Rate of Time Preference, Intertemporal Elasticity of Substitution, and Level of Wealth

Abstract

The rate of time preference (RTP) and the intertemporal elasticity of substitution (IES) are two important factors shaping intertemporal consumption decisions. Models in which the RTP and/or the IES differ systematically between rich and poor households have different empirical and policy implications for economic development, growth, and the distribution of income and consumption from those of standard models in which these parameters are constant across households. In this chapter, we estimate a model in which both RTP and IES are allowed to differ across rich and poor households using household level panel data from India. Our empirical results are consistent with the view that the RTP is constant across poor and rich households, but the IES is larger for the rich than it is for the poor.

The original article first appeared in the Review of Economics and Statistics 79, 564–814, 572. A newly written addendum has been added to this book chapter.

Appendices

Appendix

We use food including milk, sweets, and spices as the measure of food consumption. For nonfood consumption, we subtracted food and ceremonial expenses from total consumption expenditure. Ceremonial expenses are removed because they often jump from zero to large amounts. Nonfood consumption consists of narcotics, tea, coffee, tobacco, pan, and alcoholic beverages; clothing, sewing of cloth, other tailoring expenses, thread, needles, chap pals and other footwear etc.; travel and entertainment; medicines, cosmetics soap, barber service; electricity, water charges and cooling fuels for household use; labor expenses for domestic work; edible oils and fats (other than gee); and others, including complete meals in hotels, school and educational materials, stamps, stationery, grinding and milling charges, etc. Unfortunately, the ICRISAT consumption data do not include housing and transportation, because the market values of these categories of consumption are hard to measure in these villages. Total consumption expenditure is the sum of food and nonfood consumption.

To construct real consumption per male adult equivalent, nominal consumption at t is divided by the family size measure constructed by Townsend (1994) and the corresponding price index at t for each village. The price index for total consumption expenditure, food, and nonfood are the consumer price index, the price index for food, and the price index for nonfood, respectively. These real variables are valued at 1983 prices.

There are about forty households for each year in each of the three villages in the data. Some households drop out of the sample and others are added to the sample over years in the ICRISAT data. We exclude these households from our sample. There is one household in the village of Aurepalle with zero income in 1980. Because we take the log of income, this household is excluded. The number of households in our sample for the village of Aurepalle is 35; that for Shirapur, 33; and that for Kanzara, 36.

Addendum: Recent Developments

The main purpose of this chapter’s empirical work was to distinguish between the wealth-varying IES and RTP models in Sect. 3.⁹ It should be noted that the particular version of the wealth-varying IES model we study in this chapter is not a model of endogenous preferences: the IES varies with wealth because of the subsistence parameter in the model. On the other hand, the wealth-varying RTP model is a model of endogenous preferences. The empirical results of this chapter were more consistent with the wealth-varying IES model than with the wealth-varying RTP model.

We think that it is likely that the importance of consuming above the subsistence level dominates intertemporal behaviors for poor people. When a household is near the subsistence level, other considerations such as low or high real interest rates may be of much less importance. This view is consistent with our empirical results. After the publication of our paper, Ogaki and Zhang (2001) used data from Pakistani villages as well as the same data as we used for Indian villages. They found that the subsistence parameter is important in understanding risk sharing behaviors in these Indian and Pakistani villages.¹⁰

Bhatt and Ogaki (2012), a paper included as a chapter in this book, briefly discussed our empirical evidence. As they argue, it is possible that the RTP is wealth-varying for richer households as we focused on poor households in Indian villages. For example, consumption decisions of richer households in developed countries are not likely to be affected by the subsistence level considerations.