Skip to main content
Log in

Econometric methods and Reichenbach’s principle

  • Original Research
  • Published:
Synthese Aims and scope Submit manuscript

A Correction to this article was published on 16 May 2022

This article has been updated


Reichenbach’s ‘principle of the common cause’ is a foundational assumption of some important recent contributions to quantitative social science methodology but no similar principle appears in econometrics. Reiss (Philos Sci 72:964–976, 2005) has argued that the principle is necessary for instrumental variables methods in econometrics, and Pearl (In Causality: Models, reasoning and inference, Cambridge: Cambridge University Press, 2000/2009) builds a framework using it that he proposes as a means of resolving an important methodological dispute among econometricians. Through analysis of instrumental variables methods and the issue of multicollinearity, we aim to show that the relationship of the principle to econometric methods is more nuanced than implied by previous work but nevertheless may make a valuable contribution to the coherence and validity of existing methods.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Subscribe and save

Springer+ Basic
EUR 32.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or Ebook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

Change history


  1. Where \(P(A \wedge B)\) is the probability of A and B both occuring. We will use P(A|C) to represent the probability of A given that C is known to have occurred.

  2. The Spirtes et al. (2000) approach has also been employed in the analysis of time-series data.

  3. As Pearl (2009, pp. 194–195) notes, there are some nuanced distinctions to be made between confounders and confounding.

  4. As Rosenzweig & Wolpin (2000) put it, “This approach essentially assumes that some components of nonexperimental data are random” (my emphasis, Rosenzweig & Wolpin 2000, p. 827).

  5. While macroeconometrics is mostly concerned with datasets containing observations over time (‘large T, small N’ in econometricians’ parlance), microeconometrics focuses on single, or repeated, cross-sections (‘large N, small T’).

  6. See Woodward (1988) for what one might call a ‘traditional’ overview of the formalities of regression methods directed at philosophers. Our discussion relies more on the presentations by Manski (1991) and Wooldridge (2002a). A valuable additional reference is Morgan & Winship’s (2007) book, which provides an overview of regression and graph-based methods within a detailed discussion of causal inference based on counterfactuals.

  7. Linearity in the parameters allows E(y|x) to be a non-linear function of the explanatory variables. E.g. We could have \(E(y|{\mathbf {x}}) = \beta _{0} + \beta _{1} x_{1} + \beta _2 x_{2} + \beta _3 x^{2}_{2}\). The linearity assumption is convenient though not essential.

  8. The implicit assumption of causality, whether in the presentation of methods or their application, is usually observed in the interpretation or use of the estimates from the subsequent empirical analysis, as noted by Woodward (1988).

  9. Specifically, “\(\varvec{\beta }\) can be written in terms of population moments in observable variables”(Wooldridge 2002a, (53). Using the vector of explanatory variables \({\mathbf {x}}\), we can write:

    $$\begin{aligned} {\varvec{\beta }} = E[{\mathbf {x}}^{T}{\mathbf {x}}]^{-1}E({\mathbf {x}}^{T}y) \end{aligned}$$

    This is obtained by rewriting (3) as \(y = \mathbf {x} {\varvec{\beta }}+u\), premultiplying both sides by the transposed vector \({\mathbf {x}}^{T}\), applying the expectations operator to both sides and solving for \(\varvec{\beta }\) (noting that \(E[{\mathbf {x}}^{T} {\mathbf {u}}] = 0\)).

  10. See Qin (2018) for a recent critique of that approach.

  11. There is some disagreement about how this severance of relationships should be represented; see Pearl (2009, pp. 376–377).

  12. This is identical in structure to a graph by Pearl (Figure 7.8(a), 2009, p. 248).

  13. See Angrist & Pischke (2009, pp. 117–120) for a summary of this example in the context of an explanation of the instrumental variables approach.

  14. The relevant expression is:

    $$\begin{aligned} {\varvec{\beta }} = E[{\mathbf {z}}^{T}{\mathbf {x}}]^{-1}E({\mathbf {z}}^{T}y) \end{aligned}$$

    Where \({\mathbf {z}}\) is the \({\mathbf {x}}\) vector including z and excluding \(x_{j}\).

  15. Indeed IV1 is a serious concern in the empirical literature because of theoretical results showing the negative consequences of ‘weak instruments’ (small value for \(corr(z, x_{j}|\mathbf {x_{\lnot j}})\)).

  16. As another example, in a widely-cited survey article, Angrist & Krueger (2001, p. 73) state: “In our view, good instruments often come from detailed knowledge of the economic mechanism and institutions determining the regressor of interest”.

  17. Similar approaches are also utilised in macroeconometrics, however the nature of much macroeconomic data-in the form of time-series at the country level-is such that there are additional complications which are outside the scope of the present paper.

  18. Causal transitivity means that, “For any three variables A, B, and C, if A causes B and B causes C, then A causes C” (Reiss 2005, p. 969). And Reiss defines functional correctness as: “A structural equation is functionally correct if and only if it represents the true functional (but not necessarily causal) relations among its variables”. In essence, this can be understood as meaning that an equation correctly represents the relationship between the magnitudes of variables in the equation, but those relations need not be causal.

  19. The word ‘definitional’ is intended to indicate that these assumptions are not ‘implicit’ in the sense of being wholly unstated (as Reiss suggests later in his paper). Rather, they follow directly from initial definitions of the problem, such as specification of the structural equations.

  20. The second assumption could be strengthened by requiring that the structural equation specify all causes, but this is not necessary for causal inference.

  21. Hoover (2007) draws attention to an analogous problem in Reiss’s reasoning about ‘collider’ variables.

  22. The fact that an instrument cannot belong in the true structural equation is an assumption made clear in a number of texts, e.g. Wooldridge (2002b, p. 517) and Pearl (Figure 7.8 (d), 2009, p. 248). Reiss actually proposes this later in that paper (see condition CIV-2, Reiss (2005, 973)) as one of a set of assumptions that would justify the econometrician’s approach, but this assumption clearly is made both in theory and in practice.

  23. There are two additional problems with Reiss’s argument. First, his proposed stage 2 assumption is in fact the standard way of interpreting the error term in a hypothesised structural equation. Econometric textbooks often make this interpretation explicit–see for instance Greene (2003, p. 8)-and it is recognised by Woodward (1988, p. 261). Pearl endorses a conceptual understanding of such error terms as representing omitted factors since it is a useful guide “when building, evaluating and thinking about causal models” (Pearl 2009, p. 162-163). Second, Reiss fails to note that one can describe the IV logic without reference to causality per se. Contra to Reiss’s criticism that “textbooks contain ‘recipes’ for econometric inference that give the impression that econometrics can proceed without causal background assumptions” (Reiss 2005, p. 966), econometrics can proceed without such assumptions. But, as per Cartwright’s mantra: ‘no causes in, no causes out’, not if the interest is in causal inference.

  24. We exclude Woodward’s reference to these as ‘causal’ assumptions since that remains a moot point.

  25. The actual merit of algorithmic approaches in social science is a separate, contentious issue; see the contributions to McKim & Turner (1997) and the discussion of ‘automated discovery’ by Glymour (2004).

  26. It may be useful for some readers to note that within the discipline it is known that implicit in such presentations in the econometrics literature is that if \( y \) was a cause of any explanatory variables–i.e. there was ‘simultaneous causation’–then an explicit system of equations would be required.

  27. Reiss makes an argument somewhat along these lines, including a discussion of a related, intervention-based, approach by Woodward (2003)–see Reiss (2005, pp. 972–974)–but it is afflicted by some of the misunderstandings already described.

  28. No particular justification for this assertion is provided, though in principle it should be testable; standard statistical tests can be used to examine the likelihood of a given correlation being due to chance. Yet such commentary typically makes no mention of examining the presumption of spuriousness.

  29. In matrix representations the assumption of no perfect collinearity is clearly stated and known as ‘the rank condition’.

  30. The relevant derivations are available from the author, but the result should be unsurprising given that the presence of causal relationships does not alter the statistical results.

  31. The case of a common cause is omitted since it will suffice for us to demonstrate that at least one system may exist that would require a reinterpretation of estimated coefficients.

  32. Implicit in this graph is that, following Pearl (2009), all variables affecting more than one other variable in the system are shown.

  33. A partial exception is reflected in one historical practice of conducting and reporting results from multiple regressions. The practice was to report regression results in such a way as to show how the incremental inclusion of every additional explanatory variable (‘control’) altered the estimated parameter on the explanatory variable of primary interest.

  34. Pearl’s graphical representation of this issue–Pearl (Fig. 11.5, 2009, p. 339)–is very similar to our Fig. 3.iii above, except that it does not include a direct (unmediated) arrow from X to Y.


  • Angrist, J. D., & Krueger, A. B. (1991). Does compulsory schooling attendance affect schooling and earnings? Quarterly Journal of Economics, 106, 976–1014.

    Article  Google Scholar 

  • Angrist, J., Imbens, G., & Rubin, D. B. (1996). Identification of causal effects using instrumental variables. Journal of the American Statistical Association, 91, 444–455.

    Article  Google Scholar 

  • Angrist, J. D., & Krueger, A. B. (2001). Instrumental variables and the search for identification: From supply and demand to natural experiments. Journal of Economic Perspectives, 15(4), 69–85.

    Article  Google Scholar 

  • Angrist, J. D., & Pischke, J.-S. (2009). Mostly harmless econometrics. Princeton: Princeton University Press.

    Book  Google Scholar 

  • Angrist, J. D., & Pischke, J.-S. (2010). The credibility revolution in empirical economics: How better research design is taking the con out of econometrics. Journal of Economic Perspectives, 24(2), 3–30.

    Article  Google Scholar 

  • Arntzenius, F. (1992). The common cause principle. PSA Proceedings of the Biennial Meeting of the Philosophy of Science Association, 2, 227–237.

    Article  Google Scholar 

  • Arntzenius, F. (2010). Reichenbach’s common cause principle. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Fall 2010 ed.).

  • Blanchard, O. J. (1987). Comment. Journal of Business and Economic Statistics, 5(4), 449–451.

    Google Scholar 

  • Cartwright, N. (1999). The Dappled World: A study of the boundaries of science. Cambridge: Cambridge University Press.

    Book  Google Scholar 

  • Cartwright, N. (2002). Against Modularity, the Causal Markov Condition, and Any Link Between the Two: Comments on Hausman and Woodward. The British Journal for the Philosophy of Science, 53(3), 411–453.

    Article  Google Scholar 

  • DiNardo, J. (2008). Natural experiments and quasi-natural experiments. In S. N. Durlauf & L. E. Blume (Eds.), The new Palgrave dictionary of economics. Basingstoke: Palgrave Macmillan.

    Google Scholar 

  • Farrar, D. E., & Glauber, R. R. (1967). Multicollinearity in regression analysis: The problem revisited. Review of Economics and Statistics, 49, 92–107.

    Article  Google Scholar 

  • Glymour, C. (2004). The automation of discovery. Daedalus (Winter), 69–77.

  • Greene, W. H. (2003). Econometric Analysis (5th Edn.). Pearson Education.

  • Hausman, D. M., & Woodward, J. (1999). Independence, invariance and the causal Markov condition. British Journal for the Philosophy of Science, 50(4), 521–583.

    Article  Google Scholar 

  • Heckman, J. J., & Vytlacil, E. J. (2007). Econometric evaluation of social programs, part II: Using the marginal treatment effect to organize alternative econometric estimators to evaluate social programs, and to forecast their effects in new environments. Volume 6, Part B of Handbook of Econometrics, pp. 4875–5143. Elsevier.

  • Hendry, D. F., & Krolzig, H.-M. (2005). The properties of automatic Gets modelling. Economic Journal, 115, C32–C61.

    Article  Google Scholar 

  • Holland, P. W. (1986). Statistics and causal inference. Journal of the American Statistical Association, 81(3), 945–960.

    Article  Google Scholar 

  • Hoover, K. (2001). Causality in Macroeconomics. Cambridge: Cambridge University Press.

    Book  Google Scholar 

  • Hoover, K. (2003). Nonstationary time series, cointegration, and the principle of the common cause. British Journal for the Philosophy of Science, 54, 527–551.

    Article  Google Scholar 

  • Hoover, K. (2004). Lost causes. Journal of the History of Economic Thought, 26(2), 149–164.

    Article  Google Scholar 

  • Hoover, K. (2007). Probability and structure in econometric models. Paper presented at the 13th International Congress of Logic, Methodology, and the Philosophy of Science at Tsinghua University, Beijing, China, 8–16 August 2007.

  • Hoover, K. (2009). Probability and structure in econometric models. In W. W. Clark Glymour and D. W. Hl (Eds.), Logic, methodology and philosophy of science: Proceedings of the Thirteenth International Congress, pp. 497–513. College Publications.

  • Hoover, K. (2011). Counterfactuals and causal structure. In P. Illari, F. Russo, and J. Williamson (Eds.), Causality in the Sciences, pp. 338–360. Oxford University Press.

  • Keane, M. P. (2010). A structural perspective on the experimentalist school. Journal of Economic Perspectives, 24(2), 47–58.

    Article  Google Scholar 

  • Levitt, S. D. (1997). Using electoral cycles in police hiring to estimate the effect of police on crime. American Economic Review, 87(4), 270–290.

    Google Scholar 

  • Levitt, S. D. (2002). Using electoral cycles in police hiring to estimate the effect of police on crime: Reply. American Economic Review, 92(4), 1244–1250.

    Article  Google Scholar 

  • Liu, T.-C. (1960). Underidentification, structural estimation, and forecasting. Econometrica, 28(4), 855.

    Article  Google Scholar 

  • Mansfield, E. R., & Helms, B. P. (1982). Detecting multicollinearity. The American Statistician, 36(3), 158–160.

    Article  Google Scholar 

  • Manski, C. (1991). Regression. Journal of Economic Literature, 29(1), 34–50.

    Google Scholar 

  • McKim, V. R., & Turner, S. P. (1997). Causality in Crisis? Statistical methods in the search for causal knowledge in the social sciences. Notre Dame (IN): University of Notre Dame Press.

    Google Scholar 

  • Morgan, S. L., & Winship, C. (2007). Counterfactuals and causal inference. Cambridge: Cambridge University Press.

    Book  Google Scholar 

  • Muller, S. M. (2012). Econometric methods and Reichenbach’s principle. SALDRU Working Paper 85.

  • Murray, M. P. (2006). Invalid instruments and coping with weak instruments. Journal of Economic Perspectives, 20(4), 111–132.

    Article  Google Scholar 

  • Pearl, J. (2000/2009). Causality: Models,reasoning and inference. Cambridge: Cambridge University Press.

  • Qin, D. (2018). Let’s take the bias out of econometrics. Journal of Economic Methodology, 26(2), 81–98.

    Article  Google Scholar 

  • Reichenbach, H. (1956/1991). The direction of time. Berkeley: University of California Press.

  • Reiss, J. (2005). Causal instrumental variables and interventions. Philosophy of Science, 72, 964–976.

    Article  Google Scholar 

  • Reiss, J. (2007). Time series, nonsense correlations and the principle of the common cause. In F. Russo & J. Williamson (Eds.), Causality and probability in the sciences (pp. 179–196). London: College Publications.

    Google Scholar 

  • Reiss, J. (2008). Error in Economics: The Methodology of Evidence-Based Economics. London: Routledge.

    Google Scholar 

  • Rosenzweig, M. R., & Wolpin, K. I. (2000). Natural “natural experiments” in economics. Journal of Economic Literature, 38(4), 827–874.

  • Sims, C. A. (1980). Macroeconomics and Reality. Econometrica, 48(1), 1.

    Article  Google Scholar 

  • Sober, E. (2001). Venetian sea levels, British bread prices, and the principle of the common cause. British Journal for the Philosophy of Science, 52, 331–346.

    Article  Google Scholar 

  • Spirtes, P., Glymour, C. N., Scheines, R. (1993/2000). Causation, prediction and search. Cambridge, MA: MIT Press.

  • Steel, D. (2003). Making time stand still: A response to Sober’s counter-example to the principle of the common cause. British Journal for the Philosophy of Science, 54(2), 309–317.

    Article  Google Scholar 

  • Woodward, J. (1988). Understanding regression. In PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association, pp. 255–269. Chicago: University of Chicago Press.

  • Woodward, J. (1995). Causation and explanation in econometrics. In D. Little (Ed.), On the Reliability of Economic Models: Essays in the Philosophy of Economics (pp. 9–62). Dordrecht: Kluwer.

    Chapter  Google Scholar 

  • Woodward, J. (2003). Making things happen. Oxford: Oxford University Press.

    Google Scholar 

  • Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.

    Google Scholar 

  • Wooldridge, J. M. (2002). Introductory econometrics: A modern approach. Cambridge: South-Western College Publishing.

    Google Scholar 

Download references


I am grateful to two anonymous reviewers for detailed comments and suggestions that led to improvements in this manuscript, and to Martin Wittenberg for helpful comments on the very first draft of this work. Participants at a SALDRU seminar in the School of Economics at the University of Cape Town and other anonymous reviewers provided comments that informed the first working paper version (Muller 2012). The usual disclaimers apply.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Seán Mfundza Muller.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

T.C.: All Things Reichenbach Lead Guest Editor: Dr. Erik Curiel and Flavia Padovan.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Muller, S.M. Econometric methods and Reichenbach’s principle. Synthese 200, 185 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: