1. Introduction to Behavioral Economics

1. Introduction to Behavioral Economics

Note

全书导读 / Book overview 本书《行为经济学与金融笔记》(Xindi He)分两部分:Part I 行为经济学(Ch 1–7:导论与经济学史、前景理论与参考点依赖偏好、现时偏误偏好、公平、行为激励、有偏信念、注意力与显著性)、Part II 行为金融(Ch 8–12:有效市场假说、对 EMH 的表观违背、心理学、投资者交易行为、学术研究与股票收益可预测性 McLean–Pontiff 2016)。本书较前五本更偏实证:大量田野实验与论文综述,定理少、叙述多。

Note

本章导读 §1.1 如何学行为经济学:行为经济学 (BE) 是革命性的,但并非比标准经济学"更好/更正确"——它放松完全理性假设以理解日常生活中的异象,但应作为标准经济学的补充而非替代。三条心法:标准经济学已能解释我们所见的大多数现象(仅在特殊/精心设计的情形才需行为修正,例:Lacetera et al. 2011 献血田野实验,提到奖励的传单反而带来更高献血率,符合标准分析);行为经济学极易出错(要想清底层故事并做实验检验);经济学(无论行为或标准)的终极目标是改善人们的生活,应开放地结合两个领域。§1.2 经济学史:1.2.1 古典经济学(1776 Adam Smith《国富论》起;亚当·斯密已有"公正的旁观者 vs 激情"的行为思想);1.2.2 新古典经济学(约 1870"边际革命",严格公理化数学、效用最大化,仅刻画"理性"部分;Samuelson 1948);1.2.3 新古典向其他领域扩张(Becker 等"芝加哥学派"把新古典推向犯罪/婚姻/法律/金融——"经济学帝国主义");1.2.4 行为经济学(Allais 1953 悖论、Simon 1955 有限理性、Ellsberg 1961 模糊厌恶;标准经济学家以 Rabin 所谓"解释开脱"回应;1970s 起系统性偏差被讨论,Tversky–Kahneman 1974 三大启发式 [代表性/可得性/锚定],Thaler 1987 起异象专栏)。

1. Introduction to Behavioral Economics

Note

Book overview This book, Behavioral Economics and Finance Notes (Xindi He), has two parts: Part I Behavioral Economics (Ch 1–7: introduction and history of economics, prospect theory and reference-dependent preferences, present-biased preferences, fairness, behavioral incentives, biased beliefs, inattention and salience); Part II Behavioral Finance (Ch 8–12: the efficient market hypothesis, apparent violations of EMH, psychology, investor trading behavior, academic research and stock-return predictability McLean–Pontiff 2016). This book is more empirical than the previous five: many field experiments and paper summaries, few theorems, more narrative.

Note

Overview §1.1 How to learn behavioral economics: behavioral economics (BE) is revolutionary, but it is not "better or more correct" than standard economics — it relaxes the perfect-rationality assumption to understand the anomalies in daily life, but should serve as a complement to, not a replacement for, standard economics. Three things to keep in mind: standard economics already explains most of what we see (we only correct it with behavioral insights in special / intentionally crafted scenarios — e.g. Lacetera et al. 2011's blood-donation field experiment, where reward-mentioning flyers actually yield a higher donation rate, consistent with standard analysis); BE is very easy to get wrong (think carefully through the underlying stories and run the test); economics (behavioral or standard) ultimately aims to improve people's lives, so be open-minded about combining the two fields. §1.2 History of economics: 1.2.1 classical economics (from Adam Smith's 1776 Wealth of Nations; Smith already had behavioral thoughts about the "impartial spectator vs passions"); 1.2.2 neoclassical economics (the ~1870 "marginal revolution", rigorous axiomatic math, utility maximization, modeling only the "rational" part; Samuelson 1948); 1.2.3 extending neoclassical economics to other fields (Becker and the "Chicago School" pushed neoclassical into crime/marriage/law/finance — "economics imperialism"); 1.2.4 behavioral economics (the Allais 1953 paradox, Simon 1955 bounded rationality, Ellsberg 1961 ambiguity aversion; standard economists responded with Rabin's so-called "explainawaytions"; from the 1970s systematic biases were discussed, Tversky–Kahneman 1974's three heuristics [representativeness/availability/anchoring], Thaler's anomalies columns from 1987).

1.1 如何学行为经济学 / How to Learn Behavioral Economics

行为经济学是革命性的:与通常假设"完全理性的人总是求解最大化问题"的标准经济学相对,行为经济学承认主体的有限理性,以更好地理解日常生活中发生的异象。但这并不意味着行为经济学"更好"或"更正确"。学习时应始终记住以下三点。

1.1 How to Learn Behavioral Economics

Behavioral economics is revolutionary: as opposed to standard economics, which typically assumes that perfectly rational agents always solve maximization problems, behavioral economics admits the bounded rationality of agents to better understand the anomalies happening in daily life. But this doesn't mean behavioral economics is "better" or "more correct". Keep the following three things in mind.

Important

三条心法 / Three things to keep in mind (1) 标准经济学已足以解释我们在现实中所见的大多数现象,只有在特殊情形或刻意设计的场景下才需要用行为方法修正,不要轻易抛弃标准理论(2) 行为经济学极易出错,对于直觉很强的人尤其如此——务必想清底层故事,并做实验来检验,不要只听信单方面的故事,要考虑所有可能方向。(3) 无论行为或标准,经济学的终极目标都是以好的方式改善人们的生活;要意识到这一目标,并开放地在必要时结合两个领域(1) Standard economics is good enough to explain most things we see in reality; only in special cases or intentionally crafted scenarios do we need to correct it with the behavioral approach, so don't throw away standard theories. (2) Behavioral economics is very easy to get wrong, especially for those with great intuitions — think carefully through the underlying stories and run the test; don't rely on only one side of the story, and consider all possible directions. (3) Whether behavioral or standard, economics ultimately aims to change people's lives in a good way; be aware of this goal and be open-minded about combining the two fields when necessary.

Tip

例:Lacetera et al. (2011) 献血田野实验 / Example: the Lacetera et al. (2011) blood-donation field experiment 向 10 万人随机邮寄传单,部分传单不提奖励、部分提到礼品卡(如 USD 10 奖励)。一个倾向行为经济学的人可能立刻论证:提到奖励的传单应带来更低的献血率,因为外在的金钱激励会挤出内在的利他动机——无奖励组有 0.63% 的人前来献血,按此逻辑有奖励组应更低。但这条看似合理的逻辑链并未成真:知道奖励的人实际以更高的比例 1.11% 前来献血——这在标准分析里再明显不过,因为金钱确实提供了激励。另一种故事:献血供给的短缺很严重,这本身反而可能激发更高的内在动机。结论:不要只依赖单方面的故事,想清所有方向,并真正做实验来看哪种直觉是对的。100,000 flyers about blood donation were randomly mailed out; some say nothing about a reward, while others mention a gift card (e.g. a USD 10 reward). Someone inclined toward behavioral economics might immediately argue that reward-mentioning flyers should yield a lower donation rate, because extrinsic financial incentives crowd out intrinsic altruistic motivation — 0.63% of the no-reward group showed up to donate, so by this logic the reward group should be lower. But this reasonable chain of logic doesn't turn out to be true: people who know about the reward actually donate at a higher percentage of 1.11% — which is obvious in standard analysis, since money does provide incentives. An alternative story: the shortage in blood supply is severe, which might itself trigger even higher intrinsic motivation. The lesson: don't rely on only one side of the story, think about all directions, and actually run the test to see which intuition turns out to be right.

1.2 经济学史 / History

1.2.1 古典经济学 / Classical Economics

  • 1776 年 Adam Smith 出版《国富论》(The Wealth of Nations),标志这一时期的开端。
  • 从 18 世纪晚期到 19 世纪中期,古典经济学主要在英国发展。
  • 这一时期的代表思想包括:供求分析、市场作为"看不见的手"、基于劳动时间的价值理论。
  • 代表性经济学家:Adam Smith、David Ricardo、John Stuart Mill、Thomas Malthus。
  • 一个值得注意之处:早在亚当·斯密时代,他在论证中就已有行为经济学的思想——人内心的挣扎介于"公正的旁观者" (impartial spectator,理性部分) 与"激情" (passions,非理性部分) 之间。
  • In 1776, Adam Smith's The Wealth of Nations was published, which marks the beginning of this period.
  • Between the late 18th century and the middle 19th century, classical economics was developed primarily in Britain.
  • This period features ideas such as demand-supply analysis, the market as the invisible hand, and a theory of value based on labor time.
  • Representative economists: Adam Smith, David Ricardo, John Stuart Mill, Thomas Malthus.
  • One thing notable: as early as Adam Smith's time, he already had thoughts of behavioral economics in his argument that human struggles between their "impartial spectator" (the rational part) and "passions" (the irrational part).

1.2.2 新古典经济学 / Neoclassical Economics

  • 约 1870 年的"边际革命"把经济学推入新阶段,称为新古典经济学。
  • 新古典经济学高度依赖严格的公理化数学,假设具有可容许偏好的个体在每个决策中都最大化自身效用。这一思路只聚焦于"公正的旁观者"(理性)部分,更易建模。
  • 1948 年 Paul Samuelson 出版名著《经济学:入门分析》(Economics: An Introductory Analysis),进一步改变了人们对正统经济学的理解;此后经济学界形成了强调严格数学推导与证明的新规范。
  • 新古典方法表现极好的领域(部分列举):消费者与生产者问题、博弈论、贸易问题、金融、外部性与公共品问题。
  • 代表性经济学家:Léon Walras、Alfred Marshall、Vilfredo Pareto、John Hicks、Paul Samuelson、John von Neumann。

1.2.3 把新古典经济学扩展到其他领域 / Extending Neoclassical Economics to Other Fields

  • Gary Becker 等经济学家把新古典方法应用到犯罪、婚姻、毒品、教育等其他学科:Becker → 社会学;George Stigler → 监管;Eugene Fama 与 Lars Hansen → 金融;Ronald Coase 与 Richard Posner → 法律。
  • 这些"芝加哥学派"经济学家把新古典经济学的理念推到极限,将其扩展到经济学之外几乎一切领域,称为"经济学帝国主义" (economics imperialism)。

1.2.4 行为经济学 / Behavioral Economics

在早期(1950–1960 年代),一些质疑开始投向新古典方法。

1.2.2 Neoclassical Economics

  • A "marginal revolution" moved economics into a new period around 1870, called neoclassical economics.
  • Neoclassical economics relies heavily on rigorous math with axioms, and assumes that individuals with admissible preferences maximize their own utility in every decision. This idea focuses purely on the "impartial spectator" (rational) part, which is easier to model.
  • In 1948, Paul Samuelson published the famous textbook Economics: An Introductory Analysis, which further changed people's understanding of canonical economics; since then a new norm in economics academia took shape, emphasizing rigorous math derivations and proofs.
  • Topics where the neoclassical approach turns out to be very good (a partial list): consumer and producer problems, game theory, trade problems, finance, externality and public-good problems.
  • Representative economists: Léon Walras, Alfred Marshall, Vilfredo Pareto, John Hicks, Paul Samuelson, John von Neumann.

1.2.3 Extending Neoclassical Economics to Other Fields

  • Gary Becker and other economists applied the neoclassical approach to other disciplines such as crime, marriage, drug, education and so on: Becker → sociology; George Stigler → regulations; Eugene Fama and Lars Hansen → finance; Ronald Coase and Richard Posner → law.
  • These "Chicago School" economists pushed the idea of neoclassical economics to its limit, extending it to almost everywhere outside economics, which is called "economics imperialism".

1.2.4 Behavioral Economics

In the early years (1950–1960), some doubts were cast upon the neoclassical approach.

Important

Allais (1953) 悖论 / The Allais (1953) paradox Allais (1953) 提出如下两组实验(见下表)。一个满足期望效用 (EU) 性质、效用 \(u\) 凹的人通常会在实验 1 中偏好 1A 甚于 1B,即 \(u(1)>0.89u(1)+0.1u(5)+0.01u(0)\) (1.1);又在实验 2 中偏好 2B 甚于 2A,即 \(0.11u(1)+0.89u(0)<0.1u(5)+0.9u(0)\),整理得 \(u(1)<0.89u(1)+0.1u(5)+0.01u(0)\) (1.2)。显然 (1.1) 与 (1.2) 相互矛盾。Allais (1953) presented the following two experiments (see the tables below). A person who satisfies the expected-utility (EU) property with a concave utility \(u\) would typically prefer 1A to 1B in Experiment 1, i.e. \(u(1)>0.89u(1)+0.1u(5)+0.01u(0)\) (1.1); and prefer 2B to 2A in Experiment 2, i.e. \(0.11u(1)+0.89u(0)<0.1u(5)+0.9u(0)\), which rearranges to \(u(1)<0.89u(1)+0.1u(5)+0.01u(0)\) (1.2). Clearly (1.1) and (1.2) contradict each other.

表 1.1(实验 1)/ Table 1.1 (Experiment 1)

Table 1.1 (Experiment 1)

Gamble 1A (Outcome / Prob.) Gamble 1B (Outcome / Prob.)
USD 1 million / 0.89 USD 1 million / 0.89
USD 1 million / 0.11 USD 0 / 0.01
USD 5 million / 0.1

表 1.2(实验 2)/ Table 1.2 (Experiment 2)

Table 1.2 (Experiment 2)

Gamble 2A (Outcome / Prob.) Gamble 2B (Outcome / Prob.)
USD 0 / 0.89 USD 0 / 0.89
USD 1 million / 0.11 USD 0 / 0.01
USD 5 million / 0.1
Tip

悖论的来源:独立性公理 / Source of the paradox: the independence axiom 悖论之所以出现,是因为人们"1A 优于 1B"与"2B 优于 2A"的偏好不满足独立性公理:两个实验中较低的两行其实是相同的两个赌局,加上第三行(每个表的第一行)本不应改变排序,但在大多数人的偏好里它确实改变了排序,从而违背了独立性。脚注:独立性 / 无关备选独立性 (IIA) 指:若有三个赌局 \(G_1,G_2,G_3\),把 \(G_3\) 以相同方式与 \(G_1,G_2\) 组合,不应改变 \(G_1\) 与 \(G_2\) 的排序。The paradox arises because people's preferences of 1A over 1B and 2B over 2A don't follow the independence axiom: the lower two rows in each experiment are actually the same two gambles, and adding the third row (the first row in each table) should not change the ranking, but in most people's preferences it does change the ranking, which breaks independence. Footnote: Independence / Independence of Irrelevant Alternatives (IIA) means that if there are three gambles \(G_1,G_2,G_3\), then the ranking of \(G_1\) and \(G_2\) won't change when \(G_3\) is combined the same way to both.

接下来的早期质疑还包括:

  • Simon (1955) 提出有限理性 (bounded rationality) 的思想,认为主体在决策中会犯随机性错误,但他没有识别出这些错误的底层原因。
  • Ellsberg (1961) 讨论了模糊厌恶 (ambiguity aversion):不确定的不确定性比确定的不确定性更糟,人们更愿意知道面对的是哪种不确定性。这与新古典分析相对——后者认为人们只关心最终结果的概率分布。

然而,大多数标准经济学家(下称"他们")以各种理由拒绝非理性,这些理由被 Matthew Rabin 总结为"解释开脱 (explainawaytions)":

The early doubts also include:

  • Simon (1955) proposed the idea of bounded rationality, believing that agents make stochastic errors in decision making, but he did not identify the underlying reasons for those errors.
  • Ellsberg (1961) discussed ambiguity aversion: uncertain uncertainty is worse than certain uncertainty, so it is preferable to know what uncertainty it is. This is in contrast to neoclassical analysis, in which people only care about the probability distribution of the final outcome.

However, most standard economists (referred to as "they" below) rejected irrationality with reasons summarized by Matthew Rabin as "explainawaytions":

Note

Rabin 的"解释开脱" / Rabin's "explainawaytions" "似乎" (as if):专家(如台球高手)的行为"似乎"做过复杂计算。更大的赌注:当赌注足够大时,主体会投入足够的努力去做复杂计算。市场竞争:市场竞争足够激烈,最终只有那些有足够经验("似乎"会算)或真能做复杂计算的人才会被雇用。福利经济学:若非理性成立,则人们不揭示其偏好,福利经济学就不复存在,所以我们需要假设理性。相互抵消:由非理性导致的错误最终会彼此抵消。"As if": an expert (say a billiard player) acts "as if" he has done complicated calculation. Bigger stakes: agents will put in enough effort to do the complicated calculation when the stake is big enough. Market competition: the market is competitive enough so that only the guys with enough experience ("as if") or who can do the complicated calculations will finally be hired. Welfare economics: if irrationality holds, then people don't reveal their preferences and welfare economics is gone, so we have to assume rationality. Canceling out: the errors due to irrationality will finally cancel each other out.

从 1970 年代起,系统性偏差 (systematic bias) 开始被讨论,这是行为经济学的一大进展(脚注:系统性偏差不只是一个随机误差参数,而是很多人都有、且可预测的偏差)。

  • Tversky and Kahneman (1974) 提出三大基本启发式与偏差 (heuristics and biases)(脚注:即便今日,这些偏差仍被认为由这三种底层启发式驱动;例如投射偏差 (projection bias)——人们把当前状态投射到未来——便由锚定驱动):
  • 代表性 (Representativeness):人们只关注他们认为有代表性的那部分,忽略他们认为不代表该事物的信息。例:被问问题时,人们倾向选听起来正确的那部分答案、忽略其他部分。
  • 可得性 (Availability):人们的回答取决于其脑海中可得的信息。例:被问密歇根州的谋杀案时人脑中没有危险画面,故报小数字;但被问底特律的谋杀案时脑中有危险场景,故报出比密歇根州还大的数字——这不可能是理性的,因为底特律是密歇根州的一部分。
  • 锚定 (Anchoring):人很难摆脱一个锚。例:先给人一个 1 到 100 之间的随机数,再问一个他们不知道的比例问题(如联合国中非洲国家的占比),结果其回答与那个随机数强相关,即便他们知道这数字是随机的。
  • 从 1987 年起,Richard Thaler 撰写"异象 (anomalies)"专栏,逐渐说服传统经济学家相信:标准经济学确实有时会失灵。

Starting from the 1970s, systematic bias began to be discussed, which is a major advance in behavioral economics (footnote: a systematic bias is not just a stochastic error parameter, but a bias a lot of people have and that is predictable).

  • Tversky and Kahneman (1974) proposed three fundamental heuristics and biases (footnote: even now these biases are still thought to be driven by these three underlying heuristics; e.g. projection bias — people tend to project the current status to the future — is driven by anchoring):
  • Representativeness: people pay attention only to the part of a thing they think is representative, and ignore information they don't think represents the thing. Example: when asked a question, people tend to pick the answer that sounds correct and ignore the other part.
  • Availability: people's responses are determined by what is available in their mind. Example: when asked about murders in Michigan people have nothing dangerous in mind, so they say small numbers; but when asked about murders in Detroit they have dangerous scenes in mind, so they say an even bigger number than for Michigan — which can't be rational, since Detroit is part of Michigan.
  • Anchoring: it's hard to move away from an anchor. Example: give people a random number between 1 and 100, then ask about a percentage in a problem they don't know (e.g. the percentage of African countries in the United Nations); they report answers strongly correlated to the random number even though they know it's random.
  • Starting in 1987, Richard Thaler wrote anomalies columns, which gradually convinced traditional economists that standard economics did fail sometimes.

参考文献 / References

  • Allais, M. (1953). Le comportement de l'homme rationnel devant le risque. Econometrica, 503–546.(Allais 悖论)
  • Ellsberg, D. (1961). Risk, Ambiguity, and the Savage Axioms. Quarterly Journal of Economics, 643–669.(模糊厌恶)
  • Lacetera, N., Macis, M., & Slonim, R. (2011). Rewarding Altruism? A Natural Field Experiment. NBER.(献血激励田野实验)
  • Simon, H. A. (1955). A Behavioral Model of Rational Choice. Quarterly Journal of Economics, 69(1), 99–118.(有限理性)
  • Tversky, A., & Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases. Science, 185(4157), 1124–1131.(启发式与偏差)

References

  • Allais, M. (1953). Le comportement de l'homme rationnel devant le risque. Econometrica, 503–546. (the Allais paradox)
  • Ellsberg, D. (1961). Risk, Ambiguity, and the Savage Axioms. Quarterly Journal of Economics, 643–669. (ambiguity aversion)
  • Lacetera, N., Macis, M., & Slonim, R. (2011). Rewarding Altruism? A Natural Field Experiment. NBER. (the blood-donation incentive field experiment)
  • Simon, H. A. (1955). A Behavioral Model of Rational Choice. Quarterly Journal of Economics, 69(1), 99–118. (bounded rationality)
  • Tversky, A., & Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases. Science, 185(4157), 1124–1131. (heuristics and biases)