←
Hossein Alidaee807 posts
Hossein Alidaee avatar

Hossein Alidaee

@halidaee

Economics postdoc researching info frictions + ext validity in technology adoption at @HarvardHBS. PhD @NorthwesternU. All opinions are wrong, some are useful.

◷ Joined November 2008▧ 10 Photos and videos
Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 14

Related to this point

Maybe in the future alongside a paper submission I'll include a repo with an AI-written FAQ that points out why some things flagged by LLMs are not actually issues, so the reviewer's LLM has that. Like an AI online appendix that dispenses with the false positives while keeping the draft clean and human-readable.

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 14

It would be nice to, alongside a paper, submit a Refine report and just say "here is me addressing each technical point. let's now converge to referees judging responses and otherwise focusing on non-technical issues."

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 11

10% is lower than we think as a cap. Treasury rates are low not only bc lack credit risk but also liquidity and other premia. The more relevant benchmark for opportunity cost of investment is maybe zero beta rate of @KurlatPablo et al. Frequently >10%! tinyurl.com/dt3ayujy

Post attachment

Let’s cap credit card rates at 10%. Take action on affordable housing. Protect rural hospitals. End price gouging. Democrats have the bills ready. @SenWarren says now it’s up to Trump to work with us and get things done.

0:47
Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

@VC31415 Getting two AutoDML working papers applied to labor in one day! Glad to see this method taking off.

Amazing new paper by Chetty and the Opp Insights team — digital twins can recover causal effects They validate the digital twins (ie rich ML applied to generate counterfactuals from lots of covariates) with RCTs, allowing for scalable program evaluation opportunityinsights.org/wp-c…

Post attachmentPost attachment
Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Platforms therefore present the opportunity to do a form of Hayek (1945) in silico. More on this point in the future! 22/22

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Platforms present a different path. Instead of one central model, imagine everyone using a common platform to develop their own model, capturing their own preferences. A central computer could then use these models, paired with constraints, to compute optimal allocations. 21/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Second, these comparisons illustrate a path forward for AI. A lot of discussion around AI and centralization problem is having a single AI compute the equilibrium. (@msiniscalchi I think you were discussing this.) 20/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

(a) Funding many small assessment providers (like some grantmakers are doing in dev context) is less useful because firms need to learn how to map assessment to their needs. (b) Firms each developing models on proprietary assessments would make these comparisons impossible. 19/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Beyond labor, we want to highlight a few implications. First, although this mechanism is different from the data pooling and network externalities documented by others, it gives yet another reason why large platforms add value. 18/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

This is highly valuable to firms. Often, their priors in the low-skill setting is that everyone is going after the same workers. These results show firms are differentiated in which low-skill workers they value, and should differentiate their recruitment strategies. 17/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Our results more closely resemble Lazear. There are a handful of traits with statistically significant derivatives. For these traits, like ethics, one firm associates higher ethics with better retention, while another firm believes retention is worsened. 16/n

Post attachment
Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Because these models are highly non-linear random forests, we don't have clean coefficients to inspect. Instead, using the double machine learning literature, we compute the marginal impact of each trait on retention as we move along that trait's distribution. 15/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Finally, we ask what the source behind firms' distinct preferences are. Do they value distinct traits? Or, per Lazear (2009), do they value the same traits but value them differently? 14/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

These results are v important for policymakers! It implies that programs helping low skill workers with search versus training programs impacting the trait distribution have distinct welfare implications. 13/n

Hossein Alidaee avatar
Hossein Alidaee@halidaee·Nov 3

Doing this reveals heterogeneity in how applications are impacted by this "firm fixed-effect". Applications with low average scores across traits are barely affected; for applications in the top quintile, this explains a fifth of the gains we calculated 12/n

Post attachment