Economic Growth Second Edition
Household Heterogeneity
Download 0.79 Mb. Pdf ko'rish
|
BarroSalaIMartin2004Chap1-2
2.6.7
Household Heterogeneity Our analysis thus far has considered a single household as representing the entire economy. The consumption and saving decisions of the representative agent are supposed to capture the behavior of the average agent in a complex economy with many families. The important question is whether the behavior of this “representative” or “average” household is really equivalent to what we would get if we averaged the behavior of many heterogeneous families. Growth Models with Consumer Optimization 119 Caselli and Ventura (2000) have extended the Ramsey model to allow for various forms of household heterogeneity. 28 Following their analysis, we assume that the economy contains J equal-sized households, each of which is an infinitely lived dynasty. The population of each household—and, therefore, the overall population—grow at the constant rate n. Pref- erences of each household are still given by equations (2.1) and (2.10), with the preference parameters ρ and θ the same for each household. In this case, it is straightforward to allow for differences across households in initial assets and labor productivity. Let a j (t) and π j represent, respectively, the per capita assets and productivity level of the j th household. The wage rate paid to the j th household is π j w, where w is the economy- wide average wage, π j is constant over time, and we have normalized so that the mean value of π j equals 1. The flow budget constraint for each household takes the same form as equation (2.3): ˙a j = π j · w + ra j − c j − na j (2.45) In this representation, each household could have a different value of initial assets, a j (0). The optimal growth rate of each household’s per capita consumption satisfies the usual first-order condition from equation (2.9): ˙c j /c j = (1/θ) · (r − ρ) (2.46) The household’s level of per capita consumption can be found, as in the analysis of the first section of this chapter, by solving out the differential equation for c j and using the transver- sality condition (of the form of equation [2.12]). The result, analogous to equation (2.15), is c j = µ · (a j + π j ˜ w) (2.47) where µ is the propensity to consume out of assets (given by equation [2.16]) and ˜w is the present value of the economy-wide average wage. The economy-wide value of per capita assets is a = ( 1 J )· J 1 a j , and the economy-wide value of per capita consumption is c = ( 1 J ) · J 1 c j . Since the population growth rate is the same for all households, aggregation is straightforward: sum equation (2.45) over the J households and divide by J to compute the economy-wide budget constraint: ˙a = w + ra − c − na (2.48) This budget constraint is the same as equation (2.3). 28. Stiglitz (1969) worked out a model with household heterogeneity under a variety of nonoptimizing saving functions. 120 Chapter 2 We can also aggregate the consumption function, equation (2.47), across households to get the economy-wide value of consumption per person: c = µ · (a + ˜w) (2.49) This relation is the same as equation (2.15). Finally, we can use equations (2.48) and (2.49) to get ˙c/c = (1/θ) · (r − ρ) (2.50) which is the standard economy-wide condition for consumption growth. When combined with the usual analysis of competitive firms, this description of aggregate household behavior—equations (2.48) and (2.50)—delivers the standard Ramsey model. Hence, the model with the assumed forms of heterogeneity in initial assets and worker productivity has the same macroeconomic implications as the usual, representative-agent model. In other words, if the households in the economy differ in their level of wealth or productivity, and if their preferences are CIES with identical parameters and discount rates, the average consumption, assets, income, and capital for these families behave exactly as the ones of a single representative household. Hence, the representative-agent model provides the correct description of the average variables of an economy populated with the assumed forms of heterogenous agents. Aside from supporting the use of the representative-agent framework, the extension to include heterogeneity also allows for a study of the dynamics of inequality. Equation (2.46) implies that each household chooses the same growth rate for consumption. Therefore, relative consumption, c j /c, does not vary over time. The model does imply a dynamics for relative assets, a j /a. Equations (2.45), (2.47), (2.48), and (2.49) imply that relative assets change in accordance with d dt a j a = (w − µ ˜w) a · π j − a j a (2.51) We can show that, in the steady state (where w grows at the rate x and r = ρ + θx), the relation w = µ ˜w holds. Therefore, relative asset positions stay constant in the steady state. Outside of the steady state, equation (2.51) implies that the relative asset position does not change over time for a household whose relative labor productivity, π j , is as high as its relative asset position, a j /a. For other households, the behavior depends on the sign of w − µ ˜w. Imagine that w > µ ˜w. Roughly speaking, this condition says that the propensity to save out of (permanent) wage income is positive. In this case, equation (2.51) implies that a j /a would rise or fall over time depending on whether relative labor productivity exceeded or fell short of the relative asset position— π j >(or <) a j /a. Thus a convergence Growth Models with Consumer Optimization 121 pattern would hold, whereby relative assets moved toward relative productivity. However, the opposite pattern applies if w < µ ˜w. Outside of the steady state, the sign of w −µ ˜w depends on the relation of interest rates to growth rates of wages and is ambiguous. Hence, the model does not have clear predictions about the way in which a j /a will move along the transition. Caselli and Ventura (2000) also allowed for a form of heterogeneity in household pref- erences. They assumed that preferences involved the felicity function u (c + β j g ), where they interpret g as a publicly provided service. The parameter β j > 0 indicates the value that household j attaches to the public service. The variable g could also represent the ser- vices that households get freely from the environment, for example, from staring at the sky. The main result from this extension is that the aggregation of individual behavior still corresponds to a representative-agent model, in the sense that the economy-wide average variables, a and c, evolve as they would with a single agent who had average values of initial assets, labor productivity, and preferences. In this sense, the results from the Ramsey model are robust to this extension to admit heterogeneous preferences. Download 0.79 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2025
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling