Econometric Instruments and Colombian Economic Policy

Guillermo L. Gómez M. and Gerhard Tintner’s “Sobre una aplicación de la econometría a la política económica de Colombia,” published in Comercio Exterior in March 1977, develops an aggregate econometric model to inform national economic policy. Its five sections move from the Tinbergenian distinction between policy objectives and instruments through model specification, statistical estimation, numerical results, and economic interpretation. The central argument is that Colombia’s employment, prices, consumption, and production must be understood as interdependent outcomes: policy requires coordinated changes in instruments, not isolated responses to individual problems. The article supplies estimated relationships for that task without completing the optimization of a specified welfare function.

The model combines a Keynesian consumption function with neoclassical production and labor-demand relationships. Private consumption, nominal and real national product, prices, and employment are endogenous; population, wages, capital, government spending, investment, inventories, and foreign trade enter as exogenous variables. Consumption and income are expressed in real per-capita terms, while government expenditure, investment, inventory changes, and net exports are grouped as autonomous expenditure. Four production specifications explore different assumptions about returns to scale and technical change. This compact architecture connects demand expansion to productive capacity, factor remuneration, and employment.

The statistical analysis distinguishes disturbances in structural equations from errors in observed variables. Identification restrictions, independence assumptions, and normality provide the framework for estimation. Because consumption and income determine one another, the authors use a variant of indirect two-stage least squares that incorporates observational errors, with likelihood criteria helping select admissible coefficients. Their treatment of production is more tentative:

El hecho de que los resultados para la función de producción sean poco concluyentes puede justificarse cuando se tiene en cuenta que los datos para capital, empleo y salarios son poco confiables y que han sido parcialmente producidos por simulación dentro del modelo.

English translation: The fact that the results for the production function are inconclusive can be explained by taking into account that the data for capital, employment, and wages are unreliable and have been partly generated through simulation within the model.

This admission qualifies the policy interpretation. Multiple-regression estimates encounter multicollinearity, and the preferred constant-returns specification draws on the assumed equality between labor’s marginal product and its real remuneration. Its coefficients therefore depend on economic restrictions as well as empirical observations. The estimated production function assigns exponents of 0.86 to labor and 0.14 to capital; the consumption function yields a marginal propensity to consume of approximately 0.69.

The authors interpret that propensity as evidence of an economy disproportionately oriented toward consumption goods and insufficiently capable of producing capital goods. Inflation, currency depreciation, and consumer credit justify, in their account, omitting lagged consumption. Yet they acknowledge that distinguishing durable from nondurable goods would require unavailable time series. Their developmental diagnosis links weak capital profitability to inadequate investment and a preference for sectors offering rapid returns. Political insecurity, economic instability, and insufficient government incentives help explain why investment neglects productive capacity, reinforcing unemployment and low productivity.

For the period 1950–1967, the preferred specification produces average marginal productivities of 4.013 for labor and 0.0825 for capital. Comparing labor productivity with real wages, the authors argue that remuneration rising faster than productivity discourages hiring. This interpretation rests explicitly on profit maximization under perfect competition, rather than constituting an independent demonstration of wage effects. They caution:

Es muy importante no olvidar que se trata de un modelo mixto neoclásico-keynesiano muy agregado para así evitar conclusiones apresuradas e inadmisibles.

English translation: It is very important not to forget that this is a highly aggregated mixed neoclassical-Keynesian model, so as to avoid hasty and inadmissible conclusions.

Aggregation permits a national diagnosis but cannot locate the industries with the weakest productivity or the greatest investment potential. Alternative production specifications do not resolve this limitation: increasing returns produce results the authors regard as misleadingly favorable, while decreasing returns yield an economically implausible negative marginal productivity of capital.

The final conceptual move is from structural coefficients to simultaneous elasticities. Because nonlinear relationships obstruct a simple final-form solution, relative changes measure how each instrument affects several objectives together. A 1% wage increase is associated with roughly 0.88% higher prices, 0.72% lower real output, and 0.83% lower employment, while investment increases generally improve output, employment, and consumption. These are conditional model responses, not unrestricted forecasts; the reported numerical examples also contain discrepancies between prose and the elasticity table.

Evidentemente, el papel del economista-político es encontrar la combinación de cambios en las variables instrumento, que introduzca una mejora en la función objetivo, función que debe escogerse muy cuidadosamente.

English translation: Evidently, the role of the economic policymaker is to find the combination of changes in the instrument variables that brings about an improvement in the objective function, a function that must be chosen very carefully.

Mathematical programming is proposed for this subsequent stage, but choosing the welfare function remains unresolved. Political feasibility also matters: rather than endorsing wage freezes mechanically, the authors recognize union pressure and recommend compensating wage increases through investment or public expenditure. Export promotion should likewise expand production rather than restrict domestic consumption. Finally, aggregate recommendations require sectoral allocation:

Cuando el economista-político ha encontrado una combinación óptima de variaciones en las variables exógenas, es necesaria nuevamente una desagregación de éstas, para distribuir dichas variaciones de manera óptima entre sus componentes.

English translation: Once the economic policymaker has found an optimal combination of changes in the exogenous variables, these must again be disaggregated in order to distribute those changes optimally among their components.

The article’s relevance lies in making policy trade-offs quantitatively explicit while revealing the limits of that exercise. Its contribution is a framework for coordinating interventions; reliable data, sectoral analysis, and an explicitly chosen welfare objective remain necessary to turn the framework into an operational plan.