Anticipation, learning and welfare: the case of distortionary taxation

Abstract

We study the impact of anticipated fiscal policy changes in the Ramsey economy when agents form expectations using adaptive learning. We extend the existing framework by distortionary taxes as well as elastic labour supply, which makes agents' decisions non-predetermined but more realistic. We detect that the dynamic responses to anticipated tax changes under learning have oscillatory behaviour. Moreover, we demonstrate that this behaviour can have important implications for the welfare consequences of fiscal reforms. JEL Classification: E32, E62, D84Keywords: Fiscal Policy, Adaptive Learning, Oscillations c 2011 by Emanuel Gasteiger and Shoujian Zhang. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including c notice, is given to the source.$ We are indebted to Alejandro Cuñat, Robert Kunst and Gerhard Sorger for sound supervision and patience. We are also grateful to Szabolcs Deák, Seppo Honkapohja, Torsten Lisson, Paul Pichler and Monika Turyna as well as the participants at the 4th RGS Doctoral Conference in Economics, the 2011 Annual Conference of the Scottish Economic Society and the Quantitative Economics Doctorate (QED) Jamboree 2011 for valuable comments. All remaining errors are the responsibility of the authors. Leeper et al. (2009) is another good example of empirical evidence of fiscal foresight. Therein they also demonstrate the challenges for econometricians that aim to quantify the impact of fiscal policy actions and at the same time account adequately for fiscal foresight. In addition, our second key contribution is to numerically examine fiscal policy reforms, in the presence of several tax instruments.Note that there are fundamental differences between lump-sum taxation and distortionary taxation: a labour income tax under inelastic labour supply does not affect household margins and therefore causes no distortion, but under elastic labour supply the labour income tax affects the intra-temporal choice between consumption and leisure of the household and may cause an intra-temporal distortion. Next, a capital income tax has the potential to cause up to two types of distortion. First, the capital income tax in any case affects the inter-temporal household Euler equation. In case of elastic labour supply, the capital income tax also affects the intra-temporal choice between consumption and leisure of the household due to its distortion of the consumption choice. Finally, a consumption tax may also cause an inter-temporal distortion by affecting the household Next to the analytical derivations, we also calibrate our model and calculate welfare consequences for several policy experiments under perfect foresight as well as under learning. For this purpose, we make use of the welfare measure proposed by Lucas (1990) and also applied by Cooley and Hansen (1992) (for discrete time), which takes into account the whole transition path between the initial and new steady-states associated with initial ...

Adaptive Learning, Social Security Reform, and Policy Uncertainty

“…Following the decrease in the population growth rate, the adaptive learning agents update their expectations and extrapolate the change they have observed forward. This leads the capital stock to fall below the new steady state and then converge with an oscillatory pattern, consistent with the related literature (e.g., Mitra, Evans, and Honkapohja 2013, Gasteiger and Zhang 2014). These fluctuations reflect periods of optimism or pessimism followed by subsequent corrections.…”

“…The endogenous cycles are larger in amplitude and magnitude when the gain parameter is smaller. A similar pattern is observed in Gasteiger and Zhang (2014), Mitra, Evans, and Honkapohja (2013), and Evans, Guesnerie, and McGough (2019) which are focused on permanent changes to fiscal policy. In contrast, in much of the learning literature the changes in models are transitory shocks, and so a smaller gain parameter leads to smaller swings in aggregate variables.…”

“…4. See also Honkapohja (2013, 2019), Gasteiger andZhang (2014), andCaprioli (2015). The adaptive learning model developed in this paper is a special case of Finite Horizon Learning in an overlapping generations model as developed in Cottle Hunt (2019), based on Branch, Evans, and McGough (2013).…”

Animal Spirits and Fiscal Policy

“…This argument is strongly supported by the fact that empirical evidence of agents' heterogeneity and limited cognitive abilities has been provided by using both laboratory and survey data (see Carroll, 2003;Branch, 2004;Pfajfar and Santoro, 2010;Froot, 1987, 1988;Hommes, 2011). Still, researchers have started incorporating behavioral economics elements in dynamic macro models only recently (see Driscoll and Holden, 2014), and this is especially true in relation to fiscal policy (see Evans et al, 2009Evans et al, , 2012Gasteiger and Shoujian, 2014;Caprioli, 2015;Gabaix, 2016). Furthermore, some contributions have highlighted how the linearity implied by rational expectations DSGE models makes them not fully suitable for fiscal policy analyses.…”

Learning and the Size of the Government Spending Multiplier

“…An interesting result is that their learning model can generate a hump-shaped response in consumption. Gasteiger and Zhang (2014) study the impact of fiscal policy in a deterministic version of the RBC model with distortionary taxation.…”