Just received a scan of a thank you card from a student who's starting his PhD in IS at Maryland. It made my day. Also I realized that a scan of handwriting somehow touches me more than an email.
Special thanks to
@dbroockman
@avicgoldfarb
for kindly sharing their course slides.
Our paper on when and why do buyers rate just came out as a CEPR Discussion Paper. This is joint work with my brilliant coauthors
@kleintob
and Konrad Stahl.
Why another paper on this old topic? In this thread, I'd like to highlight the novelty in our findings 1/9
Study by
@ce_tucker
@ccatalini
: last mile problem between digital and physical world. Some comments argue that technology, e.g. retina scanning, can solve the verification problem. But technology alone never solves the incentive problem for truth-telling!
We find that buyers dislike search ads, but quality certificates mitigate this aversion. Results based on a field experiment which randomizes the salience of ad disclosure, and a system glitch which eradicates a quality certificate for all listings on eBay. Joint w/
@MengLiu_WUSTL
My friend always went the extra mile at work. That's why he lost his job as a taxi driver :) Check out this new MS paper on how digital platforms mitigate moral hazard in the transportation industry by
@MengLiu_WUSTL
@erikbryn
@JasonFKD
.
Uber vs. taxis, which one is more likely to rip you off by taking you on a detour? Our new paper in Management Science shows Uber drivers do better on this front, except in times of high surge pricing. Joint work with the amazing
@erikbryn
and
@JasonFKD
Why did Google's ads platform switch to using first-price auctions? Can an auction be credible, strategy-proof, and static at the same time? Check out this awesome thread on the paper by
@akbarpour_
and
@ShengwuLi
.
A very interesting paper by Meng Liu,
@erikbryn
, and
@JasonFKD
that shows Uber passengers are less likely to be taken on a detour and ripped off than taxi passengers.
New working paper from
@mit_ide
on the spillover effect of shelter-in-place policies across geographies and networks.
(Another reason why a simple DiD across geography to study COVID-related shocks does not work.)
[🚨NEW WORKING PAPER🚨] As local governments reopen, we study how policies in one region affect mobility & social distancing in other regions & the consequences of uncoordinated policy responses to
#Covid_19
. 1/
👏👏50k+ upvotes & 4k+ comments on Reddit in 1 day. Forecasting the spread of COVID by
@MengLiu_WUSTL
@RThomadsen
@songyao
(and those who attacked the UW model in comments, this one is from Washington University, which is in St Louis!!! 😫)
@KirbyKNielsen
I feel you. I had no time for research when I taught a new course for the first time. I designed the first five courses from scratch but eventually I was fortunate to discover a related course from a kind instructor and have built my course based on that since then. Huge relief!
Hong Kong postpones election for a year citing the pandemic concern. If the government is so concerned why isn't it restricting travel or enforcing work from home? Trump must be so jealous...
Happy to announce our website (Meng Liu, Song Yao and myself) that gives our forecast for the spread of covid under different social distancing levels. We also provide graphs of what the social distancing levels are in each state.
@wytham88
@vkbostwick
@SarahHBana
Hey folks this is awesome, thanks for doing this. To me whether a postdoc involves teaching or not is a BIG factor. Teaching a new prep involves lots of time devotion even for very good teachers and it's difficult to recoup the benefit within the short time span of postdoc.
I do a lot of interdisciplinary work. Here are 4 hard lessons I've learned. They may help you be more successful (than me) if you want to work across fields.
@deaneckles
@AndreyFradkin
@benthompson
@johnjhorton
In addition to great answers in the thread, a tiered loyalty program that gives greater future value for sellers with larger number of transactions on the platform could help.
Having a very bad feeling that this guy is going to take full advantage of his case by arguing that covid is nothing, the democrats have overreacted, and he has been right since the beginning.
@avi_collis
@wytham88
Yep, and one needs to retake written and road tests if the time it takes to renew the status (which can be outside our control) is longer than the grace period that DMVs set. My guess is more internationals needed to retake the exams/get a new license after the Covid disruptions.
We know these cases b/c seller needs to pick a reason for rating revision, and buyer needs to accept it. B/c seller has *1* chance for revision, it's unlikely for seller to pick the 2nd reason, implying buyer fault,if that's not true.(this holds even if there's seller bribery)6/9
@pauljhealy
I'm not sure but I doubt it. I think recruiters at top chinese schools know the hiring tactics that their competitors use from their network or the candidates they interview. I think most people are reasonable and will sympathesize with candidates that get shady offers.
@deaneckles
Putting aside all the endogeneity problems, it's very hard to interpret the result without knowing what people do to replace time on FB. Do they switch to Tiktok? Do they start baking? Perhaps the takeaway is that watching more dog videos on tiktok releases stress. Who knows
@OrgStratProf
The one due soon? Journals have two important stats on new submission: average decision time and %decisions above some threshold. If you were a journal, then finish the one due soon doesn't change average decision time but improve the second measure.
@KirbyKNielsen
Sorry I meant five lectures in a single course. I would not have survived prepping for five courses 😂 good luck with your spring semester!
@pauljhealy
Never seen a conditional interview request like this. I've heard of schools that would extend an offer only if the candidate verbally commits to come. As far as I know reneging any informal contract has no legal consequence in china.
@deaneckles
@erikbryn
Thanks
@deaneckles
. Indeed, language barrier should cause even more frictions for offline trade, as it will not only affect search and communicating product attributes, but also establishing the entire trade infrastructure, eg trust,logistics, payment etc
@erikbryn
@DigEconLab
In cases where not knowing fakes poses an externality, e.g. fake news, we need mandatory disclosure. But in other cases, disclosure if asked is good since some people who enjoy interacting with robots don't always want to be reminded that they are not humans, e.g. companionship.
@KurtLavetti
@pauljhealy
The causal question would be: if I score 1 pt higher, all else equal, how will that affect my placement. The question doesn't quite make sense to me, and my guess is a null effect based on logic. Just my two cents tho.
@sebbie88
Once I bought UVXY, it dropped by 40% on the first day, up by 40% the second day, drop by 50% the next day, and up by 50% on the fourth day. Sad story...
@danielascur
Your notes remind me of Laga distillers edition 2001. Because it's double matured it has an extra layer of sweetness from sherry casks, and the smoke becomes much more subtle (compared to Laga 16). It's one of my favorite islay scotch. Is the scotch you reviewed sherry casked?
@marcfbellemare
@KirbyKNielsen
@SarahHBana
These are great points. I wanna add that even if ratings do not give extra predictive power cndtn on revealed pref, firms may still want to learn the correlation matrices to hedge the risk of data regulation that may prohibit tracking of some but not all types of personal data.
@grazzidad
@gzervas
Interesting. I wonder how quality of the papers changes over time. Also an inverse U? when you start off, editors from good journals arent sure of you so you don't get good papers. This improves over time, but when you become an editor, the quality drops bc of a larger pool.
First, a quick summary of our paper: 1.we write a Bayesian model of buyer’s rating decision. 2.we test model predictions using eBay data. 3.based on calibrated model, our simulation suggests that buyers learn about seller type in ~100 transactions even if ratings are biased. 2/9
I really like the list of case studies of where data misleads, since it give you a chance to test your reasoning about data. For example: why does this data show rap & metal musicians die so much earlier than blues & jazz musicians?
@SeilerStephan
@grazzidad
@gzervas
That's good to know. It happened to me as well. I don't think it's you mellowing out. I think it's bc asymmetric information about your quality as a reviewer and as a researcher gradually reveals.
@Prof_Yildirim
Ex ante, there's a tradeoff between explaining for a longer time vs explaining for a shorter but involving more thinking. If you mean ex post, I guess it depends on their prior belief of whether a not so simple communication is sufficient to get the idea though.
Intuitively, although buyers see very positive aggregate seller ratings, they do not interpret it literally to the extent that they understand the process through which ratings are given. Therefore, ratings can be still quite informative despite it being biased. 8/9
The "subtle" difference leads to dramatically different predictions. E.g. suppose a very reputable seller receives a negative rating, our model predicts a higher likelihood of receiving any rating, especially so for a negative rating. The other model predicts the opposite. 4/9
@yesimorhun
I find making smoothies stress relieving. I enjoy taking some time to wash, cut, and blend different veges and fruits. I know some colleague find vacuuming in the morning calming. Hope you'll find something that works for you!
@gzervas
I see. I've never heard or seen the use of this word in my region (northern china). Anyways, the bread you made looks soft and delicious. And nice picture!