Brazil’s economy (Part 2): Employment

“Industry and Idleness – Plate 1: The Fellow Prentices at their Looms” William Hogarth (c. 18th century)

This second article in the series follows my previous article titled “What’s causing Brazil’s economic crisis? (Part 1)”, now renamed “Brazil’s economic downturn (Part 1): A crisis?“.  This article begins to answer the question by examining what is causing changes in Brazilian employment levels, which seems to be what most Brazilians (very naturally) worry most about.

Analyses suggest Brazil’s economic downturn actually began in late 2012 in the mining, minerals, manufacturing, and building materials sectors; over a full year earlier than the mid-2014 downturn in the broader overall economy.  What about the causes of the downturn?  Econometric analyses–among other things–suggest that over the long-run …

(1)  97% of Brazil’s employment growth can ultimately be attributed to (i) domestic production, (ii) domestic capital investment, (iii) debt financing of such investment, (iv) exchange rate, and, to a very minor extent, (v) exports.

(2) 58% of employment growth can be attributed to domestic capital investment and lending growth, taken together.

(3) 13% of employment growth can be attributed to domestic production growth, independent of capital investment and lending growth.

(4) 29% of employment growth can be attributed to employer expectations; predominately those related to Brazil’s foreign exchange rate.

These  findings, particularly the last, are interesting given that the World Economic Forum has recently (circa 2015) posed the question, Why is Brazil’s economy closed to trade?  Hmmm.  Which is it then?  Are Brazil’s economic problems primarily domestic because it has a closed economy, or do they stem from foreign exchange which implies significant international trade? 

Interested?  Please read on …

1. Brazil’s employment trends by industry sector, 1998-2016

Before exploring Brazil’s employment trends, it’s helpful to have some historical background on Brazil’s political economy …

Export-import economy trends.  Most modern economies are comprised of two semi-related segments; one of which might be termed an internal economy–related to what economists call an autarky–and the other being termed an external economy characterized by more or less free cross-border trade.  Indeed, Brazil was an almost entirely closed economy from 1964 to 1985 (!), including “interesting” conditions such as–I am told–laws against teaching any language other than Portuguese in the public education system.  (My separation of Brazil’s economy into internal and external segments should not be taken too literally; the two segments interact.  It’s simply a way of temporarily simplifying discussion of what’s happening in Brazil’s economy and what’s causing it.)

Brazil’s history and the distinction between internal and external economy are important because Brazil, being a large country, is still adjusting to its integration into the global economy, which seems to have begun in earnest in the early 1990s.  So, Brazil still has an internal economy that is a quasi-autarky, even though it has expanded its exports (and imports too) at a rapid pace since 1995:

Before the reader becomes upset at my use of current local currency units to construct the graph rather than constant, real exchange rate adjusted, currency units, please understand that I’ll examine exports in more detail in later articles.  The graph is simply used to demonstrate …

Brazil’s exports in terms of their nominal, internal purchasing power have grown quite rapidly; noting that Brazilian employees–the subject of this article–are necessarily paid in, and spend, only current local currency units.

I know: I should at least use export volume adjusted for inflation in the Brazilian currency.  But this is not a perfect measure either because it’s not clear that price inflation affects all economic classes in Brazil equally.  So, again, I will address these issues in future articles.

The rapid growth in Brazil’s import-export economy since 1995 suggests that imports and exports–in one way or another–might have an important influence on the Brazilian economy in general, and on it’s employment levels in particular.  To simplify analysis of Brazil’s employment trends, I’ve separated employment trends into industry segments roughly  characterized as those related to the internal economy and those related to the external economy.

Brazil’s internal economy.  Consider a graph showing Brazil’s industry employment trends from quarter 1998.1 through 2016.2 for a set of industry segments, with the exception of agriculture, perhaps having little to do directly with the external economy:

As seen, this graph shows that Brazil’s economic downturn in terms of internal economy employment levels began in mid-2014 as previously shown in Part 1 of the series.  It can also be seen that the downturn in agricultural employment seems to have begun in 2008, but I will set this issue aside until future articles.  (Incidentally, even though agricultural products represent Brazil’s largest export product group, I put agricultural employment in this graph only to avoid cluttering up the graph below.)

Although history suggests any economic contraction tends to be painful, similar to ideas presented in Part 1, the above graph seems to suggest something a bit less dramatic than a full-blown economic crisis.  After all, it can be seen that there were similar Brazilian employment downturns between 1998 and 2001.  So, painful? Yes. Crisis? Perhaps not.

Brazil’s external economy.  In contrast to the relatively mild contraction in employment in Brazil’s so-called internal economy industry sectors shown above, now consider employment trends in what might be termed Brazil’s external economy sectors:

This graph includes the building materials industry sector, even though perhaps most building materials are incorporated into Brazilian fixed capital investment rather than exported.  Building materials sector employment was included here because of its high correlation to mining, minerals, and manufacturing, which actually do represent some of Brazil’s leading exports.

The graph shows two important aspects of Brazil’s employment trends that seem to suggest some kind of crisis:  (1) The downturn in the export-related industry sectors of mining, minerals, and manufacturing began substantially earlier that downturns in Brazil’s internal economy-related sectors.  (2) The downturns in these sectors–and building materials sector–are far steeper than in the internal sectors.  One reasonable interpretation of these things is that sharp downturns in external economy sectors has perhaps had knock-on effects on Brazil’s internal economy.

Industry sector-weighted employment growth.  The graphs presented above are indexed to 1998.1 to show relative growth among industry sectors, which has the effect of obscuring the relative importance of each sector to overall employment (e.g., the sharp contraction in building materials sector employment would be devastating if it represented 95%+ of the economy, but it doesn’t … ).  To see more clearly the relative importance of each industry sector, I’ve summarized each sector’s influence on the overall 1998-2016 change in Brazil’s total employment level:

To be clear, each number on the left-hand side of the equation represents the weight each sector carries in Brazil’s overall employment growth from 1998 to 2016.  It can be seen, for example, that the steep contraction in building materials sector employment does not have nearly the effect on total employment as does contractions in manufacturing and services.

Most importantly, the expression shows that between 1998 and 2016 Brazil’s internal economy sectors represented about 73%, and the external economy sectors represented about 27%, of its total change in employment levels.  Interestingly, this suggests there is no particular evidence consistent with the suggestion that Brazil’s economy was beginning to suffer from the “Dutch Disease“, made by The Economist in 2010.  Brazil’s oil and gas industry (in the mining and minerals sector) is one of the fastest growing in recent years, but this is the only sector where employment is contracting.  The strong growth in manufacturing and services is precisely the opposite of over-reliance on natural resources for economic growth.

What’s needed now is to figure out what economic factors reasonably have a causal relationship with employment levels, and then test whether empirical data is consistent with the hypothesis that the factors have a significant negative influence on Brazil’s employment levels.

2. An econometric model of overall employment growth

Econometric methods and models.  There is a branch of mathematical statistics related to economics, termed econometrics, that is critically useful in estimating and making inferences about causal relationships between economic variables.  Because we’re interested in understanding the causes of changes in Brazilian employment levels, this means it’s necessary to estimate and test relationships between employment levels and potential causal factors.  So, I will use econometric methods here.  It’s usually a bad idea to simply hypothesize causal relationships between economic variables in an ad hoc way in contrast to developing them carefully from economy theory (it leads to the bad side of data mining), I will do it now and just ask the reader to trust that I’m not being deceptive in my choice of causal relationships to examine when estimating and testing econometric models.

For a variety of technical reasons relating to econometric methods, it is best to develop econometric models of employment and its causal factors in terms of employment growth rates and causal factor growth rates.  In the models, I will use logarithmic growth rates having the following general form:

Total employment growth rate model.  Basic economic theory suggests that the demand for employees (“labor”) depends on current domestic economic output (GDP).  GDP, in turn, depends on domestic capital investment (I), and current and expected lending (debt, D), Brazilian Real X US Dollar exchange rate (BRLUSD), and Brazilian exports (EXP).  After testing about 10-12 different model variations of Brazil’s total employment growth rate and such causal factor growth rates using 78 quarters of data from 1998 to 2016 (obtained from Banco Central do Brasil with details available on request), the following model seemed to be the most accurate, reliable representation of the effects of such factors on employment growth:

The model includes the contemporaneous quarter effects of GDP, I, and D on employment growth as well as  indicator variables, Q1, Q2 and Q3, to capture seasonal effects.  Prior quarter total employment growth rate is included to capture effects of prior quarter causal factors having a lagged effect on the growth rate and the prior quarter interaction between Brazil’s foreign exchange rate and export volume, which has a differential effect from other lagged causal factors.

As shown, the model also includes the effects of the forward-quarter Brazilian exchange rate and lending/debt, which can be interpreted as (proxies for) the effects of employers’ exchange rate expectations and its influence on profits, and lending availability expectations and the related effects on business financing, both of which are likely to influence employer’s hiring decisions.

It’s important not to regard the model as uniquely true; there are many other models of, or ways to model, employment growth that might explain things better.  For example, I tested a model that included foreign direct investment growth but–mainly because I only had about 5 years of recent data–the statistical association with employment growth was weak.  So, this model just estimates and tests relationships loosely suggested by economic theory and, so, includes variables economists generally expect to see in an model of employment growth.

Parameter estimate interpretations.  The estimated parameter values related to GDP, I, D, BRLUSD, and EXP show the estimated average marginal effect of the variables on Brazil’s overall employment growth rate during the 1998-2016 sample period, which I believe are reasonable estimates of their effects into the near- or mid-range future.  Parameter estimates are interpreted as the estimated average effect of a unit change in the causal factor on the overall employment growth rate; e.g., a 1% increase in GDP has an estimated effect of a .208% increase in the overall employment growth rate; a 1% increase in I combined with a 1% increase in D has an estimated effect of a .934% increase in the overall employment growth rate, etc.

Overall model interpretation.  It can be seen that–setting aside seasonal factors–there are five factors that significantly influence overall employment growth: overall economic output, capital investment, lending/debt, exports, and exchange rate.  This will, no doubt, surprise no one familiar with economics or the Brazilian economy.  It does, however, provide very useful information for our analysis of the causes of Brazil’s economic downturn:

(i) The Rsquared statistic of .943 suggests the model explains about 97% (~ .943^-2) of the variation in overall employment growth rates, which can be seen fairly clearly in the following graph of actual growth rates and model predictions:

This means that if we can accurately and reliably predict overall economic output, capital investment, lending/debt, and the BRLUSD exchange rate, then we can reasonably predict about 97% of the variation overall employment growth in Brazil.

(ii) Even if we “knew” these were the causal factors influencing employment growth, to my knowledge we did not know the relative size of their marginal effects, or the extent to which they explained such growth.

So, we do indeed learn something from the above model, even if what we learn is not completely surprising.  With this model, we can now derive some basic conclusions about what is (failing at) causing significant employment growth in Brazil.

Relative effect magnitudes of causal factors.  Based on the model parameter estimates and ignoring the basic lagged and seasonal effects of causal factors, and evaluating the effects based on 1998-2016 average values of the factors, it can be shown that the relative effect magnitudes of the casual factors are …

As can be seen, GDP and capital investment in combination with lending explain about 71% of overall employment growth in Brazil, while the effects of foreign exchange rate and exports–again in combination with lending–explain about 29% of the growth, with exports being a very minor factor.

Because exports as a percent of GDP average only about 12% over the 1998-2016 period, I believe it’s reasonable to conclude that exports are not an important factor in overall Brazilian employment growth.  Recalling the econometric model explains about 97% of Brazilian overall employment growth, I also believe it’s reasonable to conclude that the causal factors included in the model capture all substantial factors influencing such growth.

3. Inferences about Brazilian employment growth

I will now summarize as concisely as possible inferences that can be made from the econometric analyses and the overall employment growth model.  The econometric results suggest:

(1) Approximately 97% of Brazil’s overall employment growth can ultimately be attributed to domestic production, domestic capital investment, debt financing of such investment, exchange rate, and exports.

(2) There is little evidence that other non-domestic factors including foreign direct investment and exports-imports have substantial effects on Brazil’s overall employment levels (e.g., results suggest exports have no independent effect on employment growth, only a relatively minor effect mediated by exchange rate fluctuations).

(3) Domestic capital investment and lending growth, taken together, appear to explain about 58% of overall employment growth.

(4) Domestic production growth (increase in GDP), appears to explain about 13% of overall employment growth.

(5) Expectations of other factors–but predominately the exchange rate between the Brazilian Real and the US Dollar–appear to explain about 29% over overall employment growth.

Perhaps most importantly when we consider that Brazil’s exports over the 1998-2016 period accounted for only between 8%-14% of total economic output (imports represent 8%-16% of GDP over the same period), I think it’s reasonable to conclude that employment problems other than those resulting from profitability associated the Real x US Dollar exchange rate are limited to things happening in Brazil’s internal economy.

I’m familiar with computer equipment and other types of manufacturers whose profitability was destroyed by exchange rate conditions, thus leading to insolvency and essentially closing manufacturing facilities in Brazil.  But does the exchange rate really explain what went wrong in Brazil in mid-2012?  I don’t think we know yet … .

4. Preliminary conclusions: Brazilian employment

I think the econometric model and related analyses of Brazil’s overall employment growth between 1998 and 2016 allow us to clearly (but preliminarily) conclude a few things:

Domestic production, investment, and lending matter to Brazil’s economy, and so does exchange rate volatility to the extent it influences Brazilian industrial profits.

But Brazil is rapidly changing, so these factors need to be explored in substantially more detail to understand what aspects of Brazil’s economy are stable and what aspects are changing.  I will continue to examine these issues in my next article.

Então, até mais! Abraços!

São Paulo

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