Month: November 2023

Can ChatGPT assist in picking stocks?

by  Tyler Cowen in

I don’t believe this result will hold up, but I am happy to present the opposing point of view:

This paper studies whether ChatGPT-4 with access to the internet is able to provide valuable investment advice and evaluate financial information in a timely manner. Using a live experiment, we find a positive correlation between ChatGPT-4 ratings and future earnings announcements and stock returns. We find evidence that ChatGPT-4 adjusts ratings in response to earnings surprises and news events information in a timely manner. An investment strategy based on “attractiveness ratings” by ChatGPT-4 yields positive returns.

That is from a new paper by Matthias Pelster and Joel Val.

Same-sex marriage and priestly employment

by  Tyler Cowen in

We study the effect of legalization of same-sex marriage on coming out in the United States. We overcome data limitations by inferring coming out decisions through a revealed preference mechanism. We exploit data on enrollment in seminary studies for the Catholic priesthood, hypothesizing that Catholic priests’ vow of celibacy may lead gay men to self-select as a way to avoid a heterosexual lifestyle. Using a differences-in-differences design that exploits variation in the timing of legalization across states, we find that city-level enrollment in priestly studies fell by about 15% exclusively in states adopting the reform. The celibacy norm appears to be driving our results, since we find no effect on enrollment in deacon or lay ministry studies that do not require celibacy. We also find that coming out decisions, as inferred through enrollment in priestly studies, are primarily affected by the presence of gay communities and by prevailing social attitudes toward gays. We explain our findings with a stylized model of lifestyle choice.

That is from a new working paper by Avner Seror and Tohit Ticku.

Monday assorted links

by  Tyler Cowen

1. Big changes in energy and climate policy over the last five years, a list.

2. My 2014 post on immigration, and cosmopolitanism at the margin.  And my 2005 post “Tyler Cowen pretends he is a Democrat.”

3. Dustin Sebell reviews the BAP dissertation (persist and you can get to the actual review).

4. WaPo runs a front page story about the first freed American hostage, so I am linking to it.

5. New prize fund for mathematically intelligent AI.

6. Would your cat eat your corpse?

Bona Vacantia

by  Alex Tabarrok in

Bona vacantia are unowned assets–primarily the assets of people who die without a will and without heirs. In Britain, as in the United States, unowned assets generally pass to the government but Britain is full of ancient and strange traditions and one of them is that the estates of people resident in the “ancient county palatine of Lancashire” who die without a will or heirs go to the Duchy of Lancaster.

The Duchy of Lancaster is “a portfolio of lands, properties, and assets held in trust for the sovereign.” The Duchy dates back to 1265 but in 1461 Edward IV made it a distinct and private inheritance of the reigning monarch. As such the reigning monarch is not allowed to sell the Duchy but is due any returns. To be clear, this is all different from the Crown Estate which are a bunch of land and property nominally owned by the monarch and dedicated to funding the monarchy but yet not owned by the monarch privately. Having fun yet?

Periodically someone discovers the duchies and kicks up a fuss and there is a debate about making the revenues public and giving the monarch a stipend instead. The latest fuss was kicked up by the Guardian with an article that describes the whole thing in accurate but lurid terms:

The king is profiting from the deaths of thousands of people in the north-west of England whose assets are secretly being used to upgrade a commercial property empire managed by his hereditary estate, the Guardian can reveal.

Great stuff! It could only have been improved by adding “the king and his son” are profiting from the deaths of thousands of people because there is another Duchy in Cornwall which is owned by the eldest son of the reigning monarch, in this case Prince William, where the same rules apply. Prince William also has the rights to all the royal fish caught off the shores of the Duchy—namely to whales, porpoises, and sturgeon (obviously).

Despite having once read a book on the subject, I am obviously not an expert in the intricacies of British inheritance law. Who is? But I found the back story very amusing.

Carrying opioids in legal imports?

by  Tyler Cowen in

The U.S. opioid crisis is now driven by fentanyl, a powerful synthetic opioid that currently accounts for 90% of all opioid deaths. Fentanyl is smuggled from abroad, with little evidence on how this happens. We show that a substantial amount of fentanyl smuggling occurs via legal trade flows, with a positive relationship between state-level imports and drug overdoses that accounts for 15,000-20,000 deaths per year. This relationship is not explained by geographic differences in “deaths of despair,” general demand for opioids, or job losses from import competition. Our results suggest that fentanyl smuggling via imports is pervasive and a key determinant of opioid problems

That is from a new NBER working paper by Timothy J. Moore, William W. Olney, and Benjamin Hansen.  One core lesson seems to be that interdiction is largely a futile endeavor.

My favorite non-classical music of the year 2023

by  Tyler Cowen in

I liked these best:

Lankum, False Lankum.  Claims to be Irish folk music, but has ambient textures and is designed to alienate its core audience.

Yaeji, With a Hammer.  A mix of English and Korean, house and hip hop.  She lives in Brooklyn.

Boygenius, The Rest.  Four songs, twelve minutes.

Christine and the Queens, Paranoia, Angels, True Love.  Three CDs, weird, still growing on me.  By some French person.

Paul Simon, Seven Psalms.  Now he is partly deaf, and he was already singing about dying.

Arooj Aftab, Vijay Iyer, and Shahzad Ismaily, Love in Exile.

Cecile McLorin Salvant, Mélusine.  Her track record (and consistency) at this point is simply staggering, and you can put her on a par with the very greatest of female jazz vocalists.

Irreversible Entanglements, Protect Your Light.  From a free jazz collective, still vital.

Your Mother Should Know: Brad Mehldau Plays the Beatles.

Ches Smith and We All Break, Path of Seven Colors.  From the year before, but I discovered it this year, a blend of Haitian vodou and jazz.

I will be doing a separate post on classical music.  What do you all recommend in these categories?

Addendum: And, via Brett Reynolds, here is a Spotify playlist for those.

EconGoat for Claude 2.1

by  Tyler Cowen in

Claude 2’s large context window and easy file upload feature have made this one of the best LLMs.

As of November 21, 2023, Anthropic’s Claude 2.1 has the largest context window of any of the leading LLMs. Paid users can upload and query up to around 150,000 words of text, more than enough to fit this 95,000 word book. But even the free plan allows you to analyze about 75,000 words, which in this book equates to about six chapters.

Here are more details on how to use Claude 2.1 to interrogate GOAT: Who is the Greatest Economist of All Time, and Why Does It Matter?

Does studying economics make you more selfish?

by  Tyler Cowen in

Maybe not:

It is widely held that studying economics makes you more selfish and politically conservative. We use a difference-in-differences strategy to disentangle the causal impact of economics education from selection effects. We estimate the effect of four different intermediate microeconomics courses on students’ experimentally elicited social preferences and beliefs about others, and policy opinions. We find no discernible effect of studying economics (whatever the course content) on self-interest or beliefs about others’ self-interest. Results on policy preferences also point to little effect, except that economics may make students somewhat less opposed to highly restrictive immigration policies.

That is from a newly published paper by Daniele Girardi, Sai Madhurika Mamunuru, Simon D. Halliday, and Samuel Bowles.  This is a good example of a myth that got started with relatively little basis in fact.  At the very least it now has to be filed under “unconfirmed, likely false.”  Via Stefan Schubert.

Follow the science?

by  Tyler Cowen in

We report the results of a forecasting experiment about a randomized controlled trial that was conducted in the field. The experiment asks Ph.D. students, faculty, and policy practitioners to forecast (1) compliance rates for the RCT and (2) treatment effects of the intervention. The forecasting experiment randomizes the order of questions about compliance and treatment effects and the provision of information that a pilot experiment had been conducted which produced null results. Forecasters were excessively optimistic about treatment effects and unresponsive to item order as well as to information about a pilot. Those who declare themselves expert in the area relevant to the intervention are particularly resistant to new information that the treatment is ineffective. We interpret our results as suggesting that we should exercise caution when undertaking expert forecasting, since experts may have unrealistic expectations and may be inflexible in altering these even when provided new information.

Even at current margins, researcher fallibility remains an undertreated topic.  It should be at the center of any approach to method or philosophy of science, rather than the abstract principles we are usually fed.  In any case, that is from a new paper by Mats Ahrenshop, Miriam Golden, Saad Gulzar, and Luke Sonnet.

Via the excellent Kevin Lewis.

AI and real estate revaluations

by  Tyler Cowen

That is the topic of my latest Bloomberg column, and here is one key sentence:

Over time, the more change we see, the more that real estate decisions will be determined by where these in-demand workers want to live.

A general point that is worth keeping in mind over the next few decades.  It applies to more than just real estate, and for that matter, applies to more than just AI.

Saturday assorted links

by  Tyler Cowen

1. “The World Bank estimates that crime costs South Africa 10% of GDP annually.”  Link here.

2. An analysis of Geert Wilders.  People don’t like deductibles!

3. An unreasonable and offensive rant against Millennials.

4. Advance raves for Godzilla Minus One.  And AARP now sponsoring the Rolling Stones.

5. Consider Before Discarding, by Marti Leimbach.

6. Menger mackerel money SBF.

7. AGI and excess backwards induction.  Knowing when to apply backwards induction should be a skill we teach in school.  And speculation about Q*.

8. “The $VIX has fallen 41% over the last 4 weeks (from 21.27 to 12.46), the 9th largest 4-week decline in history.”  Link here.  Lowest in about four years.

9. John Cochrane on dollarization.