Um Modêlo Econométrico Aplicado à Economia Brasileira / An Econometric Model Applied To The Brazilian Economy

Gerhard Tintner, Isabella Consigliere, and José T. M. Carneiro’s 1970 journal article presents a compact Keynesian model for Brazilian economic analysis and policy, in parallel Portuguese and English texts. Its eight-part organization moves from the model and parameter estimation to numerical results, simultaneous elasticities, interpretation, a data appendix, and bibliography. Adapting an earlier model developed for OEEC countries, the United States, and Canada, the authors seek to translate estimated economic relationships into a usable account of how policy instruments affect several outcomes together.

Êste estudo se propõe a aplicar à economia brasileira um modêlo econométrico que possa servir a fins de análise e política econômica.

English translation: This study proposes to apply to the Brazilian economy an econometric model that can serve purposes of analysis and economic policy.

The practical ambition is inseparable from statistical scarcity. The model contains five endogenous variables: private consumption, nominal gross national product, the price index, real gross national product, and total employment. Two structural equations describe consumption and production; three further equations establish nominal expenditure, deflation, and employment demand. Government expenditure, investment, inventories, exports, imports, population, industrial wages, and time are treated as exogenous, although the authors concede that only time fully meets the independence requirement. Thus, the distinction between policy instruments and economic outcomes is an analytical assumption, not an established property of every included variable.

Consumption estimation supplies the first methodological problem. Relating consumption directly to current income produces an apparently strong fit, but exposes the estimates to simultaneous-equation bias: consumption helps determine the income used to explain it. The authors therefore employ previous-year real income per person as a predetermined explanatory variable. Their reported consumption equation has an intercept of 5.39, a slope of 0.4014, and a correlation coefficient of 0.52. They also identify inflation as a structural difficulty. The choice favors an attempt to address endogeneity over a superficially more impressive correlation.

Production estimation makes the limitations still more consequential. Without capital-stock data, the authors use a Cobb–Douglas-type function relating real output solely to employment. Rather than regress output directly on an endogenous labor variable, they adopt Klein’s indirect procedure. Assuming that the real wage equals labor’s marginal product, they estimate the production elasticity from labor’s share of output, using a geometric mean, and then calculate the intercept. The resulting employment elasticity is 0.695 and the logarithmic intercept is 4.05. Yet manufacturing wages stand in for wages throughout the economy, which the authors acknowledge overstates labor’s contribution and its estimated elasticity. Their warning establishes the appropriate status of the numerical results:

The conclusions obtained from the available data should be used with caution.

The article’s central conceptual move is to convert this small structural system into simultaneous elasticities. Instead of estimating isolated responses, it asks how consumption, prices, nominal and real output, and employment change together when one designated external variable increases by one percent. Government spending, investment, inventory changes, and net exports are grouped into a non-private-consumption expenditure aggregate. Differentiating the equations, expressing changes proportionally, and substituting mean values yields recursive response systems for changes in population, wages, and expenditure.

It was necessary, therefore, to operate in terms of relative and not absolute variables.

This transformation gives the model its proposed policy relevance: the same intervention can be evaluated across several objectives. A one-percent increase in government expenditure produces estimated increases of 0.31 percent in nominal output, 0.20 percent in real output, 0.23 percent in private consumption, 0.11 percent in wholesale prices, and 0.30 percent in employment. Investment has nearly identical effects within the model. Exports have positive effects and imports negative ones, with similar magnitudes. Population growth raises every endogenous variable, but real output rises less than population; the reported response therefore does not imply an improvement in output per person.

The wage experiment reveals how strongly the results depend on the model’s construction. A one-percent nominal wage increase raises prices by 0.82 percent while reducing real output by 0.42 percent and employment by 0.60 percent. The authors recognize that these negative responses may appear unrealistic. Their explanation invokes the ceteris paribus restriction, marginal-productivity theory, and a production function containing labor alone. These are conditional implications of the assumed system, not independently established effects of wage policy.

Certainly the results obtained could have been better if we had more and better sources of information and data to build a more complete model.

The appendix makes that qualification concrete. Its annual observations cover 1952–1964. National expenditure, population, and wholesale-price series come from IMF statistics; manufacturing wages are constructed from Brazilian statistical yearbooks and checked against other wage publications. Total employment is reconstructed using benchmark employment data and an industrial employment index. These procedures extend sparse observations into an economy-wide dataset, but also transmit industrial-sector assumptions into aggregate estimates.

The parallel texts require care: the English model initially prints lagged consumption where the Portuguese equation and later estimation use lagged income, and both elasticity derivations subsequently write consumption against contemporaneous income without explaining the removal of the lag. Such discrepancies limit literal replication. The article’s enduring contribution is its explicit bridge from structural estimation to joint policy responses, together with unusually direct acknowledgment of the informational and theoretical restrictions governing those responses.