Abstract
This paper tests how subjects behave in an intertemporal consumption/saving experiment when borrowing is allowed and whether subjects treat debt differently than savings. Two treatments create environments where either saving or borrowing is required for optimal consumption. Since both treatments share the same optimal consumption levels, observed consumption choices can be directly compared across treatments. The experimental findings imply that deviations from optimal behavior are higher when subjects have to borrow than when they have to save in order to consume optimally, suggesting debt aversion. Signifiant under-consumption is observed when subjects have to borrow in order to reach optimal consumption. In line with previous experiments, weak evidence is found suggesting that subjects over-consume when saving is necessary for optimal consumption.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
An early reference is Hey and Dardanoni (1988). Noussair and Matheny (2000), Lei and Noussair (2002), Ballinger et al. (2003, 2011), Carbone and Hey (2004), Carbone (2006), Brown et al. (2009) as well as Carbone and Duffy (2013) further substantiate the experimental analysis of intertemporal consumption/saving problems.
See Duffy (2012) for an excellent survey of intertemporal consumption/savings experiments.
The experiment by Fenig et al. (2013) has treatments without borrowing constraints. They find that prohibiting borrowing for speculation in an experimental general equilibrium economy leads to increased precautionary saving and smaller asset price bubbles.
Brown et al. (2009) use an income stream that is increasing in expectations. However, their experiment also uses habit formation utility and therefore saving is optimal in early periods even with an increasing income stream.
Note that together with the ending condition \((a_{20}=0)\), this also implies that total consumption must equal total income: \(\sum ^{20}_{t=1}c_t=\sum ^{20}_{t=1}y_t\).
CARA utility was chosen because this class of utility functions is defined on the negative domain, i.e. allows negative utility. Why this is of importance will be explained later in this section. However, other papers in the literature on life-cycle consumption/savings problems also make use of CARA utility. See, for instance Carbone and Hey (2004).
See the appendix of Meissner and Rostam-Afschar (2014) for the detailed derivation of optimal consumption.
Subjects did not know about the role change when playing the first three repetitions. They received new instructions after the third round.
See supplementary material for the instructions and the quiz.
Note that consumption and savings are two sides of one coin: a subject who consumes too little also saves too much and vice versa.
I use the median as a descriptive statistic because I use non-parametric tests, such as the Mann–Whitney U Test to compare treatments. These tests typically provide information about the median of a random sample.
Note that for Measure 1 and 2, conditional optimal consumption is the only meaningful benchmark. With unconditionally optimal consumption as a benchmark this measure would always be zero, since \(\sum ^{20}_{t=1}c^*_t(w_t^*)=\sum ^{20}_{t=1}c_t\), where \(c^*_t(w_t^*)\) denotes unconditionally optimal consumption at period \(t\) as a function of optimal wealth \(w_t^*\), always holds.
Subjects may perceive the role change after round three to reveal new information about the structure and purpose of the experiment, which may lead them to reconsider their strategy. Therefore, the fact that the differences in \(m_2\) and \(m_3\) seize to be significant after the role change is not overly surprising.
Appendix 1 contains an example that illustrates how myopia affects deviations from full horizon optimal behavior.
I also estimated \(h^*\) using absolute deviations instead of squared deviations. The results are almost identical.
The treatment effect here is the change of the income process, which determines whether saving or borrowing is necessary for optimal consumption.
Figure 3 appears to suggest that there are different learning rates in the two treatments, since the median \(m_2\) is declining in the saving treatment and somewhat constant in the borrowing treatment. With respect to learning, this figure may be misleading because the median of (within subject) differences between two rounds is not necessarily the same as the difference between the medians of two rounds, i.e.: median \((m_2^{r-1}-m_2^{r})\lesseqgtr \) median \((m_2^{r-1})-\) median \((m_2^{r})\). The right hand side of this inequality can be obtained from Fig. 3 and Table 1, and the left hand side in Table 2. In order to analyze learning effects, I am focusing on within subject differences between rounds, i.e the left hand side of the above inequality.
In particular, it is hard to identify causality. Abstaining from debt could be due to debt aversion as well as lack of borrowing experience. Lack of borrowing experience, however, could also be caused by debt aversion.
References
Ballinger, T. P., Palumbo, M. G., & Wilcox, N. T. (2003). Precautionary saving and social learning across generations: An experiment. The Economic Journal, 113(490), 920–947.
Ballinger, T. P., Hudson, E., Karkoviata, L., & Wilcox, N. T. (2011). Saving behavior and cognitive abilities. Experimental Economics, 14(3), 349–374.
Brown, A. L., Chua, Z. E., & Camerer, C. F. (2009). Learning and visceral temptation in dynamic saving experiments. The Quarterly Journal of Economics, 124(1), 197–231.
Caballero, R. J. (1990). Consumption puzzles and precautionary savings. Journal of Monetary Economics, 25(1), 113–136.
Caballero, R. J. (1991). Earnings uncertainty and aggregate wealth accumulation. The American Economic Review, 81, 859–871.
Caetano, G., Palacios, M., & Patrinos, H. (2011). Measuring aversion to debt: An experiment among student loan candidates. World Bank Policy Research Working Paper Series.
Carbone, E. (2006). Understanding intertemporal choices. Applied Economics, 38(8), 889–898.
Carbone, E., & Duffy, J. (2013). Lifecycle consumption plans, social learning and external habits: Experimental evidence. Working paper.
Carbone, E., & Hey, J. D. (2004). The effect of unemployment on consumption: An experimental analysis. The Economic Journal, 114(497), 660–683.
Duffy, J. (2012). Macroeconomics: A survey of laboratory research. Handbook of Experimental Economics 2.
Eckel, C. C., Johnson, C., Montmarquette, C., & Rojas, C. (2007). Debt aversion and the demand for loans for postsecondary education. Public Finance Review, 35(2), 233–262.
Fenig, G., Mileva, M., & Petersen, L. (2013). Asset trading and monetary policy in production economies. Working Paper.
Field, E. (2009). Educational debt burden and career choice: Evidence from a financial aid experiment at NYU law school. American Economic Journal, 1(1), 1–21.
Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171–178.
Hey, J. D., & Dardanoni, V. (1998). Optimal consumption under uncertainty: An experimental investigation. The Economic Journal, 98(390), 105–116.
Lei, V., & Noussair, C. N. (2002). An experimental test of an optimal growth model. The American Economic Review, 92(3), 549–570.
Meissner, T., & Rostam-Afschar, D. (2014). Do tax cuts increase consumption? An experimental test of Ricardian Equivalence. Freie Universität Berlin, School of Business & Economics Discussion Paper 2014/16.
Noussair, C., & Matheny, K. (2000). An experimental study of decisions in dynamic optimization problems. Economic Theory, 15(2), 389–419.
Acknowledgments
I am grateful for valuable comments by Frank Heinemann, Dorothea Kübler, Dietmar Fehr and participants of the Berlin Behavioral Economics Colloquium and the Barcelona GSE Summer Forum Workshop on Theoretical and Experimental Macroeconomics. I also thank Satpal Nijjar for help with programming the experimental software. Financial support from the Deutsche Forschungsgemeinschaft (DFG) through CRC 649 “Economic Risk” is gratefully acknowledged.
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendix 1: Myopia and deviations from optimal behavior
Appendix 1: Myopia and deviations from optimal behavior
Example
Assume we are in the first period of the 20-period life-cycle consumption environment described in Sect. 2. Further assume that the decisionmaker is myopic and has a planning horizon of 2, that is when making a decision she acts as if the the next period is the final period. Myopic optimal consumption \(c^M\) in period 1 follows from the two-period optimization problem:
As explained in Sect. 3, different optimal myopic consumption levels arise in the different treatments. Since optimal consumption is a function of period wealth, optimal consumption levels also depend on the income shock. Long-term (fully rational) optimal consumption only depends on the income shock and does not differ between treatments.ϕ
-
1.
Borrowing treatment (increasing income stream): Short-term (myopic) optimal consumption is:
$$ c^M_B= {\left\{ \begin{array}{ll} 9.503 & \text{if}\; y_1=0 \, \text{(negative\;income\;shock) }\\ 19.503 & \text{if}\; y_1=20 \; \text{(positive \;income \;shock)} \end{array}\right. } $$(9) -
2.
Saving treatment (decreasing income stream): Short-term (myopic) optimal consumption is:
$$ c^M_S= {\left\{ \begin{array}{ll} 189.503 & \text{if}\; y_1=190\; \text{(negative\;income\;shock)}\\ 199.503 & \text{if}\; y_1=210 \; \text{(positive\;income\;shock) } \end{array}\right. } $$(10)
Long-term optimal consumption does not differ between treatments and is given by:
It is straight-forward to see that deviations from long-term optimal behavior caused by myopia are higher in the Borrowing treatment than in the Saving treatment when an negative income shock occurs:
Analogously, with a positive income shock deviations in the Saving treatment are higher than in the Borrowing treatment:
Rights and permissions
About this article
Cite this article
Meissner, T. Intertemporal consumption and debt aversion: an experimental study. Exp Econ 19, 281–298 (2016). https://doi.org/10.1007/s10683-015-9437-0
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10683-015-9437-0