Economic Preferences In Developing Countries: Intergenerational Transmission And The Role Of Cognitive Ability

This paper constitutes the first attempt to empirically test for the determi- nants of individual’s risk and time preferences through four possible mechanisms in [r]

ECONOMIC PREFERENCES IN DEVELOPING COUNTRIES:

INTERGENERATIONAL TRANSMISSION AND THE ROLE OF COGNITIVE

ABILITY

Yanjun Li Anh Nguyet Tran T

Abstrac

Recent theories endogenize individual’s economic preferences, postulated that

they are transmitted from parents and role models, as well as related to

individual’s cognitive ability This paper constitutes the first attempt to

empirically test for the determi- nants of individual’s risk and time

preferences through four possible mechanisms in the context of

developing countries: (1) direct transmission of preferences from parents

to children; (2) positive assortative matching of spouses, which improves

the efficiency of the transmission process; (3) the role of local

environment on prevailing children’s preferences; (4) children’s cognitive

capacity on increasing willingness to take risks in advantageous situations

and be more patient over time horizon Using the longitudi- nal data from

the Indonesia Family Life Survey, we find strong, persistent empirical

support for all four mechanisms over a test-retest period of seven years A

more thor- ough investigation of the underlying mechanisms further

suggests that gender plays a key role in the transmission process, so as

various family background factors We also provide collateral evidence

that the measured risk and time preferences are strongly associated with a

wide range of individual’s outcomes Among them, risky choices such as

smoking and child marriage, as well as career choices (self-employment) are particularly crucial issues in Indonesia

Keywords Risk preferences; Trust; Intergenerational transmission; Cultural eco- nomics; Family economics; Assortative mating; Social interactions; IFLS; Development economics JEL classification D1, D8, J12, J13, J62, O12, Z13

Extended Abstract

1 Introduction and motivation

Preferences, beliefs, and attitudes that guide personal behaviors are partially formed based on heritable genetic factors, and partially passed across generations by social learn- ing (Bisin and Verdier, 2010); parents are deeply involved throughout the whole process Preferences transmitted from parents to their offspring may include many fundamental ones, such as risk or time preferences These two dimensions of individual preferences have been empirically shown to have a significant impact on

a wide range of essential choices and behaviors, including personal habits ( ubi cka

et al., 2001), educational attainment (Castillo et al., 2011; Golsteyn et al., 2014), occupation and labor market outcomes (Heck- man et al., 2006; Deming, 2017), financial behavior (Agarwal and Mazumder 2013; Lusardi and Mitchell 2014), etc., and thus are of great value to researchers and policymakers

This paper examines the intergenerational transmission of time and risk preferences

in Indonesia As of our best knowledge, this paper is the first attempt to investigate the transmission of risk and time preferences in the context of developing countries Previous studies relating to the transfer of these traits focus explicitly on rich, western countries (Kimball et al., 2009; Dohmen et al., 2012; Alan et al., 2017)

However, as our results demonstrate below, the intergenerational transmission of risk attitudes and impatience in the setting of a developing country leads to several interesting and meaningful implications We provide evidence on three distinct aspects of the preference transmission process First, we analyse the preferences transmission from parents to children using direct mea- sures of risk and time preferences for individuals and their parents Second, we test for another aspect of the transmission process through assortative matching of parental prefer- ences Within this framework, we strengthen the findings by controlling for fixed effects of local environment on individual?s preferences The theoretical models by Bisin and Verdier (2000) and Bisin et al (2004) predict that spouses may engage in strategic, assortative matching in order to more efficiently transmit their preferences/attitudes

to offspring

Finally, we address a recent question on the relationship between cognitive ability and economic preferences: whether risk and time preferences are related to individual cognitive skills in the context of developing countries As before, from our best knowledge, this is the first study on the preference transmission while simultaneously investigating the cognition-preference gradient in a developing country

2 Results

Our results indicate a strong evidence of the three transmission mechanisms of risk and time preferences - parents to children, local environment to individuals, and parental assor- tative matching; as well as the positive relation from higher cognition

to higher willingness to take risks (in advantageous situation such as winning a lottery), and more patience The first main result is that time and risk preferences are strongly positively correlated between parents and children Speaking differently, offspring of parents who are more willing to take risks and more patient develop similar preferences to their parents This result is evident over time (both in 2007 and 2014) These correlations are essentially strongly significant when we control

for further individual characteristics (gender, age, and cognitive ability) A second finding is that, the magnitude of parents-child preference transmission lowers by almost by a half once we control for the fixed effects of residential communities This is true for both time and risk preferences and over time (in 2007 and in 2014) These results suggest that local environment (and therefore, average preferences at the local level) play a role in shaping individual preferences, apart from parental effects

The third main finding is a strong pattern of positive correlation of risk and time preferences among spouses (i.e., parents of the respondents), consistent with the theoretical models on assortative matching We find a comparably large correlation between spouses’ preferences, again qualitatively unchanged over a period of seven years Combined with the previous findings that both mother’s and father’s preferences are transmitted to children, this result of assortative matching strongly is highly consistent with the implications of attitude transmission theories

Finally, as we already emphasize, our study also speaks to the relationship between cognitive ability and risk preference and impatience in the context of developing countries Our results suggest that both risk and time preferences vary systematically with cognitive ability, conditional on parental transmission of preferences, local environment, as well as individual characteristics For example, in 2007, 50 percent increase in cognitive ability are associated with 7.5 percent increase in individual’s willingness to take risks, and 10 percent increase in patience over the yearlong time horizon More remarkably, we obtain qualitatively similar results on the preferences-cognition gradient of the same sample of individuals after 7 years

3 Data description: the Indonesian Family Life Survey 1997-

2015

We use data from the Indonesian Family Life Survey (IFLS) conducted in 2007 (IFLS4) and 2015 (IFLS5) The IFLS is a household and community longitudinal

survey, and contains information on risk and time preferences of the respondents and their parents, and individual’s cognitive ability as well as other demographic backgrounds

3.1 Individual risk and time preferences

In these waves, the study administered a series of questions to elicit attitudes towards risk and time preference on all individuals aged 15 and older We use the information from these modules to construct the measures for respondents and their parents in 2007 and 2015.Both measures of risk preferences consist of a series of eight interdependent hypothetical binary choices between lotteries and safe financial rewards The first four pairs of lotteries give a higher expected payment for the riskier choice, whereas the latter four pairs of lotteries give a higher expected payment for the safer choice We took the number of risky decisions the respondents made as the measure of individual risk attitudes (Holt and Laury, 2002; Lonnqvist et al., 2015; Dohmen et al., 2018) The measure of time preferences is similarly derived from eight hypothetical binary choices between immediate and delayed payment in a “staircase” procedures (Cornsweet, 1962; Dohmen et al., 2018) In each of eight questions, participants, respondents decide between receiving a reward today or larger payments in one year and then in five years We took the number of delayed choices as the measure of individual’s patience (Dohmen et al., 2018)

3.2 Explanatory variable: cognitive ability

Starting from the IFLS4 in 2007, all respondents aged 15 or older were asked to take cognitive assessments, which were divided into two parts: (i) a set of four tasks adapted for the Indonesian population from the similar tests in the Health and Retirement Survey (HRS) in the U.S; (ii) an abridged version of the Ravens test These tests contain mea- sures of quantitative abstract reasoning ability and episodic memory (Ofstedal et al., 2005; McArdle et al., 2007; Strauss et al., 2016)

I use the test scores as manifestation measures of the unobserved fluid intelligence The unobserved cognitive factor is recovered using a measurement system widely used in the psychology and the economics literature on human capital development (for example, Bollen, 1989; Cunha and Heckman, 2008; Cunha et al., 2010; Agostinelli and Wiswall, 2018)

4 Tables and graphs

Child’s risk and time preferences as a linear function of parents’ preferences using the IFLS4 data in 2007 The graphs on the upper panel show children’s average scores of risky choices (standardized), i.e., willingness to take risks for a given level

of willingness to take risks of fathers (right graph) and of mothers (left graph) The lower graphs show children’s average scores of delayed rewards (standardized), i.e., patience, for a given level of father’s patience (left graph) and mother’s patience (right graph)

Child’s risk and time preferences as a linear function of parents’ preferences using the IFLS5 data in 2015 The graphs on the upper panel show children’s average scores of risky choices (standardized), i.e., willingness to take risks for a given level of willingness to take risks of fathers (right graph) and of mothers (left graph) The lower graphs show children’s average scores of delayed rewards (standardized), i.e., patience, for a given level of father’s patience (left graph) and mother’s patience (right graph)