Economic Growth Second Edition
Results under Isoelastic Utility
Download 0.79 Mb. Pdf ko'rish
|
BarroSalaIMartin2004Chap1-2
2.7.4
Results under Isoelastic Utility In the standard analysis, where φ(t −τ) = 0 for all t, consumption is not a constant fraction of wealth unless θ = 1. However, we know, for any value of θ, that the first-order condition for consumption growth at time τ is given from equation (2.11) by ˙c c (τ) = (1/θ) · [r(τ) − ρ] (2.72) A reasonable conjecture is that the form of equation (2.72) would still hold when φ(t −τ) = 0 but that the constant ρ would be replaced by some other constant that represented the effective rate of time preference. This conjecture is incorrect. The reason is that the effective rate of time preference at time τ involves an interaction of the path of the future values of φ (t − τ) with future interest rates and turns out not to be constant when interest rates are changing except when θ = 1. Although the transitional dynamics is complicated, it is straightforward to work out the characteristics of the steady state. The key point is that, in a steady state, an increase in household assets would be used to raise consumption uniformly in future periods. This property makes it easy to compute propensities to consume for future periods with respect to current assets and, therefore, makes it easy to find the first-order optimization condition for current consumption. Only the results are presented here. In the steady state, the interest rate is given by r ∗ = x + n + 1/ (2.73) Growth Models with Consumer Optimization 131 where the integral is now defined by ≡ ∞ 0 e −{[ρ−x·(1−θ)−n]·v+φ(v)} d v (2.74) Thus, if φ(v) = 0, we get the standard result r ∗ = ρ + θx For the case of Laibson’s quasi-hyperbolic utility function in equation (2.64), the result turns out to be r ∗ ≈ ρ β − n · (1 − β) β + x · (β + θ − 1) β (2.75) where recall that 0 < β < 1. Thus, for the case considered before of log utility (θ = 1), the effect of x on r ∗ is one-to-one. More generally, the effect of x on r ∗ is more or less than one-to-one depending on whether θ is greater or less than 1. For the transitional dynamics, Barro (1999) shows that consumption growth at any date τ satisfies the condition ˙c c (τ) = (1/θ) · [r(τ) − λ(τ)] (2.76) The term λ(τ) is the effective rate of time preference and is given by λ(τ) = ∞ τ ω(t, τ) · [ρ + φ (t − τ)] dt ∞ τ ω(t, τ) dt (2.77) where ω(t, τ) > 0. Thus, λ(τ) is again a weighted average of future instantaneous rates of time preference, ρ + φ (t − τ). The difference from equation (2.62) is that the weighting factor, ω(t, τ), is time varying unless θ = 1. Barro (1999) shows that, if θ > 1, ω(t, τ) declines with the average of interest rates between dates τ and t. If the economy begins with a capital intensity below its steady-state value, r (τ) starts high and then falls toward its steady-state value. The weights ω(t, τ) are then particularly low for dates t far in the future. Since these dates are also the ones with relatively low values of ρ + φ (t − τ), λ(τ) is high initially. However, as interest rates fall, the weights, ω(t, τ), become more even, and λ(τ) declines. This descending path of λ(τ) means that households effectively become more patient over time. However, the effects are all reversed if θ < 1. The case θ = 1, which we worked out before, is the intermediate one in which the weights stay constant during the transition. Hence, in this case, the effective rate of time preference does not change during the transition. |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2025
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling