The psychology of sunk cost: A classroom experiment

Abstract

Economics and business students are taught that sunk costs are irrelevant to their decisions. Yet, there is ample evidence that managers fail to integrate this simple rule and fall prey to what is known as the sunk-costs bias. To mitigate cognitive biases, such as the sunk-cost bias, educators must raise students’ awareness of these common judgment errors. In this article, the author proposes a classroom activity that actively engages students and allows them to identify this bias in their own judgments. The activity builds on a series of experiments from the psychology literature. The author discusses how these experiments have been adapted for classroom use and presents evidence suggesting that the activity increased students’ awareness of the sunk-cost bias and improved their decision-making skills.     

On Endogeneity Bias: Some General Remarks and Correction of the ROIW Paper by Bourguignon,Ferreira, and Menéndez (2007)

This paper is a short note on the question of correcting for endogeneity bias in a regression. It also points out errors and omissions in the derivation of this bias in Bourguignon, Ferreira, and Menéndez (2007). We show that some assumptions needed for the derivation are not explicit and even under these assumptions, certain simplifications adopted are not valid. We support our points using simulation experiments.     

Resolving questions about bias in real and hypothetical referenda

Despite 10 years of research on behavior in hypothetical referenda, conflict remains in the literature on whether or not the mechanism generates biased responses compared to real referenda, and the nature and source of any such bias. Almost all previous inquiry in respect of this issue has concentrated on bias at the aggregate level. This paper reports a series of three experiments which focuses on bias at the individual level and how this can translate to bias at the aggregate level. The authors argue that only an individual approach to hypothetical bias is consistent with the concept of incentive compatibility. The results of these experiments reflect these previous conflicting findings but go on to show that individual hypothetical bias is a robust result driven by the differing influence of pure self-interest and other- regarding preferences in real and hypothetical situations, rather than by a single behavioral theory such as free riding. In a hypothetical situation these preferences cause yea-saying and non-demand revealing voting. This suggests that investigation of individual respondents in other hypothetical one-shot binary choices may also provide us with insights into aggregate behavior in these situations. Alphabetical list of authors no seniority of authorship attributed.     

Demand Revelation,Hypothetical Bias,and Threshold Public Goods Provision

We examine the issues of hypothetical bias, demand revelation, and gender differences in a threshold public goods experiment with heterogeneous induced-values. First, we find no evidence of hypothetical bias in our threshold provision public goods experiments, despite the fact this is an open-ended type mechanism. Our results support recent experimental findings that use induced-values to investigate hypothetical bias. Second, we investigate the demand revealing performance of real and hypothetical payments and find no evidence that real payment performs better than hypothetical payment in our experiments. Third, we examine whether payments, real and hypothetical, are positively related to induced-values. Our results suggest there is statistical evidence that payments are positively related to true values. Finally, we examine the effect of gender on real and hypothetical payments in our experiment. The results show that gender matters for contributions through both real and hypothetical payments, after controlling for true values as well as socio-economic variables. Results also indicate that females are more likely to truthfully reveal their value than males through hypothetical payments, but gender is not significant for truthfully revealing their value through real payments.     

The irrelevant-menu affect on valuation

Three experiments are designed to test if the level of irrelevant prizes in the menu has a positive (assimilation) or negative (contrast) effect on the perceived valuation of target objects. Familiar field prizes and binary lotteries over such prizes are placed within “more-expensive” and “less-expensive” menus. Subjects fill-in a sequence of binary choice problems to reveal their preference between given cash and a designated prize from the menu. Between-subject comparisons reveal that the prize-level in the menu positively affects perceived valuations in spite of procedural attempts to rule out menu-dependent preferences and prohibit experimenter bias. The effect also shows within-subject in auction experiments: the price that subjects are willing to pay for given monetary lotteries significantly increases with the average payoff in the irrelevant-menu. The bias finally manifests even when subjects are led to choose the target lottery, independently, from the underlying menu.     

Estimating the gravity model when zero trade flows are frequent and economically determined

ABSTRACT

Since the 1950s, we have known that the presence of zero-valued dependent variables can seriously bias econometric estimates whether the zeros are included or excluded. Yet the widely-used gravity model is frequently estimated on samples that include large fractions of zeros. An influential paper by Santos Silva and Tenreyro – based on simulations that include no economically-determined zeros – concludes that the bias problems resulting from zeros and those resulting from heteroscedasticity and nonlinearity can be solved using the Poisson Pseudo-Maximum-Likelihood (PPML) model including the zero values. This paper begins by adapting the Santos Silva and Tenreyro experimental design to include economically-determined zeros to see whether this conclusion continues to hold. With this design, it finds that alternative estimators have lower bias than PPML. Changing to a Monte Carlo design that replicates the much-higher real-world frequency of predicted values near zero restores the finding of lower bias with the PPML estimator. The results highlight the need for very careful design of Monte Carlo experiments when evaluating alternative estimators of the gravity model.     

Exchange Rules and the Incentive Compatibility of Choice Experiments

The incentive properties of stated-preference surveys continue to be a central debate in the valuation of public goods. The majority of empirical studies have focused on incentive properties of contingent valuation questions in relation to situations where answers have monetary consequences. This research explores the incentive properties of repeated, attribute-based choice questions when subjects are provided with an explicit connection between choices and outcomes. Two market/provision-rules are investigated: a posted-price market and a plurality-rule vote. These two provision rules are contrasted to treatments in which no provision rule is discussed—subjects are simply asked to choose their preferred alternative. These three hypothetical choice treatments are compared with a binding choice treatment. While none of the public good treatments are theoretically incentive compatible, we include a comparison of hypothetical and binding choices for a private-good that is incentive compatible. The private good experiments indicate that marginal willingness to pay (WTP) estimates from the hypothetical treatment are larger, but not statistically different than corresponding estimates in the binding choice treatment. Results for the public good experiments indicate that marginal WTP estimates from the hypothetical treatments are much larger, and statistically different than corresponding estimates in the binding choice treatment. The bias is largest when no provision rule is discussed. The bias is reduced with the inclusion of a provision rule, but surprisingly, there was no difference across provision rule treatments. Overall, our results indicate that choice experiments involving a public good should include a provision rule to reduce bias, but the resulting marginal WTP estimates may still be more biased, on average, than those arising from contingent valuation survey formats.     

Bootstrapping structural VARs: Avoiding a potential bias in confidence intervals for impulse response functions

Constructing bootstrap confidence intervals for impulse response functions (IRFs) from structural vector autoregression (SVAR) models has become standard practice in empirical macroeconomic research. The accuracy of such confidence intervals can deteriorate severely, however, if the bootstrap IRFs are biased. We document an apparently common source of bias in the estimation of the VAR error covariance matrix which can be easily reduced by a scale adjustment. This bias is generally unrecognized because it only affects the bootstrap estimates of the error variance, not the original OLS estimates. Nevertheless, as we illustrate here, analytically, with sampling experiments, and in an example from the literature, the bootstrap error variance bias can have significant distorting effects on bootstrap IRF confidence intervals. We also show that scale-adjusted bootstrap confidence intervals can be expected to exhibit improved coverage accuracy.     

Can biased polls distort electoral results? Evidence from the lab

We introduce a new methodological approach for studying the effect of biased polls on election outcomes and apply it to a set of new experiments with 375 participants. Voters may observe and learn about the bias by playing multiple voting rounds. While in control conditions, polls are unbiased, in treatment conditions, participants view only poll results where a particular candidate’s vote share is the largest. This candidate is consistently elected more often in the treatments than in the controls, because biased polls robustly distort voters’ expectations about vote shares. This effect holds after eighteen election rounds, out of which the first three are practice rounds, but somewhat more weakly in our main treatment where voters are explicitly informed about the bias.     

Computerizing audit studies

This paper briefly discusses the history, benefits, and shortcomings of traditional audit field experiments to study market discrimination. Specifically it identifies template bias and experimenter bias as major concerns in the traditional audit method and demonstrates through an empirical example that computerization of a resume or correspondence audit can efficiently increase sample size and greatly mitigate these concerns. Finally, it presents a useful meta-tool that future researchers can use to create their own resume audits.     

Singleton bias and lexicographic preferences among equally valued alternatives

Two experiments examined a form of singleton bias deducible from lexicographic choice and Tversky’s theory of elimination by aspects. In Experiment 1, 100 decision makers who chose from a set of job applicants defined by equal numbers of equally important qualifications tended to ignore the singleton defined by possession of a relevant attribute but tended to rank-order the attributes as predicted by lexicographic choice theory and showed various forms of singleton bias. In Experiment 2, 100 decision makers who chose from sets of unspecified alternatives, universities, and houses/apartments defined by attributes that they had individually rated as equally important ignored the attribute-defined singleton in every alternative set but manifested a different singleton bias in favor of the middle option of the on-screen display.     

An experimental study of blind proficiency tests in forensic science

We conducted a series of sender?Creceiver experiments to study the consequences of implementing a regime of blind proficiency tests in forensic science to reduce error rates and improve the criminal justice system. Senders are our surrogate for forensic laboratories and receivers, for the judge or jury. Our experimental surrogate (random audits with a penalty) for blind proficiency tests reduced sender error rates by as much as 46% depending on the level of experimentally induced bias. When penalties improve information quality, receiver error rates fell by as much as 26% depending on the level of the sender bias. We also find that the penalty must be large relative to the payoff to induce the reduction in errors. Our results suggest that a regime of blind proficiency testing has the potential to reduce forensic science errors.     

Is there selection bias in laboratory experiments? The case of social and risk preferences

Laboratory experiments are frequently used to examine the nature of individuals’ social and risk preferences and inform economic theory. However, it is unknown whether the preferences of volunteer participants are representative of the population from which the participants are drawn, or whether they differ due to selection bias. To answer this question, we measured the preferences of 1,173 students in a classroom experiment using a trust game and a lottery choice task. Separately, we invited all students to participate in a laboratory experiment using common recruitment procedures. To evaluate whether there is selection bias, we compare the social and risk preferences of students who eventually participated in a laboratory experiment to those who did not, and find that they do not differ significantly. However, we also find that people who sent less in a trust game were more likely to participate in a laboratory experiment, and discuss possible explanations for this behavior.     

Home bias in multimarket Cournot games

We explore the role of trade costs for the home bias in trade. In a series of Cournot duopoly experiments with a home and an export market, we compare output choices when firms face different levels of trade costs. We find that there is two-way trade in identical products and that firms hold the majority market share in their home market. The resulting home bias turns out to be, however, stronger than that predicted by theory, and it even occurs without trade costs. Tacit collusion contributes to the home bias observed in our experiment but does not offer a full explanation for the phenomenon.     

Exploitable actions of believers in the “law of small numbers” in repeated constant-sum games

In repeated fixed-pair constant-sum games with unique equilibria in mixed strategies, such as matching pennies, the subgame perfect equilibrium is repeating the stage-game mixed-strategy equilibrium action. In such games rational players avoid strategies that are exploitable, in that current actions either deviate systematically from the equilibrium action probabilities or fail to be serially independent of past actions. I revisit classic experiments and find that subjects’ actions are sometimes exploitable because they are serially dependent. Subjects have difficulty in producing serially independent actions and in recognizing serially dependent sequences due to a bias called local representativeness.     

Experimental approach to the determination of the demand for public goods

The paper proposes a new method of estimating the demand for public goods in view of the ‘free rider’ problem. The idea is to establish an experimental setting in which only a random sample of the population is investigated. Alternative experiments are then proposed with the intent of discovering the true individual demand. First a statistical method is proposed to estimate the bias in response. Then, a random environment is created and it is shown under what conditions a successful elicitation of the demand is possible.     

The effects of rent asymmetries in experimental auction markets

We report the results of twelve ‘double-auction’ market experiments designed to analyze the effects of asymmetric induced supply and demand configurations on the price convergence path toward a competitive equilibrium. The proposition (convergence bias) that prices tend to approach the competitive equilibrium from above (below) when consumer surplus is greater (less) than producer surplus cannot be rejected. We do, however, reject the proposition (convergence symmetry) that these convergence biases are of equal absolute magnitude. Excesses of producer over consumer surplus are found to have a more pronounced effect on the sequence of contract prices.     

Risk attitudes, randomization to treatment, and self-selection into experiments

Randomization to treatment is fundamental to statistical control in the design of experiments. However randomization implies some uncertainty about treatment condition, and individuals differ in their preferences towards taking on risk. Since human subjects often volunteer for experiments or are allowed to drop out of the experiment at any time if they want to, it is possible that the sample observed in an experiment might be biased because of the risk of randomization. On the other hand, the widespread use of a guaranteed show-up fee that is non-stochastic may generate sample selection biases of the opposite direction, encouraging more risk averse samples into experiments. We directly test these hypotheses that risk attitudes play a role in sample selection. Our results suggest that randomization bias does affect the overall level of risk aversion in the sample we observe, but that it does not affect the demographic mix of risk attitudes in the sample. We show that the common use of non-stochastic show-up fees can generate samples that are more risk averse than would otherwise have been observed.     

Eliciting Willingness to Pay without Bias using Follow-up Certainty Statements: Comparisons between Probably/Definitely and a 10-point Certainty Scale

Correction for hypothetical bias using follow up certainty questions often takes one of two forms: (1) two options, “definitely sure” and “probably sure”, or (2) a 10-point scale with 10 very certain. While both have been successful in eliminating hypothetical bias from estimates of WTP by calibrating based on the certainty of yes responses, little is known about the relationship between the two. The purpose of this paper is to compare the two using data from three field experiments in a private good, dichotomous choice format. We compare four types of yes responses that differ in the criterion used to determine if there is sufficient certainty for a hypothetical yes response to be considered a true yes response. We make several comparisons, but focus on determining which values on the 10-point scale give the same estimates of WTP as “definitely sure” hypothetical yeses and real yeses (actual purchases). Values that produce equivalence are near 10 on the certainty scale.     

IMPROVING FORECAST ACCURACY BY COMBINING RECURSIVE AND ROLLING FORECASTS*

This article presents analytical, Monte Carlo, and empirical evidence on combining recursive and rolling forecasts when linear predictive models are subject to structural change. Using a characterization of the bias–variance trade‐off faced when choosing between either the recursive and rolling schemes or a scalar convex combination of the two, we derive optimal observation windows and combining weights designed to minimize mean square forecast error. Monte Carlo experiments and several empirical examples indicate that combination can often provide improvements in forecast accuracy relative to forecasts made using the recursive scheme or the rolling scheme with a fixed window width.