Sir Clive Granger

Building Econometric Models to Consider Policy for Deforestation in the Amazon Rain Forest

Thursday, 17 August 2006
14:00 - 14:30 CEST



The world’s rain forests make substantial contributions to the world’s oxygen production and carbon dioxide capture. The largest rain forest is in the Amazon region in South America, most of which is in Brazil. This forest is 38% the size of the United States and ten times the size of Japan. It is being deforested rapidly. The amount of forest lost so far is equivalent to the size of Alaska and California.


As we are interested in policy questions it is necessary to obtain a data set over time and space. Data was obtained on a variety of environmental, commercial, and economic variables for 250 consistently defined regions of the Brazilian rain forest at five year intervals and for five time periods, producing a substantial panel data set. Because of problems of definition and measurement, data quality is varied, so robust techniques of estimation have to be used.

Most of the forest measurements are made on the ground but some make use of satellite information. The result is a panel data set.

Explanatory Variables, Policy

A model may be constructed with the amount of deforestation as a dependent variable and several possible explanatory variables such as population increase, amount of deforestation in adjacent areas, increase in road length in the areas, and changes in various policy variables.

Policies can be either economics (such as taxes or prices) or legal (actions are discouraged or are illegal and are penalized at different levels).

Who Cuts Trees?

Directly responsible for deforestation are:

a. Loggers – the sales of timber is a major export business for Brazil;

b. “Settlers” – mostly farm workers from Northeast Brazil who do not own land. They take sections of forest, cut down the trees, burn them, then attempt to farm the new cleared land.

Unfortunately, the cleared land is of poor quality and does not make successful farm land. After a few years the settlers have to cut down more trees to search for useful land.

It seems that by building new roads in areas that are already partly deforested (instead of to virgin forest), we can persuade settles to move there and then both the forest and the settlers can be better served.

New models will have to investigate this possibility and new policies developed that would achieve such an outcome.

Related Laureates