Dealing with adaptive expectations in Stock-Flow consistent models

Abstract

Stock-flow consistent (SFC) models, in their simulations of the macroeconomic dynamics, have not properly dealt with how households form their expectations about future income flows. Although the majority of SFC models assume some type of adaptive expectations, the literature has not provided so far guidelines about how to best model expected variables in general, and expect flows of income in consumption decisions in particular, thus not paying enough attention to the role of expectations in affecting the path of variables and the steady-state composition of stocks. In this article, which compares different types of modeling strategies for income expectations, it is shown that minor differences in the way that expectations are modeled may have important macroeconomic consequences, mainly in terms of the traverse and the steady-state composition of wealth.

Keywords:

• Stock-flow consistent models
• income expectations
• consumption
• wealth composition

JEL CLASSIFICATION:

• B50
• E12
• E21
• E27
• C63

Notes

1 It should be noted that Burgess et al. (2016) is a paper elaborated for the Bank of England. The authors regard their model as a “useful addition to the set (‘suite’) of models called upon by policy makers to help them in their decision making” (p. 2). As will be shown later, their formulation of households’ expectations biases the values of the steady-state norms and, as a consequence, their forecasting conclusions and any policy decided upon using their exercises. It is an example of what can happen if not enough attention is paid to a seemingly inconsequential component of an SFC model, such as income expectations of households’ consumption decisions.

2 These results are straightforward. One only needs to reformulate the equations in terms of x and x^e. For example, the formulation of Caiani will be $x_t^e=\left(1\hspace{0.17em}-\hspace{0.17em}\epsilon \right)x_{t\hspace{0.17em}-\hspace{0.17em}1}^e+x_{t\hspace{0.17em}-\hspace{0.17em}1}.$ Following the same approach with the remaining equations, the mathematical equivalence between them is easily verifiable.

3 Considering the actual parameters, in Le Heron (2011) it is 0.1. For Caiani et al. (2016) it is 0.25. Finally, for Zezza (2008) the value is 0.75.

4 We thank an anonymous referee for drawing our attention to this point.

5 This model was used because it deals with a growing economy, thus providing a valid test for the use of expectations, and also because its software codes are freely and publicly available on the internet at http://gennaro.zezza.it/software/eviews/gl2006.php

6 If we had used Caiani et al.’s formulation, these parameter values would be, respectively, 1.8. 1.5, 1.1, 0.9, 0.5, and 0.2. For Le Heron’s formulation, the equivalent values are, respectively, 0.8, 0.5, 0.1, −0.1, −0.5, and −0.8.

Additional information

Notes on contributors

Sylvio Antonio Kappes

Sylvio Antonio Kappes, ABD in the Graduate Program in Economics, Federal University of Rio Grande do Sul.

Marcelo Milan

Marcelo Milan, Assistant Professor (tenured) in the Department of Economics and International Relations and in the Graduate Program in Strategic International Studies, Federal University of Rio Grande do Sul.