Journals / Economics Business and Organization Research / 2025 / Cilt: 7 Sayı: 1

ECONOMETRIC MODEL SPECIFICATION AND D-SEPARATION: A MONTE CARLO EXPERIMENT

EKONOMETRİK MODEL BELİRLEME VE D-AYRIMI: BİR MONTE CARLO DENEYİ

Fernando Vera-sánchez(Universidad Anáhuac de Puebla)0000-0003-3501-6478
Pages
21–39
Publication Date
2025-12-27
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

While earlier studies recommend using a proxy to address the issue of missing variables, the Directed Acyclic Graphs method advises a thorough evaluation of the proxy's attributes beyond merely its strong correlation with the unobservable variable.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:The purpose of this paper is to identify how Directed Acyclic Graphs perform in a commonly occurring setting in economics, namely the use of proxy variables.
📊
Key Findings & Data:We were able to demonstrate in a Monte Carlo simulation that the proxy variable suggested by existing literature could be worse than the procedure suggested by Directed Acyclic Graphs, these findings are relevant for the specification of economic models.
💡
Conclusion & Impact:While earlier studies recommend using a proxy to address the issue of missing variables, the Directed Acyclic Graphs method advises a thorough evaluation of the proxy's attributes beyond merely its strong correlation with the unobservable variable.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Vekil Değişkenler · Yönlendirilmiş Çevrimsiz Grafikler · Monte Carlo Simülasyonu

Abstract

The purpose of this paper is to identify how Directed Acyclic Graphs perform in a commonly occurring setting in economics, namely the use of proxy variables. When estimating the coefficients that describe the relationships between variables in an economic model, theoretical foundations guide the selection of variables to include in the statistical framework. However, some important variables lack measurable scales. Variables such as education, intelligence, or price expectations are often key explanatory factors, but since they are not directly observable, researchers must choose between treating them as omitted variables or using proxy indicators like years of schooling, test scores, or expected future prices. The method of this research aims to compare the traditional use of proxy variables with the methodology based on Directed Acyclic Graphs. To highlight the distinctions between both approaches, using a Monte Carlo simulation. We were able to demonstrate in a Monte Carlo simulation that the proxy variable suggested by existing literature could be worse than the procedure suggested by Directed Acyclic Graphs, these findings are relevant for the specification of economic models. Our conclusions indicate that dealing with an unobservable variable requires careful consideration when introducing a proxy variable. While earlier studies recommend using a proxy to address the issue of missing variables, the Directed Acyclic Graphs method advises a thorough evaluation of the proxy's attributes beyond merely its strong correlation with the unobservable variable.

Özet

Bu çalışmanın amacı, ekonomi literatüründe sıklıkla karşılaşılan bir durum olan vekil (proxy) değişkenlerin kullanımı bağlamında Yönlendirilmiş Çevrimsiz Grafiklerin (Directed Acyclic Graphs, DAG) nasıl bir performans sergilediğini incelemektir. Bir ekonomik modelde değişkenler arasındaki ilişkileri tanımlayan katsayılar tahmin edilirken, istatistiksel çerçeveye dâhil edilecek değişkenlerin seçimi kuramsal temellere dayanmaktadır. Ancak bazı önemli değişkenlerin ölçülebilir ölçekleri bulunmamaktadır. Eğitim, zekâ ya da fiyat beklentileri gibi değişkenler çoğu zaman temel açıklayıcı faktörlerdir; ne var ki doğrudan gözlemlenemedikleri için araştırmacılar bu değişkenleri ya dışlanmış değişkenler olarak ele almak ya da öğrenim süresi, test puanları veya beklenen gelecek fiyatlar gibi vekil göstergeler kullanmak arasında bir tercih yapmak zorunda kalmaktadır. Bu çalışmanın yöntemi, vekil değişkenlerin geleneksel kullanımını Yönlendirilmiş Çevrimsiz Grafiklere dayalı metodoloji ile karşılaştırmayı amaçlamaktadır. Her iki yaklaşım arasındaki farkları ortaya koymak için Monte Carlo simülasyonu kullanılmaktadır. Monte Carlo simülasyonu aracılığıyla, mevcut literatür tarafından önerilen vekil değişkenin, Yönlendirilmiş Çevrimsiz Grafikler tarafından önerilen prosedüre kıyasla daha kötü sonuçlar verebileceği gösterilmiştir. Bu bulgular, ekonomik modellerin belirlenmesi açısından önem taşımaktadır. Sonuçlarımız, gözlemlenemeyen bir değişkenle karşılaşıldığında vekil değişkenin modele dâhil edilmesinin dikkatli bir değerlendirme gerektirdiğine işaret etmektedir. Önceki çalışmalar, eksik değişken sorununu gidermek için vekil değişken kullanımını önermekteyken, Yönlendirilmiş Çevrimsiz Grafikler yaklaşımı, vekil değişkenin yalnızca gözlemlenemeyen değişkenle olan yüksek korelasyonuna değil, onun yapısal özelliklerine de kapsamlı biçimde dikkat edilmesini salık vermektedir.

🔖

Cite This Article

Vera-sánchez, F. (2025). ECONOMETRIC MODEL SPECIFICATION AND D-SEPARATION: A MONTE CARLO EXPERIMENT. Economics Business and Organization Research, Cilt: 7 Sayı: 1, 21–39. https://doi.org/10.66414/ebor.100346

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (38)

  1. Aigner, D. J. (1974) MSE dominance of least squares with errors of observation. Journal of Econometrics, 2(4), 365-372. https://doi.org/10.1016/0304-4076(74)90020-7
  2. Bellemare, M. F., Masaki, T., & Pepinsky, T. B. (2017). Lagged explanatory variables and the estimation of causal effect. Journal of Politics, 79(3), 949-963. https://doi.org/10.1086/690946
  3. Bollinger, C. R., & Minier J. (2015). On the robustness of coefficient estimates to the inclusion of proxy variables. Journal of Econometric Methods, 4(1), 101-122. https://doi.org/10.1515/jem-2012-0008
  4. de Luna, X., Fowler, P., & Johansson, P. (2017). Proxy variables and nonparametric identification of causal effects. Economics Letters, 150, 152-154. https://doi.org/10.1016/j.econlet.2016.11.018
  5. Fé, E. (2012). Instrumental variable estimation of heteroskedasticity adaptive error component models. Statistical Papers, 53(3), 577-615. https://doi.org/10.1007/s00362-011-0366-5
  6. Galvao, A. F., Montes-Rojas, G., & Song, S. (2017). Endogeneity bias modeling using observables. Economics Letters, 152, 41-45. https://doi.org/10.1016/j.econlet.2016.12.021
  7. Gauger, J. (1989). The generated regressor correction: Impacts upon inferences in hypothesis testing. Journal of Macroeconomics, 11(3), 383-95. https://doi.org/10.1016/0164-0704(89)90065-7
  8. Ghosh, P., Grier, K., & Kim, J. (2021). Heterogeneous endogeneity. Statistical Papers, 62(2), 847-886. https://doi.org/10.1007/s00362-019-01116-9
  9. Greene W. H. (2007). Econometric analysis (6th edition). Prentice Hall.
  10. Griffiths, W. E., Hill, R. C., & Judge, G. G. (1993). Learning and practicing econometrics. John Wiley & Sons.
  11. Hu, Y., & Sasaki, Y. (2018). Closed-form identification of dynamic discrete choice models with proxies for unobserved state variables. Econometric Theory, 34(1), 166-185. https://doi.org/10.1017/S0266466617000081
  12. Hu, Y., & Schennach, S. M. (2008). Instrumental variable treatment of nonclassical measurement error models. Econometrica, 76(1), 195-216. https://doi.org/10.1111/j.0012-9682.2008.00823.x
  13. Kakimoto, S., & K. Ohtani. (1985). On the use of proxy variable in the test of homoscedasticity. Economic Letters, 18(2-3), 153-156. https://doi.org/10.1016/0165-1765(85)90171-5
  14. Kennedy, P. (1998). A guide to econometrics (4th edition). MIT Press.
  15. Kim, K. I., & Petrin, A. (2022). A generalized non-parametric instrumental variable-control function approach to estimation in nonlinear settings. Journal of Econometric Methods, 11(1), 91-125. https://doi.org/10.1515/jem-2021-0038
  16. Krasker, W. S., & Pratt, J. W., (1986). Bounding the effects of proxy variables on regression coefficients. Econometrica, 54(3), 641-655. https://doi.org/10.2307/1911312
  17. Kuroki M., & Pearl J. (2014). Measurement bias and effect restoration in causal inference. Biometrika, 101(2), 423-437. https://doi.org/10.1093/biomet/ast066
  18. Leamer, E. E. (1978). Specification searches ad hoc inference with nonexperimental data. John Wiley & Sons.
  19. McCallum, B. T. (1972). Relative asymptotic bias from errors of omission and measurement. Econometrica, 40(4), 757-758. https://doi.org/10.2307/1912970
  20. Murphy, K., & Topel, R. (1985). Estimation and inference in two step econometric models. Journal of Business & Economic Statistics, 3(4), 370-379. https://doi.org/10.1080/07350015.1985.10509471
  21. Namba, A., & Xu, H. (2018). PMSE dominance of the positive-part shrinkage estimator in a regression model with proxy variables. Journal of Statistical Computation & Simulation, 88(15), 2893-2908. https://doi.org/10.1080/00949655.2018.1491576
  22. Ohtani, K. (1985). A note on the use of a proxy variable in testing hypotheses. Economics Letters, 17(1-2), 107-110. https://doi.org/10.1016/0165-1765(85)90137-5
  23. Oxley, L., & McAleer, M. (1993). Econometric issues in macroeconomic models with generated regressors. Journal of Economic Surveys, 7(1), 1-40. https://doi.org/10.1111/j.1467-6419.1993.tb00158.x
  24. Pagan, A. R. (1984). Econometric issues in the analysis of regressions with generated regressors. International Economic Review, 25(1), 221-247. https://doi.org/10.2307/2648877
  25. Papadopoulos, A. (2022). Accounting for endogeneity in regression models using copulas: A step-by-step guide for empirical studies. Journal of Econometric Methods, 11(1), 127-154. https://doi.org/10.1515/jem-2020-0007
  26. Pearl J. (2009). Causality: Models, reasoning and inference (2nd edition). Cambridge University Press.
  27. Pearl, J. (1995). Causal diagrams for empirical research. Biometrika, 82(4), 669-688. https://doi.org/10.1093/biomet/82.4.669
  28. Pearl, J. (1996). The art and science of cause and effect (Technical Report No. R-248). Cognitive Science Laboratory, University of California, Los Angeles.
  29. Pindyck, R. S., & Rubinfeld, D. L. (1991). Econometric models and economic forecasts. McGraw Hill.
  30. Rothenhäusler, D., Meinshausen, N., Bühlmann, P., & Peters, J. (2021). Anchor regression: Heterogeneous data meet causality. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 83(2), 215-246. https://doi.org/10.1111/rssb.12398
  31. Sajons, G. B. (2020). Estimating the causal effect of measured endogenous variables: A tutorial on experimentally randomized instrumental variables. Leadership Quarterly, 31(5), 101348. https://doi.org/10.1016/j.leaqua.2019.101348
  32. Scheines, P., Spirtes, P., Glymour, C., & Meek. C. (1994). Tetrad II user’s manual and software. Lawrence Erlbaum Associates, Inc.
  33. Shingaki, R., Kanda, H., & Kuroki, M. (2021). Selection and integration of generalized instrumental variables for estimating total effects. Statistical Papers, 62(5), 2355-2381. https://doi.org/10.1007/s00362-020-01190-4
  34. Sperrin, M., & Martin, G. P. (2020). Multiple imputation with missing indicators as proxies for unmeasured variables: simulation study. BMC Medical Research Methodology, 20(1), 185. https://doi.org/10.1186/s12874-020-01068-x
  35. Spirtes, P. (1995). Directed cyclic graphical representations of feedback models. In P. Besnard & S. Hanks (Eds.), Proceedings of the Eleventh Conference on Uncertainty in Artificial Intelligence (pp. 491-498). Morgan Kaufmann Publishers.
  36. Spirtes, P., Glymour, C., & Scheines. R. (1993). Causation, prediction, and search. Springer. https://doi.org/10.1007/978-1-4612-2748-9
  37. Stanghellini E. (2004). Instrumental variables in Gaussian directed acyclic graph models with an unobserved confounder. Environmetrics, 15(5), 463-469. https://doi.org/10.1002/env.676
  38. Wickens, M. R. (1972). A note on the use of proxy variables. Econometrica, 40(4), 759-761. https://doi.org/10.2307/1912971