical reasons and was estimated using Banco Central do Brasil data from the 78 quarters spanning 1997.1 to 2016.3.  The model shows the estimated marginal effects of the following factors on Brazilian government spending growth rates:

  • BRED: Indicator variable to capture economic downturn effects.
  • Q1,Q2,Q3: Quarterly indicator variables to capture seasonality effects.
  • TAX: Brazilian government tax revenue.
  • LDS: Luiz Inácio Lula da Silva political regime, 2003-2010.
  • DVRDilma Vana Rousseff political regime, 2011-2016.

The BRED variable was included in the model primarily to segregate the effects of Brazil’s economic downturn from the effects of other factors.  Because political regime fiscal policy depends on tax revenues, interaction effects are included in the model, which can be used to estimate how fiscal policy changes between regimes.

Regarding the characterizations of political regimes, note that Fernando Henrique Cardoso‘s political regime spanned from 1995-2002 and–though affiliated with the nominally social democratic party, Partido da Social Democracia Brasileira–his regime is widely considered as politically moderate.  It follows that the base level quarterly growth rate of .006 and 1.000 marginal effect of tax revenue growth are associated with Cardoso’s political regime spanning 1997-2002 in the data set.  The Lula and Rousseff regimes–regarded by many as socialist–basically span the remaining 2003-2016 period in the data set.

Model explanatory power and basic growth rate trend.  The model explains about 94% of the observed variation in government spending growth rates (the square root of the R-squared statistic); much of which can be seen to be related to quarterly seasonality in the rates, as shown in following graph spanning 1997 through 2016:

The model fits reasonably well, both in terms of technical considerations (R-squared and p-values on parameter estimates) and the visual fit to the data.  There are clearly other factors influencing the growth rates as seen by differences in the vertical distance between the model predictions (in blue) and the observed growth rates (in green), but I will assume the model is adequate for our purposes here.

With respect to the general time trend (shown in yellow), we can see that Brazilian government spending is ..

  • decreasing under the Cardoso regime,
  • increasing under the Lula regime,
  • increasing under the Rousseff regime until mid-2014, and
  • decreasing under the Rousseff regime after mid-2014.

Most interestingly, the general downward trend in growth before the Rousseff regime even began in 2011 suggests the so-called socialist regimes of Lula and Rousseff either intentionally or unintentionally were required to curtail their spending programs.  A more complete examination of this would require us to consider the effects of government debt, interest rate, and perhaps foreign exchange factors, but I will defer this type of analysis until future articles.

Marginal effects analysis.  At the most basic level, the econometric model estimates suggest the following with respect to Brazilian government spending (GS) growth rates and tax revenue growth rates under each regime:

  • Cardoso regime: 1% tax growth is associated with 1% GS growth.
  • Lula regime: 1% tax growth is associated with 1.187% GS growth.
  • Rousseff regime: 1% tax growth is associated with a .399% GS decrease.

This last result is striking: Although criticized for profligate spending, the Rousseff regime–which began in 2011–was reversing (or being forced to reverse) the usual positive relationship between government spending growth and tax revenue growth.  And, referring back to the graph of the econometric model predictions and actual observed growth rates, this apparently occurred quite some time prior to the mid-2014 economic downturn.

Further analysis is needed, however, because governmental political regimes not only influence the relationship between spending and taxation, but also growth in taxation resulting from tax legislation.  Consequently, it’s necessary to not only examine the relative marginal effects as I’ve shown above, but to also evaluate the average spending growth rate conditional on tax revenue growth rate during each of the three political regimes.

Regime growth rates comparison analysis.  Although I will not explain the math, the econometric model can be used to derive the following average annual effect of each political regime on Brazilian government spending growth rates, evaluated conditional on the average tax revenue growth rate under each respective regime (FHC = Cardoso, LDS = Lula, DVR = Rousseff):

As can be seen, Brazilian government spending growth rates are really quite similar in the Cardoso and Lula regimes, ranging between 2.9% and 3.0% average annual growth.  The annual spending growth rate in the Rousseff regime is a significantly higher 4.0% on average over the the 2011-2016 regime span.

As suggested earlier, because fiscal policy includes tax policy it’s necessary and helpful to summarize these average government spending growth rates by regime in comparison to their related average tax revenue growth rates.  Multiplying the average quarterly tax revenue growth rates for each regime by four to convert to an annual average rate results in the following comparisons by political regime (FHC = Cardoso, LDS = Lula, DVR = Rousseff):

  • FHC regime:  2.9% spending growth versus .5% tax revenue growth.
  • LDS regime:  3.0% spending growth versus 4.6% tax revenue growth.
  • DVR regime:  4.0% spending growth versus 1.0% tax revenue growth.

In this last result with respect to the Rousseff regime, we can apparently see evidence consistent with what journalist Felipe Moura Brasil identifies in the quote presented in the introduction.

But also notice these results are generally inconsistent with the notion that Brazilian government fiscal policy depends on political ideology: The results suggest the Cardoso regime spending policy with respect to tax revenues is qualitatively similar to that of the Rousseff regime; and, the Lula regime policy–at least in this regard–appears more consistent with what most people think of as associated with moderate and conservative ideologies.  So, in short, the analyses presented in this article are quite inconsistent with the view that political ideology influences Brazilian government fiscal policy in a simple, direct way.

I believe the above analyses basically exhaust the basic information we can obtain the econometric model in the form presented.  So, I will end here and summarize.

3. Preliminary conclusions: Brazilian government spending

Evidence presented in the article suggests that the simplistic view that political ideology is the primary influence in Brazilian government fiscal policy is not correct.  This suggests that there are other factors that need to be explored to more fully understand Brazilian government spending policies.  Factors likely influencing such policy including privatization programs where Brazilian government assets are sold to fund spending and investment, as well as government borrowing, will be explored in future articles.  So, until then …

Para frente e para cima!

São Paulo

Disclosure.  I am a political agnostic and make no judgment about political ideologies, in general.  I like to think of myself as a realist.  I have both acquaintances and good friends in both the political parties mentioned in the article.  No offense is intended by anything I’ve written above; my intent is simply to examine and present clearly a historical reality that does not seem to be widely understood by journalists and others (myself included).  MMc

Caveats.  Please note: (i) views presented above are my own and do not reflect those of others; (ii) like anyone, I’m not infallible and am responsible for any errors; (iii) I greatly appreciate being informed of any significant errors in facts, logic, or inferences and am happy to give credit to anyone doing so; (iv) the above article is subject to revision and correction; and, (v) the article cannot be construed as investment or financial advice and is intended merely for educational purposes.  MMc