Assessing the Impacts of Non-Ricardian Households in an Estimated New Keynesian DSGE Model

Using Bayesian maximum likelihood and data for Portugal, I estimate a New Keynesian DSGE model allowing for the presence of non-Ricardian households and test the stability of the model's prediction when the fraction of liquidity-constrained households changes. In particular, I assess the impacts on: (i) the model parameters posterior distributions; (ii) the impulse responses to six types of structural shocks; and (iii) the sources of fluctuations in output, inflation and the nominal interest rate. The first interesting result is the estimated share of non-Ricardian households in the Portuguese economy, which is found to be relatively high (58%). Even under a simplistic model economy, this result seems plausible and in line with Campbell and Mankiw (1989) for the US (50% of households estimated to be liquidity-constrained) but slightly higher than for other European countries and the euro area (between 25% and 37%). I also show that different-even if relatively close-shares of non-Ricardian households provide very distinct estimates of several parameters, and uneven results and interpretations. Impulse responses to consumption preference and productivity shocks are more amplified for lower shares of liquidity-constrained households; whereas for greater proportions, the model predicts more noticeable responses to price markup and government spending shocks. Fluctuations in output growth are mainly driven by productivity shocks for a lower share of rule-of-thumb consumers and by price markup shocks in the opposite scenario. Furthermore, the presence of a high proportion of non-Ricardian households and a high degree of price stickiness makes the Taylor-type interest rate rule solution locally indeterminate as in Galí et al. (2007).