@ohadsamet
A generous rebuttal would be that those who bear risk (entrepreneurs) enjoy the fruits that the poster craves. A darker reality may be that generational wealth is impenetrable.
@ohadsamet
There's something in here about being a salaried employee not being enough to break into the upper echelons of American society. The poster experiences anxiety from having this realization after years of being top 1% in school and work.
I took a class this quarter that basically only covered one Econometrica paper, and it's been paradigm changing. The paper is Farrell, Liang, and Misra (2021), "Deep Neural Networks for Estimation and Inference".
Hi
#EconTwitter
! 📊
Interested in the intersection between causal inference, econometrics and machine learning?
Check out 👇 this fascinating 90' tutorial by Bernard Koch (
@UCLA
) on deep learning for causal inference.
Perfect for those exploring how neural networks can
“You have all these economists with 160 IQs that spend their life studying it, can you name me one super-wealthy economist that’s ever made money out of securities? No.”
— Warren Buffett
my Indian origin and Indian descent friends are surprised to learn that Bangladesh has a higher GDP per capita (and life expectancy, and gender equality, etc.) than India. apparently, the perceived wisdom is that Bangladesh is to India what Mexico is to the U.S.?
@noor_siddiqui_
“All sample test questions were screened, and questions containing visual assets such as clinical images, medical photography, and graphs were removed.” Yeah, this is selection bias.
Chicago Booth PhD Program gets $100mm. What? Just the PhD program!?
@KhoaVuUmn
we have to pay for our own Stata licenses. how many licenses can we get paid?
Today marks the second anniversary of Yiran Fan's passing. As most of our cohort is on the job market this year, I'm reminded of the theft of such a brilliant and generous mind from our community. Yiran, you are deeply missed.
Reminder (for econ/finance PhDs): in the time it takes to get tenure, you can save enough money from an industry job (assuming you're working comparable hours) to FIRE (financial independence, early retirement).
Helpful calculator:
Oh man, there's so much cool math I still want to learn. Irony of grad school (and academia) is that it selects for the naturally curious ex ante but rewards the hyper-focused and hyper-specialized ex post. My biggest struggle has been to square the ex ante and ex post selves.
In undergrad I was exposed to the covariance operator and spent too long trying to understand how it relates to stochastic processes. After many years (in my defence, not my field...) I have found just the exercise I need!
My favorite heuristic derivation of the HJB is from Dixit's Art of Smooth Pasting. Discrete time approximation is intuitive and allows you to inspect the mechanics of the dynamics. The Markovian nature of the HJB is also emphasized, and it prepares you for numerical optimization.
Another fun example of 'street math' from yesterday's Anki review. From Fernandez-Villaverde and
@NunoGalo
's excellent continuous-time notes.
Deriving the HJB using Bellman's principle of optimality and dividing by 'dt.'
MBA ✅
PhD left to go 💪🏾
One of the huge perks of the
@ChicagoBooth
Finance PhD program is the ability to concurrently complete the MBA. (David Booth was enrolled in the PhD program and famously left with the MBA.) I haven't seen this offered by any other program.
Another fun example of 'street math' from yesterday's Anki review. From Fernandez-Villaverde and
@NunoGalo
's excellent continuous-time notes.
Deriving the HJB using Bellman's principle of optimality and dividing by 'dt.'
Here's an example of 'street math' that came up today in my Anki reviews: the heuristic derivation of a viscosity supersolution condition (v(x) has a convex kink at x). See slides 9-12 of
@ben_moll
's excellent primer on the topic: .
Open Source Bond Asset Pricing (,
@dickerson_phd
) is an essential resource for those interested in US corporate bond markets and/or working with Enhanced TRACE data.
TRACE cleaning scripts: .
Related paper:
Late to the party, but I just read this great article by
@toniwhited
on reconciling the structural and reduced-form tribes:
Especially relevant to me since I believe in the merits of both and hope to present both approaches to answer a question in my JMP!
Here's another excellent syllabus with perhaps the most comprehensive textbook references I've seen across stochastic calculus, control, and mean field games:
- Payne
#EconTwitter
do you have/know of any syllabus on continuous-time macro and macro-finance? Please share!
To start:
-
@MarkusEconomist
-
@NunoGalo
- Matthieu Gomez
- Michael Sockin
#EconTwitter
do you have/know of any syllabus on continuous-time macro and macro-finance? Please share!
To start:
-
@MarkusEconomist
-
@NunoGalo
- Matthieu Gomez
- Michael Sockin
#EconTwitter
here's my survey of deep learning approaches to estimating CONTINUOUS TIME macro and macro-finance problems (alt to upwind FDM). The common benchmark models are BruSan (for macro-finance) and Huggett-Bewley-Aiyagari + Krusell-Smith (for het macro). Please contribute!
Leaving
@jpmorgan
after a promotion to Associate and great performance reviews to RA at the
@federalreserve
for 2 years before doing a PhD in 7 years.
Second maybe was rolling over my entire traditional 401K balance to a Roth IRA after markets tanked in 2020 because of COVID.
I'm hooked on profiles of academics who were late-bloomers or those who overcame adversity. Can you drop some links here? Reading these gets me amped.
Examples:
June Huh:
Yitang Zhang:
(and more recently, )
Interesting take on Einstein's infamous job market woes (toiling away at a patent office after failing to secure an academic position) actually facilitating his breakthroughs by freeing him from publication pressure.
Think
#EconTwitter
will be shocked to learn that many Stats PhD programs, including the one at UChicago, no longer require you to take grad measure theory.
I'm not submitting to SED 2024, but I find the submission statistics extremely fascinating to get sense of what's trending.
Heterogeneous agent macro and macro-finance at the top!
Surprisingly, machine learning and AI dead last. Curious for a conference on dynamics...
Congrats
@JohnHCochrane
for writing a timeless classic! The figure below from the text was very important in my personal intellectual development. A big picture A-ha moment connecting factor models to equilibrium notions of pricing.
ChatGPT is a game changer, and it's impact on my productivity is indisputable. Instead of manually downloading files from an atrocious javascript object, I used ChatGPT to write Python code using Selenium to do this for me. What could have taken me an hour, took me two. 😤
I'd also recommend Machine Learning for Factor Investing. Print edition was published in 2020, but the free online edition is continuously updated. Code in both R and Python.
Print:
Online:
Python:
@SJosephBurns
59. I"m shocked people are responding 9. Also, many incorrect interpretations of PEMDAS. Let's walk through it.
Begin with 50+10*0+7+2
P? No
E? No
M? Yes. 10 x 0 = 0. Updated expression is 50+0+7+2
D? No
A? Yes. Work left to right. 50+0 = 50. 50+7=57. 57+2=59.
S? No.
Answer: 59.
#EconTwitter
Is 'Spatial Finance' a thing? Studying the geographic distribution of bank branches seems like an obvious topic (say, 'Bank Branches across Space and Time'). But other obvious topics include dispersion in capital and entrepreneurial activity, etc.
#econtwitter
is Yong and Zhou (1999) still the best resource to study the links between recursive and non-recursive solutions to stochastic control problems?
Disagree. Performing well in 1st and/or 2nd year courses will set you up to TA and RA for future advisors. Advisors matter the most for PhD success, from what I've seen, and the feedback from advising to paper quality is strong (also cue endogeneity).
2. College grades matter for getting into econ/finance PhD. Some PhD students actually work hard in 1st year and try to ace their classes, not realizing that this basically has 0 direct effect on whether you'll find a job after PhD, it's all about writing papers
This looks like a pretty awesome, free textbook on deep learning:
Part VI covers deep learning for partial differential equations, including PINNs in Chapter 16.
I've encountered this own irritation with my own research. I would love to spend some time at the
@ecb
for a project, but the pathways I've identified are blocked to US citizens (among others). The irony here is that at the
@federalreserve
, I met more Europeans than Americans!
@ShengwuLi
Is it even possible to be a polymath in this day and age? Like quite literally bump up against the physical constraints of thoroughly knowing the relevant literatures?
The best part about working on theory/structural models is that I'm actually learning/discovering new economics from the very act of research. Don't really experience this with empirical research; there it's more about learning what implicit model the real world satisfies.
Had an hilarious exchange with a professor yesterday.
Prof: Are you a Bayesian or a frequentest?
Me: I'm a desperate grad student trying to graduate.
Prof: Okay, you're a Bayesian then.
And with that I was freed from the tyrannous yoke of frequentest inference.
.
@ChicagoBooth
provides a stark example of this, quite literally reflected in the name of the school, the doctoral program, and on dedication plaques outside lecture halls.
A more pithy statement: Booth - The House Fama Built.
Great research round-up on approaches to solve nonlinear PDEs in the conclusion of
@UncertainLars
,
@PaymonKhorrami
, and Tourre (2023). Paper itself is an excellent didactic guide to macro-finance models.
Woke up this morning with a proof in my head for a problem that I was stuck on last night. Worked out some details while still in bed and then got to writing. Peak academia. 😂
Let me get this straight JF and AR didn’t expose Gino when given the chance, and Gino capitalizes on this to effectively redirect fraud accusations their direction. Stuff to make Machiavelli proud.
I don’t care if it’s real analysis in particular. But one has to be deeply incurious not to take a proof-based math course before starting an econ PhD.
Kind of crazy to think that I may be in the Chetty et al. dataset. I was on free/reduced lunch in HS, matriculated
@Columbia
in 2008, was a Pell Grant recipient, graduated in 2012, and worked at
@jpmorgan
between 2012 to 2015.
I'm so impressed by this textbook. It's both accessible and exhaustive. Moreover, it's practical with the inclusion of sample codes.
But the authors don't stop there! There's an abundance of additional material, including courses based on the book:
Real Analysis touches every facet of the foundational knowledge in micro, macro, and metrics, if not directly, then surely indirectly. That said, it's clear that applied research far and away dominates and insisting on a RA filter is inconsistent with actual job market realities.
I agree that multivariate calculus should be a pre-requisite for economics PhD (and undergrad majors should know that), but the real analysis req is a bit much. And that’s coming from someone who loved that class (taken while a research assistant).
Semi-serious question, but has there been any other PhD program and dissertation area combination with a higher return for a school than Chicago GSB/Booth Finance? Return measured as Donations/Stipends?
(If the powers that be are reading these, please keep this in mind :P.)
Chicago Booth PhD Program gets $100mm. What? Just the PhD program!?
@KhoaVuUmn
we have to pay for our own Stata licenses. how many licenses can we get paid?
This has got to be the all-time greatest LinkedIn update on my feed!
I had the pleasure of working closely with
@SantiPenap
in Paraguay while at
@jpmorgan
. A very transformative experience.
Always a good feeling to see a technocrat and economist holding high political office!
What was Kissinger's greatest triumph? His and Nixon's breakthrough in diplomatic relations with China.
What was the cost? Genocide in Bangladesh.
Highly recommend "The Blood Telegram" by Bass. Archer Blood gave up his career to voice dissent.
"The Blood Telegram" by
@Gary_Bass
is Kissinger at his most Machiavellian, a devastating and disturbing account of how he turned a blind eye to the genocide in Bangladesh to preserve his Pakistani channel to China.
My theory on why GRE went from a (ironically) coarse 800 scale to a finer 170 scale in ~2011 (and allow selective reporting) is that it would precisely allow them to report tighter cutoffs which programs would endogenously respond to, promoting exam retakes by applicants.
A 168 in the GRE Quant section is no longer above 90th percentile. Most Econ PhD programs say that to be competitive the quant score should be above 90th percentile. Does anyone know how Econ PhD Adcoms are responding to these percentile changes, if at all?
#EconTwitter
PhD rejections:
Harvard Econ
MIT
Berkeley
Michigan
Yale
Penn
Minnesota
NYU
MIT-Sloan
Northwestern
Princeton
Stanford
Booth
Acceptances:
Harvard BusEc
Columbia
Wisconsin
Waitlist:
UChicago (no funding)
And now we have precedent for banks to take on duration risk in low interest rate environments via Treasuries and other AAA public debt securities. You basically get a free put when interest rates rise and you accumulate unrealized losses.
It’s official.
@chicagobooth
Econ PhD had a great year! 💪💪💪🥳🥳🥳
My cohort was the first of larger cohorts. Booth Econ PhD previously only had 0, 1, 2 students a year.
Booth Finance and other well-established Booth programs placed great as always:
Today marks the 3rd year since Yiran's passing. It's still with great shock and sadness to see him as an entry in Chicago's dataset on shooting victims. Yiran was only in his early 30s, and it's heartbreaking to see how young the distribution of first degree murders skews.
I have comments in my code, but this is still very impressive. also, I can code in Matlab and just use ChatGPT to translate to julia for speed gains? that's amazing.
1st order effect of ChatGPT will be to make knowledge workers more productive, not replace them entirely.
One upshot of the Fed (and US broadly) generally not having onerous nationality restrictions (e.g., ECB) is that policymaking talent here is very deep (and it shows). You could staff the entire world's central banking leadership from tapping Federal Reserve System economists.
I am not sure what it says exactly but I have heard the news: Fatih is the new governor of the Central Bank of Turkey!!! Congratulations
@yfatihkarahan
, your NY Fed former colleagues are so proud of you!!!
I've seen so, so many hyper critical takes on the Fed, but I'm honestly impressed? If a soft-landing is pulled off with stable inflation, record low unemployment, and continued real wage growth, then will we revise our narrative?
NEW:
CPI Inflation declined to 3% year-on-year, the lowest level since March 2021, growing 0.2% month-on-month
Core CPI inflation declined to 4.8% year-on-year, the lowest level since November 2021, growing 0.2% month-on-month, the lowest since August 2021
I like to scan enrollment in second year field courses to get a rough leading indicator of future JMPs. Areas that stand out (at least at Chicago, broadly):
- IO
- Experimental Economics
- Spatial Economics
If these ‘national IQ’ data float across your feed, please note they are pseudoscience. The dataset was developed by a guy given a sword by his psychologist mates for his “long-term services to eugenics”, which is political ideology, not science
It's interesting that
@QuantEcon
(
@john_stachurski
) has dedicated lectures on continuous-time Markov chains. But as suggested below, this structure emerges in the solution of the HJB (process at i jumps to i-1 or i+1). From this you get transition rates!
My favorite heuristic derivation of the HJB is from Dixit's Art of Smooth Pasting. Discrete time approximation is intuitive and allows you to inspect the mechanics of the dynamics. The Markovian nature of the HJB is also emphasized, and it prepares you for numerical optimization.
Cool workshop (on mean field games among other things?) organized by Fernando Alvarez,
@UncertainLars
, and Takis Souganidis scheduled for March.
Methods for Solving and Analyzing Dynamic Models in the Face of Uncertainty and Cross-Sectional Heterogeneity
Random presentation advice for econ/finance PhD students:
Your paper is a circle. The circle marks the territory you want to claim: your paper's contribution. Your goal is to:
1. Draw a moderate-sized circle
2. Defend your circle
Very impressed by ChatGPT's ability to guess a functional form for the solution of the HJB arising from the canonical Merton portfolio choice problem. Of course, this is the best case scenario for ChatGPT to solve HJBs with known closed-form solutions but still.
@t_holden
So the NY Fed has some bureaucratic obligation to make a public posting but really wants to hire one specific individual? And this imposes a negative externality on nervous job seekers who really have to apply to any job, remotely relevant, because of how thin the market is?
@noor_siddiqui_
Also, you want to compare the accuracy of ChatGPT against the accuracy of human responses over a similar selection of questions. We have no idea whether that would still be the same threshold. Far stretch to conclude that ChatGPT is more competent than humans on the USMLEs.
#EconTwitter
any suggestions? Here's some off the top of my mind:
- Tarek Hassan (BU): Asset Pricing PhD half-course and International Macroeconomics and Finance
-
@MarkusEconomist
: Lecture 13 of Princeton's ECO 529:
For a more mundane, but perhaps more common story of immigrant success, let me point you to the Bangladeshi community in Florida in the 1980s. Emerging from a country ravaged by war and famine, many entered the U.S. on student visas, which they overstayed, without graduating.
Today is the 30th anniversary of when my dad arrived in America. He came with 2 suitcases and 0 savings, knowing just 1 person in town. My mom and I followed 8 months later, with no English and leaving everything and everyone behind.
Very interesting set of facts that I didn't know before. From the Survey of Consumer Finances (SCF), couples with children have higher incomes than couples without children but have considerably lower net worth!