ScholarLens Back to search
1978•RePEc: Research Papers in EconomicsRequires access

Difficulties with Testing for Causation

Rodney L. Jacobs, Edward E. Leamer, Michael P. Ward

Open publisher page 1 citations

Abstract

This paper examines the use of econometric tests for causality that are based on the work of Granger and Sims. The main conclusion is that it is not possible to infer the direction of causation from observed correlations. There are two reasons why the causality tests cannot reveal the direction of causation from observed correlations. First, the null hypothesis tested is necessary but not sufficient to imply causality. Second, any specification error renders the causality test results uninterpretable.

Open-access reader

About this research paper

What this paper is about

This paper examines the use of econometric tests for causality that are based on the work of Granger and Sims. The main conclusion is that it is not possible to infer the direction of causation from observed correlations. There are two reasons why the causality tests cannot reveal the direction of causation from observed correlations. First, the null hypothesis tested is necessary but not sufficient to imply causality. Second, any specification error renders the causality test results uninterpretable.

Why it matters

OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

This paper examines the use of econometric tests for causality that are based on the work of Granger and Sims. The main conclusion is that it is not possible to infer the direction of causation from observed correlations. There are two reasons why the causality tests cannot reveal the direction of causation from observed correlations. First, the null hypothesis tested is necessary but not sufficient to imply causality. Second, any specification error renders the causality test results uninterpretable.

Key concepts: Causation, Causality (physics), Granger causality, Null hypothesis, Econometrics, Null (SQL), Economics, Epistemology

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
Difficulties with Testing for Causation — Research Paper | ScholarLens