昨日紹介したAngrist=Pischke論文にCarola Binderも反応し、その要旨を引用している。

In the 1960s and 1970s, an empirical economist’s typical mission was to “explain” economic variables like wages or GDP growth. Applied econometrics has since evolved to prioritize the estimation of specific causal effects and empirical policy analysis over general models of outcome determination. Yet econometric instruction remains mostly abstract, focusing on the search for “true models” and technical concerns associated with classical regression assumptions. Questions of research design and causality still take a back seat in the classroom, in spite of having risen to the top of the modern empirical agenda. This essay traces the divergent development of econometric teaching and empirical practice, arguing for a pedagogical paradigm shift.




One is a focus on causal questions and empirical examples, rather than models and math. Another is a revision of the anachronistic classical regression framework, away from explaining economic processes and towards controlled statistical comparisons. The third is an emphasis on modern quasiexperimental tools.




This advice on teaching regression resonates with my experience co-teaching the economics senior thesis seminar at Haverford for the past two years. Over the summer, my research assistant Alex Rodrigue read through several years' worth of senior theses in the archives and documented the research question in each thesis. We noticed that many students use research questions of the form "What are the factors that affect Y?" and run a regression of Y on all the variables they can think of, treating all regressors equally and not attempting to investigate any particular causal relationship from one variable X to Y. The more successful theses posit a causal relationship from X to Y driven by specific economic mechanisms, then use regression analysis and other methods to estimate and interpret the effect. The latter type of thesis has more pedagogical benefits, whether or not the student can ultimately achieve convincing identification, because it leads the student to think more seriously about economic mechanisms.






Francis Dieboldが、Joshua D. AngristとJörn-Steffen PischkeのNBER論文噛み付いている(H/T Economist’s View)。


How should changes in our use of econometrics change the way we teach econometrics?

Our take on this is simple. We start with empirical strategies based on randomized trials and quasi‐experimental methods because they provide a template that reveals the challenges of causal inference, and the manner in which econometric tools meet these challenges. We call this framework the design‐based approach to econometrics because the skills and strategies required to use it successfully are related to research design. This viewpoint leads to our first concrete prescription for instructional change: a revision in the manner in which we teach regression.

Regression should be taught the way it’s now most often used: as a tool to control for confounding factors. This approach abandons the traditional regression framework in which all regressors are treated equally. The pedagogical emphasis on statistical efficiency and functional form, along with the sophomoric narrative that sets students off in pursuit of “true models” as defined by a seemingly precise statistical fit, is ready for retirement. Instead, the focus should be on the set of control variables needed to insure that the regression‐estimated effect of the variable of interest has a causal interpretation.






Here's what AP get right:

(G1) One of the major goals in econometrics is predicting the effects of exogenous "treatments" or "interventions" or "policies". Phrased in the language of estimation, the question is "If I intervene and give someone a certain treatment ∂x,x∈X, what is my minimum-MSE estimate of ∂y?" So we are estimating the partial derivative ∂y/∂x.

AP argue the virtues and trumpet the successes of a "design-based" perspective for achieving G1. In my view they are largely correct as regards G1. ...the econometric simplicity of design-based methods is tremendously intoxicating. It's mostly just linear regression of y on x and a few cleverly-chosen control variables -- you don't need a full model -- with White-washed standard errors. ...

Here's what AP miss/dismiss:

(G2) The other major goal in econometrics is predicting y. In the language of estimation, the question is "if a new person i arrives with covariates Xi, what is my minimum-MSE estimate of her yi? So we are estimating a conditional mean E(y|X), which in general is very different from estimating a partial derivative ∂y/∂x.

The problem with the AP paradigm is that it doesn't work for goal G2.









*1:これはおそらく、whitewashという一般用語と、論文でその手法が参照されているHalbert Whiteを掛けている。




と題した論説をJoel Mokyrノースウエスタン大教授が書いている原題は「How Europe became so rich」;H/T Mostly Economics)。

How and why did the modern world and its unprecedented prosperity begin? Learned tomes by historians, economists, political scientists and other scholars fill many bookshelves with explanations of how and why the process of modern economic growth or ‘the Great Enrichment’ exploded in western Europe in the 18th century. One of the oldest and most persuasive explanations is the long political fragmentation of Europe. For centuries, no ruler had ever been able to unite Europe the way the Mongols and the Mings had united China.

It should be emphasised that Europe’s success was not the result of any inherent superiority of European (much less Christian) culture. It was rather what is known as a classical emergent property, a complex and unintended outcome of simpler interactions on the whole. The modern European economic miracle was the result of contingent institutional outcomes. It was neither designed nor planned. But it happened, and once it began, it generated a self-reinforcing dynamic of economic progress that made knowledge-driven growth both possible and sustainable.

How did this work? In brief, Europe’s political fragmentation spurred productive competition. It meant that European rulers found themselves competing for the best and most productive intellectuals and artisans. The economic historian Eric L Jones called this ‘the States system’. The costs of European political division into multiple competing states were substantial: they included almost incessant warfare, protectionism, and other coordination failures. Many scholars now believe, however, that in the long run the benefits of competing states might have been larger than the costs. In particular, the existence of multiple competing states encouraged scientific and technological innovation.


近代世界とその前例の無い繁栄はどのように始まったのだろうか? 近代の経済成長の過程、ないし「大繁栄」が、なぜどのように18世紀の西欧で急進展したかの説明については、歴史学者経済学者政治学者、およびその他の学者による浩瀚学術書が数多く出版されている。その中で最も古くかつ説得力のある説明は、欧州の長期に亘る政治的分断である。何世紀もの間、蒙古や明朝が中国を統一したように欧州を統一できた支配者はいなかった。


それはどのように機能したのだろうか? 簡単に言えば、欧州の政治的分断が生産的競争を促進したのである。そのため、欧州の支配者たちは、最良かつ最も生産的な知識人と職人を求めて相争った。経済史学者のエリック・L・ジョーンズは、これを「複数国家システム」と呼んだ。欧州が複数の互いに競合する国家に政治的に分かれたことの費用は高くついた。ほぼ絶え間なく続く戦争、保護主義、その他の協調の失敗などがその費用に含まれる。しかし多くの学者は、互いに競争する国家がもたらした便益は長期的には費用を上回ったのではないか、と今日では考えている。とりわけ、複数の競争する国家の存在は、科学および技術のイノベーションを促した。


A possible objection to this view is that political fragmentation was not enough. The Indian subcontinent and the Middle East were fragmented for much of their history, and Africa even more so, yet they did not experience a Great Enrichment. Clearly, more was needed. The size of the ‘market’ that intellectual and technological innovators faced was one element of scientific and technological development that has not perhaps received as much attention it should. In 1769, for example, Matthew Boulton wrote to his partner James Watt: ‘It is not worth my while to manufacture [your engine] for three counties only; but I find it very well worth my while to make it for all the world.’

What was true for steam engines was equally true for books and essays on astronomy, medicine and mathematics. Writing such a book involved fixed costs, and so the size of the market mattered. If fragmentation meant that the constituency of each innovator was small, it would have dampened the incentives.

In early modern Europe, however, political and religious fragmentation did not mean small audiences for intellectual innovators. Political fragmentation existed alongside a remarkable intellectual and cultural unity. Europe offered a more or less integrated market for ideas, a continent-wide network of learned men and women, in which new ideas were distributed and circulated. European cultural unity was rooted in its classical heritage and, among intellectuals, the widespread use of Latin as their lingua franca. The structure of the medieval Christian Church also provided an element shared throughout the continent. Indeed, long before the term ‘Europe’ was commonly used, it was called ‘Christendom’.


If Europe’s intellectuals moved with unprecedented frequency and ease, their ideas travelled even faster. Through the printing press and the much-improved postal system, written knowledge circulated rapidly. In the relatively pluralistic environment of early modern Europe, especially in contrast with East Asia, conservative attempts to suppress new ideas floundered. The reputation of intellectual superstars such as Galileo and Spinoza was such that, if local censorship tried to prohibit the publication of their works, they could easily find publishers abroad.








It is interesting to note that the advances in science were driven not only by the emergence of open science and the growing sophistication of the transnational market for ideas. They were also driven by the appearance of better tools and instruments that faci­litated research in natural philosophy. The most important ones include the micro­scope, telescope, barometer and modern thermometer. All of them were developed in the first half of the 17th century. Improved tools in physics, astronomy and biology refuted many misconceptions inherited from classical antiquity. The newly discovered notions of a vacuum and an atmosphere stimulated the emergence of atmospheric engines. In turn, steam engines inspired scientists to investigate the physics of the conversion of heat into motion. More than a century after Newcomen’s first pump (the famous Dudley Castle engine of 1712), thermodynamics was developed.

In 18th-century Europe, the interplay between pure science and the work of engineers and mechanics became progressively stronger. ...





We must recognise that Europe’s (and the world’s) Great Enrichment was in no way inevitable. With fairly minor changes in initial conditions, or even accidents along the way, it might never have happened. ...

...The world today, after all, still consists of competing entities, and seems not much closer to unification than in 1600. Its market for ideas is more active than ever, and innovations are occurring at an ever faster pace. Far from all the low-hanging technological fruits having been picked, the best is still to come.







スタンリー・フィッシャーが、ウォーリック経済サミット*1での「I'd Rather Have Bob Solow Than an Econometric Model, But ...」と題した11日の講演で、金融政策の決定過程について解説している(H/T Economist’s View)。

Eureka moments are rare in all fields, not least in economics. One such moment came to me when I was an undergraduate at the London School of Economics in the 1960s. I was talking to a friend who was telling me about econometric models. He explained that it would soon be possible to build a mathematical model that would accurately predict the future course of the economy. It was but a step from there to realize that the problems of policymaking would soon be over. All it would take was a bit of algebra to solve for the policies that would produce the desired values of the target variables.

It was a wonderful prospect, and it remains a wonderful idea. But it has not yet happened. I want to talk about why not and about some of the consequences for policymaking.





As an example of such a process, I want to discuss the important decision taken at the August 2011 meeting. At the time policymakers gathered in Washington for the meeting, the FOMC's target for the federal funds rate had been set to nearly zero for more than 2-1/2 years. And although the economy had improved from the depths of the Great Recession, the unemployment rate was still above 9 percent.

Over the summer, the economic outlook darkened considerably. In response, in August, the staff's Tealbook forecast projected that the federal funds rate would remain near zero three quarters longer than what the staff had expected in June. Figure 3, taken from the August 2011 Tealbook, illustrates how the change in the economic outlook affected FRB/US simulations of optimal monetary policy. As you know, an optimal policy is a path for the policy instrument that minimizes the shortfalls in economic outcomes relative to policymakers' goals; in this case, the optimal policy path is computed using the FRB/US model and takes the staff's baseline outlook as given. In principle, optimal policy simulations deliver better outcomes than simple policy rules, but those outcomes are conditional on some strong assumptions.

...the prescriptions of optimal policy were saying not only that the Committee's interest rate should remain at zero for some time to come, but also that that period of time should be considerably longer than previously thought.





結局、このFOMCでは、フォワードガイダンスに関する表現を強める(「for an extended period」→「at least through mid-2013」)ことが決定された。

And what do I take from this episode? The interest rate decision taken in August 2011 was unusual in that a decision was made about the likely path of future interest rates. Most often, the FOMC is deciding what interest rate to set at its current meeting. Either way, in reaching its decision, the Committee will examine the prescriptions of different monetary rules and the implications of different model simulations. But it should never decide what to do until it has carefully discussed the economic logic that underlies its decision. A monetary rule, or a model simulation, or both, will likely be part of the economic case supporting a monetary policy decision, but they are rarely the full justification for the decision. Sometimes a monetary policy committee will make a decision that is not consistent with the prescriptions of standard monetary rules--and that may well be the right decision. Further, in modern times, the policy statement of the monetary policy committee will seek to explain why the committee is making the decision it is announcing. The quality of those explanations is a critical part of the policy process, for good decisions and good explanations of those decisions help build the credibility of the central bank--and a credible central bank is a more effective central bank.


このエピソードの教訓は何だろうか? 2011年8月に採られた金利に関する決定は、将来の金利の想定経路についての決定という点で異例であった。FOMCはその回の会合での金利を決定するのが通例である。いずれにせよ、FOMCは決定に到達するまでに、異なる金融ルールの下での政策、および、異なるモデルシミュレーションの含意を吟味する。しかし、何をするかを決定するのは、決定の基盤となる経済的論理を注意深く議論してから初めて行うべきである。金融ルール、もしくはモデルシミュレーション、もしくはその両方は、金融政策決定を支持する経済的主張の一部とはなるであろうが、決定を完全に正当化するものとなることは稀である。金融政策委員会が標準的な金融ルールから導かれる政策と矛盾する決定を下し、かつ、その決定が大いに正しい、ということもある。また、現代においては、金融政策委員会の政策ステートメントは、公表する決定を委員会が下した理由を説明しようとする。そうした説明の質は、政策プロセスの極めて重要な部分である。というのは、優れた決定と、そうした決定の優れた説明は、中央銀行の信認を構築するのに役立つからである。そして、信認の高い中央銀行は、より効果的な中央銀行なのである。


As the August 2011 meeting illustrates, the eureka moment I thought I had 50-plus years ago was a chimera. Why is that? First, the economy is very complex, and models that attempt to approximate that complexity can sometimes let us down. A particular difficulty is that expectations of the future play a critical role in determining how the economy reacts to a policy change. Moreover, the economy changes over time--this means that policymakers need to be able to adapt their models promptly and accurately in real time. And, finally, no one model or policy rule can capture the varied experiences and views brought to policymaking by a committee. All of these factors and more recommend against accepting the prescriptions of any one model or policy rule at face value.

And now to the bottom line: The title of my speech is an incomplete quotation of something Paul Samuelson once said. What Samuelson said was this, "I'd rather have Bob Solow than an econometric model, but I'd rather have Bob Solow with an econometric model than without one." And Samuelson, who was a shameless eclectic, would almost certainly have said essentially the same thing about policy rules.


2011年8月の会合が示すように、50云年前に私が分かったと思った瞬間は、幻想であった。それはなぜか? 第一に、経済は非常に複雑であり、その複雑さを近似しようとするモデルが期待に応えられないこともしばしばある。特に難しいのは、政策変更に経済がどのように反応するかを決める上で、将来への期待が極めて重要な役割を演じる点である。また、経済は時間とともに変化する。そのことが意味するのは、政策担当者も自分のモデルをリアルタイムで即座かつ正確に適応させることができねばならない、ということである。そして最後に、一つのモデル、もしくは一つの政策ルールだけで、委員会の政策決定に持ち込まれる様々な経験や見解を捉えることはできない。以上の要因すべてが他の要因と相俟って示しているのは、一つのモデル、もしくは一つの政策ルールから導かれる政策を額面通り受け入れるべきではない、ということである。





タイラーコーエンが、自動化の進展について悲観的な見方を表題のブルームバーグ論説(原題は「Industrial Revolution Comparisons Aren't Comforting」)で示している

“Why should it be different this time?” That’s the most common response I hear when I raise concerns about automation and the future of jobs, and it’s a pretty simple rejoinder. The Western world managed the shift out of agricultural jobs into industry, and continued to see economic growth. So will not the jobs being displaced now by automation and artificial intelligence lead to new jobs elsewhere in a broadly similar and beneficial manner? Will not the former truck drivers, displaced by self-driving vehicles, find work caring for the elderly or maybe fixing or programming the new modes of transport?

As economics, that may well be correct, but as history it’s missing some central problems. The shift out of agricultural jobs, while eventually a boon for virtually all of humanity, brought significant problems along the way. This time probably won’t be different, and that’s exactly why we should be concerned.


By the estimates of Gregory Clark, economic historian at the University of California at Davis, English real wages may have fallen about 10 percent from 1770 to 1810, a 40-year period. Clark also estimates that it took 60 to 70 years of transition, after the onset of industrialization, for English workers to see sustained real wage gains at all.


「今回が違うとどうして思うんだい?」というのが、私が自動化と仕事の未来についての懸念を示す時の最も一般的な反応で、それは非常に簡単な反論である。西洋世界は仕事について農業から工業への移行を成し遂げ、経済成長を続けた。だから今回自動化人工知能によって置き換えられる仕事も、概ね同様の有益な形で、何か別の新たな仕事につながるのではないか? 自動運転車で置き換えられた元トラック運転手は、高齢者介護や、あるいは新たな形態の輸送の補修や計画の仕事を見つけるのではないか?