Behavioral Economics and GPT-4: From William Shakespeare to Elena Ferrante
There is a new paper on LLMs by Gabriel Abrams, here is the abstract:
We prompted GPT-4 (a large language model) to play the Dictator game, a classic behavioral economics experiment, as 148 literary fictional characters from the 17th century to the 21st century.
Of literary interest, this paper analyzed character selfishness by century, the relative frequency of literary character personality traits, and the average valence of these traits. The paper also analyzed character gender differences in selfishness.
From an economics/AI perspective, this paper generates specific and quantifiable Turing tests which the model passed for zero price effect, lack of spitefulness and altruism, and failed for human sensitivity to relative ordinal position and price elasticity (elasticity is significantly lower than humans). Model updates from March to August 2023 had relatively minor impacts on Turing test outcomes.
There is a general and mainly monotonic decrease in selfish behavior over time in literary characters. 50% of the decisions of characters from the 17th century are selfish compared to just 19% of the decisions of characters from the 21st century. Overall, humans exhibited much more selfish behavior than AI characters, with 51% of human decisions being selfish compared to 32% of decisions made by AI characters.
Historical literary characters have a surprisingly strong net positive valence across 2,785 personality traits generated by GPT-4 (3.2X more positive than negative). However, valence varied significantly across centuries. The most positive century, in terms of personality traits, was the 21st — over 10X the ratio of positive to negative traits. The least positive century was the 17th at just 1.8X. “Empathetic,” “fair” and “selfless,” were the most overweight traits in the 20th century. Conversely, “manipulative,” “ambitious” and “ruthless” were the most overweight traits in the 17th century.
Male characters were more selfish than female characters: 35% of male decisions were selfish compared to just 24% for female characters. The skew was highest in the 17th century where selfish decisions for male and female were 62% and 20% respectively.
This analysis offers a specific and quantifiable partial Turing test. In a few ways, the model is remarkably human-like; The key human-like characteristics are the zero price effect, lack of spitefulness and altruism. However, in other ways, GPT-4 reflects unusual or inhuman preferences. The model does not appear to have human sensitivity to relative ordinal position and has significantly lower price elasticity than humans.
Model updates in GPT-4 have made it slightly more sensitive to ordinal value, but not more selfish. The model shows preference consistency across model runs for each character with respect to selfishness.
To which journal might you advise him to send this paper?
Evan Soltas on the job market from MIT
Here is his home page, the job market paper is “Tax incentives and the supply of low income housing”:
Subsidies to developers are a core instrument of housing policy. How do they affect housing markets, and who benefits? I assess their impacts and incidence with a dynamic model of housing markets and new data on developers competing for Low-Income Housing Tax Credits. I estimate the model using three sources of variation: quasi-random assignment of subsidies, shocks to subsidy generosity, and nonlinearities in scoring rules for subsidy applications. I find that, due to displacement of unsubsidized housing, subsidies add few net units to the housing stock and instead reallocate units progressively. Households benefit from developer competition for subsidies, but competition also results in high entry costs, and developers still capture nearly half of the welfare gains. In counterfactuals, a stylized voucher program can generate the same household benefits at less fiscal cost.
Recommended!
Minimum wages and rents
This topic remains underdiscussed in the minimum wage debates, here are some recent results from Atsushi Yamagishi:
I analyze the effect of minimum wage hikes on housing rents using exogenous variation in minimum wages across local labor markets in Japan. I estimate that in low-quality rental housing market, a 10% minimum wage increase induces a 2.5%–4.5% increase in rents. Minimum wage hikes benefit workers in light of a spatial equilibrium model showing that changes in housing market rents work as a sufficient statistic for measuring utility changes arising from changes in minimum wages. The increase in housing rents also implies an unintended benefit for homeowners.
Atsushi Yamagishi is from Princeton economics, but that is not his job market paper, here is the whole portfolio, which looks quite interesting.
Friday assorted links
1. How did the railway affect the diffusion of ideas in 19th century Germany? (Caterina Chiopris is a job market candidate from Harvard).
2. Yiong Huang on “breaking the spiral of silence,” job market candidate from Harvard.
3. “Notably, at selective institutions, new [course] content focuses on societal issues, while at less selective institutions, new content emphasizes job-relevant skills.” Jacob Light is a job market candidate from Stanford.
4. My 2015 MR post on the future coming first to Israel (and Singapore).
Good Developments in Africa
The Guardian: Visas to visit Kenya are to be scrapped for other African nationals from next year as part of a movement towards opening up trade and travel within the continent.
“By the end of this year, no African will be required to have a visa to come to Kenya,” Kenya’s president, William Ruto, said at a climate change conference in Congo-Brazzaville.
Costly and time-consuming visa requirements, as well as high air fares, have long created barriers to inter-African travel for African passport holders; 32 out of 54 African countries still require the nationals of half or more countries on the continent to obtain a visa.
Good. Kenya should also scrap visas for US and European citizens!
This is part of Africa’s move to free trade with the The African Continental Free Trade Area (AfCFTA), the largest free trade area in the world measured by the number of countries participating. Eventually AfCTA will cover 1.3 billion people across 55 countries.
Visas can have very high costs in reducing travel and trade. Africa is moving in the right direction. The US and Europe in contrast are adding visa requirements and the wait times to get a US visa are an absurd embarrassment.
My Conversation with Harriet Karimi Muriithi
This is another CWT bonus episode, recorded in Tatu City, Kenya, outside of Nairobi. Harriet is a 22-year-old waitress. Here is the audio, video, and transcript. Here is the episode summary:
Harriet is a 22-year-old hospitality professional living and working in Tatu City, a massive mixed-used development spearheaded by Jennings. Harriet grew up in the picturesque foothills of Mount Kenya before moving to the capital city as a child to pursue better schooling. She has witnessed Nairobi’s remarkable growth firsthand over the last decade. An ambitious go-getter, Harriet studied supply chain management but and wishes to open her own high-end restaurant.
In her conversation with Tyler, Harriet opens up about her TikTok hobby, love of fantasy novels, thoughts on improving Kenya’s education system, and how she leverages AI tools like ChatGPT in her daily life, the Chinese influence across Africa, the challenges women face in village life versus Nairobi, what foods to sample as a visitor to Kenya, her favorite musicians from Beyoncé to Nigerian Afrobeats stars, why she believes technology can help address racism, her Catholic faith and church attendance, how COVID-19 affected her education and Kenya’s recovery, the superstitions that persist in rural areas, the career paths available to Kenya’s youth today, why Nollywood movies captivate her, the diversity of languages and tribes across the country, whether Kenya’s neighbors impact prospects for peace, what she thinks of the decline in the size of families, why she enjoys podcasts about random acts of kindness, what infrastructure and lifestyle changes are reshaping Nairobi, if the British colonial legacy still influences politics today, and more.
Here is one excerpt:
COWEN: How ambitious are you?
MURIITHI: On a scale of 1 to 10, I will say an 8.5.
This episode is best consumed in combination with the episode with the village elder Githae Gitinji. The contrast between the two perspectives is startling. And here is my CWT episode with Stephen Jennings, concerning Tatu City itself.
My Conversation with Githae Gitinji
This is a special bonus episode of CWT, Githae is a 58-year-old village elder who mediates disputes and lives in Tatu City, Kenya, near Nairobi. Here is the audio, video, and transcript. Here is the episode summary:
In his conversation with Tyler, Githae discusses his work as a businessman in the transport industry and what he looks for when hiring drivers, the reasons he moved from his rural hometown to the city and his perspectives on urban vs rural living, Kikuyu cultural practices, his role as a community elder resolving disputes through both discussion and social pressure, the challenges Kenya faces, his call for more foreign investment to create local jobs, how generational attitudes differ, the role of religion and Githae’s Catholic faith, perspectives on Chinese involvement in Kenya and openness to foreigners, thoughts on the devolution of power to Kenyan counties, his favorite wildlife, why he’s optimistic about Kenya’s future despite current difficulties, and more.
Excerpt:
COWEN: What do you do that the court system does not do? Because you’re not police, but still you do something useful.
GITHINJI: What we normally do, we as a group, we listen to one another very much. When one person reaches that stage of being told that you are a man now, you normally have to respect your elder. Those people do respect me. When I call you, when I tell you “Come and we’ll talk it out,” with my group, you cannot say you cannot come, because if you do, we normally discipline somebody. Not by beating, we just remove you from our group. When we isolate you from our group, you’ll feel that is not fair for you. You come back and say — and apologize. We take you back into the group.
COWEN: If you’re isolated, you can’t be friends with those people anymore.
GITHINJI: When we isolate you, we mean you are not allowed to interact in any way.
COWEN: Any way.
GITHINJI: Any business, anything with the other community [members]. If it is so, definitely, you have to be a loser, because you might be needing one of those people to help you in business or something of the sort. When you are isolated, this man tells you, “No. Go and cleanse yourself first with that group.”
If you find his Kikiyu accent difficult, just read the transcript instead. This episode is best consumed in a pair with my concurrently recorded episode with Harriet Karimi Muriithi, a 22-year-old Kenyan waitress — the contrasts in perspective across a mere generation are remarkable.
Basil Halperin on the job market from MIT
Here is the home page, here is his job market paper (with Daniele Caratelli) “Optimal policy under menu costs”:
We analytically characterize optimal monetary policy in a multisector economy with menu costs, and show that inflation and output should move inversely after sectoral shocks. That is, following negative shocks, inflation should be allowed to rise, and vice versa. In a baseline parameterization, optimal policy stabilizes nominal wages. This nominal wage targeting contrasts with inflation targeting, the optimal policy prescribed by the textbook New Keynesian model in which firms are permitted to adjust their prices only randomly and exogenously. The key intuition is that stabilizing inflation causes shocks to spill over across sectors, needlessly increasing the number of firms that must pay the fixed cost of price adjustment compared to optimal policy. Finally, we show in a quantitative model that, following a sectoral shock, nominal wage targeting reduces the welfare loss arising from menu costs by 81% compared to inflation targeting.
Noteworthy!
Apple Ad From 1999
Thursday assorted links
1. Maxims from Larry Gagosian.
2. How might California regulate AI?
3. Why has growth in medical knowledge been so stagnant? Northwestern job market paper by Megumi Murakami.
4. My AI webinar for Macmillan. And also from Macmillan here is an Eric Parsons profile video.
5. Are Singaporean couples who are funny more satisfied with each other?
6. Pakistan starts to expel 1.7 million undocumented Afghani migrants. That is perhaps the largest forced expulsion since the 1950s?
7. Paul’s latest (video). And the song itself.
What should I ask Fuchsia Dunlop?
Yes I will be doing a third (!) Conversation with her. She has written some of the best books and cooked some of the best food. Here is her Wikipedia page. Here is her home page.
Here is her new book — self-recommending if anything was — Invitation to a Banquet: The Story of Chinese Food.
Here is my previous podcast with Fuchsia, and here is my first podcast with her. This time around — what should I ask?
Marriage sentences to ponder
Increased exposure to opposite-sex members of other class groups generates a substantial increase in interclass marriage, but increased exposure to other race groups has no detectable effect on interracial marriage.
That is from the job market paper of Benjamin Goldman, a job market candidate from Harvard University.
My excellent Conversation with Stephen Jennings
Recorded in Tatu City, Kenya, not far from Nairobi, Tatu City is a budding Special Enterprise Zone. Here is the transcript, audio, and video. Here is the episode overview:
Stephen and Tyler first met over thirty years ago while working on economic reforms in New Zealand. With a distinguished career that transitioned from the New Zealand Treasury to significant ventures in emerging economies, Stephen now focuses on developing new urban landscapes across Africa as the founder and CEO of Rendeavour.
Tyler sat down with Stephen in Tatu City, one of his multi-use developments just north of Nairobi, where they discussed why he’s optimistic about Kenya in particular, why so many African cities appear to have low agglomeration externalities, how Tatu City regulates cars and designs for transportation, how his experience as reformer and privatizer informed the way utilities are provided, what will set the city apart aesthetically, why talent is the biggest constraint he faces, how Nairobi should fix its traffic problems, what variable best tracks Kenyan unity, what the country should do to boost agricultural productivity, the economic prospects for New Zealand, how playing rugby influenced his approach to the world, how living in Kenya has changed him, what he will learn next, and more.
Here is one excerpt:
COWEN: Just give us some basic facts. Where is Tatu City right now, and where will it be headed when it’s more or less finished?
JENNINGS: Tatu City is the only operational special economic zone [SEZ] in the country. It is 5,000 hectares of fully planned urban development. It is at quite an advanced stage. We have 70 large-scale industrial companies with us, including major multinationals and many of the regional leaders. We have 3,000 students come on site every day to our four new schools. We’re advanced in building the first phase of the first new CBD for the region. We have tens of thousands of core center jobs moving into that area, together with other modern office amenities. All of the elements — we have many residential modules, thousands of new residential units at a wide range of price points — all of the elements of a new city are in place.
COWEN: How many people will end up living here?
JENNINGS: Around 250,000.
COWEN: And how many businesses?
JENNINGS: There’ll be thousands of businesses.
And delving more deeply into matters:
COWEN: What do you think is the book [on economic development] that has influenced you most?
JENNINGS: It’s a very good question. I think I’ve read just about everything in development. There’s nothing I really like very much. Development is a black box. I don’t think there’s anything that has much predictive power. There’s a lot of ex post explanations, whether they be policy settings, location, culture. I think 90% of them are ex post; very few of them are predictive. Some of them are just tautologies. I really like factualization.
It’s descriptive more than analytical, but it just makes it clear that most of the world has been on a very similar development trajectory. It’s just not sequenced. Sweden started early; Ethiopia started late. But the nature of the transition and the inevitability of that transition, other than very extreme circumstances, is kind of the same.
COWEN: What do you think economists get wrong?
JENNINGS: I don’t think we really understand development at all, because if we could, we could predict it. We can predict virtually nothing. It’s just too complicated. It’s too connected with politics. I think there’s a lot of feedback loops and elements of development that we don’t understand properly. We certainly can’t quantify them because the development’s happening in such a wide range of settings, from communism dictatorships through to very liberal systems and with all different kinds of industrial — on every dimension, there’s a huge range of variables.
Excellent and interesting throughout.
Wednesday assorted links
1. Jake Seliger on the clinical trial system, from the point of view of a patient. And from his wife Bess Stillman.
2. The war in Sudan, and its economics.
3. Progress on the dengue front (link corrected).
4. Joy Buchanan review of GOAT.
5. Zvi’s review of the AI Executive Order, recommended if you care.
Why aren’t the American hostages receiving more attention?
Yesterday Senator Marsha Blackburn tweeted: “The White House admitted Hamas is holding nearly 500 Americans hostage in Gaza.” To be clear, those are Americans not allowed to leave Gaza (NYT), they are not being held in a compound. As for hostages in the narrower sense of that term, there seem to be about ten. Neither are instances of liberty, nor are they safe positions to be in. Personally, I would consider both groups to be hostages.
No matter which definition of hostage you prefer, I don’t see so many major MSM articles about these hostages. I remember the much earlier Iranian hostage crisis, when many Americans even knew the identities and life stories of individual hostages. It was a front page item almost every day. As I am composing this post (the day before), I don’t see it on the NYT front page at all. Same with WaPo, though they do have “Biden hosts Trick or Treaters at the White House.” If you consider this New Yorker story, well yes it covers the American hostages somewhat, but it is nothing close to what I might have expected. They are not even the article lead. So why so little coverage? I have a few candidate hypotheses, which may or may not be true:
1. The MSM wants Biden’s reelection, and they don’t want him ending up painted as “another Jimmy Carter” who cannot rescue the hostages.
2. The young Woke staffers at MSM don’t want to make Hamas look too bad, or to make the Israeli retaliation look too good.
3. These hostages are themselves not “The Current Thing,” even though the war itself seems to be The Current Thing. Has “The Current Thing” become so narrowly circumscribed?
4. For some national security reasons, MSM has been told by our government that too much hostage coverage would endanger their possible release or rescue.
5. The Biden administration is pressuring MSM not to cover the hostages too much, out of fear that the domestic pressures for America to intervene will become too strong.
6. It simply hasn’t happened yet, due to noise in the system.
What else? I am not saying any of those are true, and some of them are more conspiratorial than the kinds of explanation I usually find persuasive. So why don’t the American hostages receive more attention?