Hikaru Saijo

@lucasian76 I think the rule of thumb is you increase the number of layers until the test errors stop decreasing. Unfortunately I found no good intuition for which activation function to use, although I found LeakyReLu or Tanh tends to perform reasonably well for econ applocations

In particular, I cannot find a reliable link between economic intuition and the right activation function or number of layers. Is a different problem completely as, say thinking with the model to calibrate.

Simple Hopenhayn with DL imitating JFV pytorch codes. I am unable to make it converge to the VFI solution. Not clear at all how to devise the architecture at all. With VFI is way more direct. Still fun to play with it!

@xajpdtpjunpad あとVS codeにはtex extension（tex workshop）もあるので、ドラフトもClaudeにコードを参照してもらったりして書いてもらうこともできます。ここまで来るともはや自分の存在意義が分からなくなりますが（笑）

@xajpdtpjunpad 僕はVS codeにClaude code extensionを入れて作業してるのですが、これだと結構スムーズに文脈伝えられる気がします。Cursorは使ったことないので分からないですが。

Every journal editor should read this: causalinf.substack.com/p/claude-code-…

@t_holden That I don’t know. Never tried git:)

@hpfilter Does it give you the nice per line change accept reject even without git commits like antigravity?

This seems a good time to remind people that Antigravity provides a far more natural way to work than Claude Code. The entry cost is far lower. It is particularly suited for tasks for which clean tests of correctness don't exist (inc. econ tasks). Then you want to see the code!

Yes thank you

Saturday, I gave a talk at ASSA 2026 on my new paper ``Macroeconomic Effects of Production Networks: A Deep Learning Approach,’’ with @cmatthes_econ, @hpfilter, and @MacroInPieces. Several attendees asked for a copy of the slides I used (sorry, we are still a few weeks away from circulating a completed draft), so I am linking them here: sas.upenn.edu/~jesusfv/Slide… There has been a lively debate over the last few weeks on X about what deep learning can do in macro. I think this is a good example of what we can now study. Dynamic stochastic models with non-trivial input-output structures are a beast to compute. They are inherently nonlinear, and you need to carry not only aggregate state variables but also all the sector-specific variables, such as sectoral productivity and sector-specific capital. If you have, as we do in the paper’s calibration to match U.S. input-output tables, 37 sectors, you end up with at least 74 state variables. Even very aggressive sparse-grid allocations, such as Smolyak, will struggle with that. A deep neural network can deal with this problem, especially if you build it using the symmetry (or, if you prefer, exchangeability) ideas I discuss here: nber.org/papers/w28981 with @MahdiKahou, @jlperla, and Arnav Sood. We also show that adopting a deep learning approach changes the answers you get relative to more traditional solution methods. More concretely, we capture endogenous changes in relative prices across sectors that perturbation solutions largely miss. That leads to materially different elasticities of output, investment, and labor supply to sectoral shocks. In our benchmark calibration, we find substantial attenuation of sectoral shocks, precisely because of these relative price changes. When steel becomes too expensive following a productivity shock in the steel industry, car manufacturers switch to more plastic. This result resonates with my own earlier work on the macroeconomics of wars. One lesson from wars over the last 120 years is how resilient modern economies have been to large disruptions from mobilization, blockades, or bombing. What looked like bottlenecks that would stop production turned out to be inconveniences, because there are many ways to produce a tank or an airplane. Some alternatives were less efficient than the optimal one, but the productivity losses were not as large as pre-war planners envisioned when they took a static view of resource allocation. Our results replicate that intuition inside a standard stochastic multi-sector business cycle model. Of course, we could throw wrenches into the economy’s reallocation process, but, at this moment, I think the insight is robust.

@xajpdtpjunpad 単に鍵付きの人をブロックしているだけでは？僕の憶測ですが。

@gp_mihalache Yes I think so

@cmatthes_econ Frank Schorfheide alert

I just had a scary thought: When I talk to junior people today about Matlab code, do they feel the same way I did back in the day when people would show me their Gauss code?

New paper with @CosminIlut and @hpfilter. Under Smooth Diagnostic Expectations, agents' over-reaction depends on the level of uncertainty. This provides a parsimonious explanation for over-reaction and over-confidence and key business cycle facts

Embedding smooth diagnostic expectations in a real business cycle model accounts for overreaction and overconfidence as well as key properties of the business cycle, from @Francesco_Bia, @CosminIlut, and @hpfilter nber.org/papers/w32152

We (Yuichiro Kamada, Toshihiko Mukoyama, and I) will tweet about Japanese job candidates! This is an initiative by a few volunteers of Japanese economists. We were helped by many senior Japanese colleagues when we were juniors/grad students.

#EconTwitter, I'm super excited to say I'm finally on the #EconJobMarket ! I'm a #EconJMC from @ucsc Ever wonder how Chinese monetary policy matters on the global stage? Answer? Supply Chains! Here's a brief 🧵on my JMP, coauthored with my classmate @anirbansanyal83

@ToshiMukoyama 同感です。というか、オフィスがもらえないのは流石に辛いですね。僕の意見はオフィスがもらえる、という前提のもとです。

@hpfilter 共著まで行かなくても毎日学校で顔を見ていれば何となく「同僚感」が出てくるし、コンファレンスに行ったりしても（一度会っただけとかではなくて）そういう馴染みの知り合いがいると気が楽とかありません？　カレッジタウンにいても週末観光に出て行くことはできるわけだし。

僕の意見はちょっと違って、僕はUVAにいた時にはもっとこういうところに来ればいいのに、と思っていた。カレッジタウンの方がオフィスももらいやすいし、人が学校に出てくるのでなじみやすい。（車が運転できない人には大変かもしれないけれど。）サバティカルの成果が観光っていうのは寂しいと思う。

（１）多少無理目なジャーナルかな、と思っても気にせず投稿してみても良いかと。可能性無しならデスクリジェクトされるし、良いところに掲載された時のメリットはかなり大。例えばトップフィールドに一本ある人とフィールド２番手に二本ある人だと前者の方が高く評価される傾向にあると思う。

@carlos_ed_sg Very nice. I never had a good intuition for hump-shaped consumption (and never have seen it written up).

dropbox.com/s/ca5do2wanqkw… Short informal note on two things you might have overlooked when studying the RBC model (the ultra bare bones versions of it) 🧐

@xajpdtpjunpad Juliaは使ったことないですが、PythonだとPytorchがあったりして確かに結構充実している印象ですね。