Government spending and consumer attitudes toward risk, time preference, and intertemporal substitution: an econometric analysis.

• Author: Hatzinikolaou, Dimitris

1. Introduction

   During the past three decades, the size of government, measured by the share of government expenditure in GDP, has been growing in many countries, notably those in the OECD. The consequences of this phenomenon have not been fully assessed, however, and the issue is still fervently debated among economists and policy-makers.

   In this paper, we construct a model that considers the direct effects, if any, of government growth on the attitudes of a typical consumer toward risk, time preference, and intertemporal substitution in consumption. Our null hypothesis is that a growing government sector does not affect these attitudes, and the alternative is that it causes a typical consumer to become less risk averse, more impatient to consume now rather than in the future, and less responsive to changes in real interest rates. If the alternative hypothesis is correct, then government growth may cause economic growth to decline by causing private saving and market efficiency to decline. For if a consumer becomes less risk averse, then he will insure less against uncertain future income by saving less. And if he becomes less willing to postpone consumption, then saving will fall at every given level of the real interest rate. As a result, current and future real interest rates will rise, thus lowering investment in physical capital. Productivity and long-run economic growth will decline. To our knowledge, no formal model exists that examines these effects, although previous studies, e.g., Bean [4], explicitly consider the effects of government expenditure on consumer behavior.

   We use an intertemporal utility maximization model to address the issue. Following Bean [4], we include government expenditure in the utility function but not in the budget constraint of a typical consumer, i.e., the model is non-Ricardian.(1) Using Greek annual aggregate data, 1960-1990, we obtain reasonable parameter estimates that lead to the rejection of the null hypothesis. Our motivation to use Greek data comes from the spectacular growth of the size of government during the 1980s, which is widely believed to have caused the recent macroeconomic imbalances [19].(2) It also comes from the suggestion of the OECD [18, 38] that fast-growing government transfers may have caused a "rentier mentality" in the Greek society, and may have created a "climate of complacency, dissociating income from work effort, reducing work incentives and favouring consumption." We present the model in section II and the empirical results in section III. Section IV concludes the paper.

2. The Model

   The Consumer's Problem

   We consider a representative consumer who maximizes a multi-period utility function. Having taken all the relevant constraints into account and having formed his expectations about the future, he plans current and future consumption expenditure at the beginning of each period. Government spending on goods, services, and welfare schemes directly affects consumer behavior, since the consumer derives utility from public goods, services, and transfers in kind. For example, public roads, parks, schools, libraries, and hospitals satisfy some of the consumer's wants. We assume that although he chooses the level of consumption by participating in free markets, he takes government expenditure as given. We also assume that he takes leisure as given.(3) We start our analysis with the one-period utility function

   u(c,l,g) = (

   , (1)

   where c is real consumption expenditure per worker, l is leisure time per worker, g is real government expenditure per worker, and [Alpha], [Beta], and [Gamma] are parameters. This utility function is a simple representation of preferences that are nonseparable in consumption, leisure, and government expenditure, implying that the consumer considers a geometric weighted average of these variables. If this weighted average is thought of as a composite commodity, then (1) is a constant relative risk aversion utility function, provided that [Gamma] [less than] 1. Moreover, this specification leads to an econometrically tractable estimating equation that involves the growth rates of c, l, and g, which are likely to be stationary. Note that if [Alpha] [greater than] 0, [Beta] [greater than] 0, and 1 - [Alpha] - [Beta] [greater than] 0, then this utility function is strictly increasing in c, l, and g (nonsatiation), and if [Alpha] [greater than] 0 and 1 - [Alpha][Gamma] [greater than] 0, then it is strictly concave in the choice variable c. Note also that if [Gamma] [greater than] 0 ([Gamma] [less than] 0), in addition to the nonsatiation restrictions, then consumption, leisure, and government expenditure are pairwise Edgeworth complements (substitutes).(4) Clearly, if 1 - [Alpha] - [Beta] = 0, then consumer's behavior will not directly depend on government expenditure. Assuming that an increase in g, other things equal, does not decrease utility, the statistical hypothesis we aim to test is [H.sub.0]: 1 - [Alpha] - [Beta] = 0 against [H.sub.1]: 1 - [Alpha] - [Beta] [greater than] 0.

   In each period t, the consumer maximizes the lifetime utility function

   [Mathematical Expression Omitted]

   subject to the lifetime budget constraint

   [Mathematical Expression Omitted],

   where [E.sub.t] is the mathematical expectation conditional on information available at time t; N is the length of the consumer's planning horizon; [Delta] is a subjective discount rate, assumed to be a positive constant; [a.sub.t] is the consumer's real financial wealth (measured in consumption units) at the beginning of period t; T is the time endowment (in hours); [w.sub.t] is the hourly real wage (in consumption units); [r.sub.t] is a real interest rate; and T - [l.sub.t] = [L.sub.t] is work effort in period t. We assume time-separable preferences because we use annual data to estimate the model, so substitutability between goods in different periods is limited [14, 237]. Equation (3) says that the sum of current wealth and current and (discounted) future dissaving is zero, implying that the consumer leaves no bequests. Using Hall's [6, 986] perturbation argument, the first-order condition is given by(5)

   [(1+[Delta]).sup.-1][E.sub.t][(1 + [

   .sup.[Beta][Gamma]][([g.sub.t+1]/[g.sub.t]).sup.(1-[Alpha]-[Beta])[Gamma]]] = 1. (4)

   To interpret equation (4), we first calculate from equation (2) the marginal rate of substitution

   [Mathematical Expression Omitted].

   Equation (4) is the intertemporal efficiency condition, or Euler equation, for consumption. It says that when the consumer is at the optimum, he cannot become better off by transferring one unit of consumption from time t to time t + 1 (by saving) or from time t + 1 to time t (by dissaving). It requires that at the optimum the conditional expectation of the marginal benefit from the consumption of an additional unit in period t + 1 be equal to its marginal cost. The former is measured by [Mathematical Expression Omitted] [given by equation (5)], and the latter is measured by 1/(1 + [r.sub.t]).(6) Equation (4) can be expected to hold regardless of labor market constraints, as long as quantity constraints are either not binding or absent from the goods and capital markets, so that intertemporal substitution in consumption is feasible.

   Behavior toward Risk

   We measure risk aversion by the insurance or risk premium, [Phi], which is the amount of private consumption that the consumer would give up to avoid a fair gamble. If he is a risk averter, then [Phi] is positive. To derive an expression for [Phi], let [Epsilon], [Zeta], and [Eta] be the nonsystematic components of c, l, and g, respectively, and assume that E([Epsilon]) = E([Zeta]) = E([Eta]) = 0. Assume also that [Epsilon], [Zeta], and [Eta] take on small values, so that moments higher than their variances, denoted by [Mathematical Expression Omitted], [Mathematical...