海外における「エビデンスに基づく政策」 - University of Chicago

政策の効果をどう測定するか？
海外における「エビデンスに基づく
政策」の最新動向
伊藤公一朗
シカゴ大学公共政策大学院ハリススクール助教授
Email: [email protected]
Web: www.koichiroito.com
October 2016
本日お話したい内容
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
Conclusion
Evidence-based
policymaking
Example 1: Income
taxation （エビデンスに基づく政策形成）とは
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
具体例１：オバマ大統領が設置したエビデンスに基づく政策のための評議会
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
Introduction
Research
Design
Estimation
Welfare
よくある誤解：「データ＝エビデンス」ではない
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Conclu
エビデンスとは政策効果を因果関係として示したデータ分析結果
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is widely used in many important economic policies
Marginal income tax rates (%) in 2010 in the US
35
30
30
Research Design
l Tax
Rate
(%)
Marginal
Tax
Rate
nal
ax
arginal
Rate
Tax Rate
Tax
(%)
Rate
(%)Marginal
(%)Marginal
Tax
Marginal
Rate
Tax Rate
Tax
(%)
Rate
(%)(%)
(%)
Introduction
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
25
Introduction
政策効果を正しく測定する方法の紹介
Example 1: Income taxation
Conclusion
Conclusion
25
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
25
20
最良の解決法はRCT（ランダム化比較試験）
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is20
widely
used income
in many important
economic
20
15
25
Introduction
Research
Design
Estimation
Welfare
具体例２：経産省、企業、大学研究者が主導したRCTによる政策評価
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
15
10
20
具体例３：RCTが実施できない場合でも既存のデータを賢く使う手法がある
Example
1: Income
taxation
15
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is15
widely
used in many important economic policies
10
5
Introduction
Research
Design income tax rates
Estimation
Conclusion
Marginal
(%) in 2010 in the Welfare
US
30
35
10
30
5
0
10
Introduction
Research
Design
Estimation
Conclusion
Taxable
Income
Nonlinear pricing is35
widely
used in many
important
economicWelfare
policies
25
30
政府機関と研究者のパートナーシップが不可欠
Example 1: Income
taxation
05
5
25
Taxable
Income
Nonlinear pricing is30
widely used in many
important
economic policies
20
25
2 / 86
具体例４：シカゴ大学
Crime
Labの取り組み
Example
1: Income
taxation
0
Marginal
income
taxIncome
rates (%) in 2010 in the US
Taxable
0
20
15
25
20
2 / 86
具体例５：シカゴ大学
EPICの取り組み
Example
1: Income
taxation
Taxable
Marginal
income
tax rates (%)
in 2010 inIncome
the US
10
20
15
2 / 86
15
35
Marginal income tax rates (%) in 2010 in the US
35
Introduction
Conclu
2
具体例１：オバマ大統領が設置した「エビデンスに基づく政策のための評議会」
Introduction
Research Design
Estimation
Welfare
Conclu
Nonlinear pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclusion
オバマ大統領 Paul
Ryan（共和党） Patty
Murray（民主党）
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Evidence-Based
Policymaking
Example 1: Income
taxation Commission Act of 2016
Conclusion
Nonlinear pricing is widely used in many important economic policies
民主党・共和党の共同法案として成立
Example
1: Income
taxation
Introduction
Research Design
Estimation
Welfare
Conclusion
Marginal income tax rates (%) in 2010 in the US
現在、政策担当者の多くは、どれだけの予算を政策に支出できたかという
Example
1: Income taxation
income
tax rates
(%) in 2010
in the US
Nonlinear pricing is Marginal
widely used
in many
important
economic
policies
「Input」を主眼として政策形成をしている。
ate
ate
ax
Rate
(%)(%)
(%)
35
Marginal income tax rates (%) in 2010 in the US
35
今後は、政策がどれだけの効果（雇用創出、環境汚染改善など）を生んだか
Example
1: Income taxation
35
30
30
という「Outcome」を物差しとして政策を形成すべき。
Marginal income tax rates (%) in 2010 in the US
30
25
25
35
25
20
3
具体例１：オバマ大統領が設置した「エビデンスに基づく政策のための評議会」
Introduction
Research Design
Estimation
Welfare
Conclu
Nonlinear Jeﬀery
pricing
is widely used in many importantMeyer（シカゴ大学教授）
economic policies
Liebman（ハーバード大学教授） Bruce
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
Conclusion
ハーバード大やシカゴ大の教授ら１５人の評議会委員が任命された主な使命
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
政府が持つ詳細な行政データを研究者へ利用させ分析させる体制を整える
Example
1: Income taxation
Marginal income tax rates (%) in 2010 in the US
RCT（ランダム化比較試験）などの厳密な科学的手法により政策が評価
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
され、政策効果の因果関係がデータ分析により解明される仕組みを作る
))
35
Marginal income tax rates (%) in 2010 in the US
35
35
30
30
4
よくある誤解：「データ＝エビデンス」ではない
Nonlinear
pricing isResearch
widely
used in many important
economicWelfare
policies
Design
Estimation
Introduction
Conclu
Introduction
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Design
Estimation
なぜオバマは「厳密な科学的手法により評価する」ことに力点を置くのか
Example 1: Income taxation
Conclu
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Design
Estimation
この理由は、「数字やデータ分析が掲載されいる＝エビデンスではない」から
Marginal
income tax rates (%) in 2010 in the US
Example 1: Income
taxation
Conclu
Introduction
Nonlinear pricing is widely used in many important economic policies
政策形成の現場ではこの点が誤解されていることが少なくない
Marginal
income tax rates (%) in 2010 in the US
Example 1: Income
taxation
35
Introduction
Research Design
Estimation
Welfare
Conclu
ginal
Rate
Marginal
Tax
(%)
Marginal
Rate
Tax
Marginal
Rate
(%)
Tax Rate
(%)
Tax Rate
(%) (%)
30
Marginal
income tax rates (%) in 2010 in the US
「エビデンスとは政策（A）がoutcome（B）へ影響を及ぼした因果関係」
Example 1: Income
taxation
35
Nonlinear pricing is widely used in many important economic policies
であり、この因果関係を示すことのできるデータ分析が求められる
25
30
Marginal income tax rates (%) in 2010 in the US
35
この点を解説するために架空のケーススタディーを考えてみる
Example 1: Income
taxation
20
25
30
35
15
Marginal income tax rates (%) in 2010 in the US
20
25
30
10
15
20
35
25
5
10
15
30
20
0
5
10
25
15
Taxable Income
0
5
20
5
ケーススタディ：あなたが霞ヶ関の政策担当官だったら
l  あなたの上司
l 
日本では電力の供給が切迫したままだ。価格政策は有効なのか？
l 
電力の料金を上げると、消費者がどのくらいの量の節電するのかを知りたい
l  あなた
l 
「価格」と「平均消費量」のデータを集めてみることに
2010年
(Before)
電力料金の
上昇があった
2012年
(After)
電力消費者
6
ケーススタディ：あなたが霞ヶ関の政策担当官だったら
Before （２０１０夏）
A(er （２０１２夏） Diﬀerence
２０円
２５円
＋５円
２０
１５
− ５
電力価格（円/kWh）
平均消費量
Introduction
（kWh/day）
Research Design
Estimation
Welfare
Conclu
Introduction
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Design
Estimation
Conclu
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Design
Estimation
あなた：価格変化があったBeforeとAfterのデータからこう結論づけた
Example 1: Income taxation
Conclu
Introduction
Nonlinear pricing is widely used in many important economic policies
「価格が５円上がったため、消費量が５単位減った。価格政策は有効である」
Marginal
income tax rates (%) in 2010 in the US
Example 1: Income
taxation
(%)
Marginal
income tax rates (%) in 2010 in the US
質問：この分析・結論のどのような点が誤っている可能性があるか？
Example 1: Income
taxation
35
30
Marginal income tax rates (%) in 2010 in the US
35
7
社会科学データで「causality（因果関係）」を言うのはとても難しい
A
B
l  AとBのデータが、一緒に動いている様子が観察できたとする（相関関係）
l  多くの場合、私たちは相関関係を見ると、以下のことを言ってしまいがち
AがBを引き起こした（因果関係）
8
社会科学データで「causality（因果関係）」を言うのはとても難しい
A
B
l  ところが、実験室で作られたデータでない場合、別の可能性もあり得る
可能性（２）BがAを引き起こした
9
社会科学データで「causality（因果関係）」を言うのはとても難しい
A
B
C
可能性（３）CがAとBの両方に影響を与えただけで、
AとBに直接の関係性はない
10
社会科学データで「causality（因果関係）」を言うのはとても難しい
A
B
C
これらの矢印の可能性を排除できないと
AからBへのCausality（因果関係）は言えない
11
社会科学データで「causality（因果関係）」を言うのはとても難しい
l  私たちのケーススタディでは、Aが電力料金、Bが消費量でした。
l  Cは何が考えられますか？どうしたらCの影響は排除できますか？
12
社会科学データで「causality（因果関係）」を言うのはとても難しい
l  私たちのケーススタディでは、Aが電力料金、Bが消費量でした。
l  Cは何が考えられますか？どうしたらCの影響は排除できますか？
l  社会科学データの厄介な点（そして誤解されがちな点）
l 
ビッグデータのような大量のデータがあれば解決するのでは？
l 
統計学のスペシャリストが高度なモデルを解析すれば解決するのでは？
l 
どちらも答えは「No」
l 
データの観測数が増えても、この「Bias」の問題は解決しない
l 
高度なモデルを使っても解決は難しいことが80年代の経済学研究で明らかに
13
社会科学データで「causality（因果関係）」を言うのはとても難しい
l  私たちのケーススタディでは、Aが電力料金、Bが消費量でした。
l  Cは何が考えられますか？どうしたらCの影響は排除できますか？
l  社会科学データの厄介な点（そして誤解されがちな点）
l 
ビッグデータのような大量のデータがあれば解決するのでは？
l 
統計学のスペシャリストが高度なモデルを解析すれば解決するのでは？
l 
どちらも答えは「No」
l 
データの観測数が増えても、この「Bias」の問題は解決しない
l 
高度なモデルを使っても解決は難しいことが80年代の経済学研究で明らかに
l  残念ながら、多くの「政策分析」が本ケーススタディのような「単なる
Before-After分析」という誤った手法で行われている
14
本日お話したい内容
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
Conclusion
Evidence-based
policymaking
Example 1: Income
taxation （エビデンスに基づく政策形成）とは
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
具体例１：オバマ大統領が設置したエビデンスに基づく政策のための評議会
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
Introduction
Research
Design
Estimation
Welfare
よくある誤解：「データ＝エビデンス」ではない
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Conclu
エビデンスとは政策効果を因果関係として示したデータ分析結果
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is widely used in many important economic policies
Marginal income tax rates (%) in 2010 in the US
35
30
30
Research Design
l Tax
Rate
(%)
Marginal
Tax
Rate
nal
ax
arginal
Rate
Tax Rate
Tax
(%)
Rate
(%)Marginal
(%)Marginal
Tax
Marginal
Rate
Tax Rate
Tax
(%)
Rate
(%)(%)
(%)
Introduction
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
25
Introduction
政策効果を正しく測定する方法の紹介
Example 1: Income taxation
Conclusion
Conclusion
25
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
25
20
最良の解決法はRCT（ランダム化比較試験）
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is20
widely
used income
in many important
economic
20
15
25
Introduction
Research
Design
Estimation
Welfare
具体例２：経産省、企業、大学研究者が主導したRCTによる政策評価
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
15
10
20
具体例３：RCTが実施できない場合でも既存のデータを賢く使う手法がある
Example
1: Income
taxation
15
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is15
widely
used in many important economic policies
10
5
Introduction
Research
Design income tax rates
Estimation
Conclusion
Marginal
(%) in 2010 in the Welfare
US
30
35
10
30
5
0
10
Introduction
Research
Design
Estimation
Conclusion
Taxable
Income
Nonlinear pricing is35
widely
used in many
important
economicWelfare
policies
25
30
政府機関と研究者のパートナーシップが不可欠
Example 1: Income
taxation
05
5
25
Taxable
Income
Nonlinear pricing is30
widely used in many
important
economic policies
20
25
2 / 86
具体例４：シカゴ大学
Crime
Labの取り組み
Example
1: Income
taxation
0
Marginal
income
taxIncome
rates (%) in 2010 in the US
Taxable
0
20
15
25
20
2 / 86
具体例５：シカゴ大学
EPICの取り組み
Example
1: Income
taxation
Taxable
Marginal
income
tax rates (%)
in 2010 inIncome
the US
10
20
15
2 / 86
15
35
Marginal income tax rates (%) in 2010 in the US
35
Introduction
Conclu
15
最良の解決法：RCT（ランダム化比較試験）
Nonlinear
pricing is widely used in many important economic policies
医学研究では、新薬が開発された際、効果の検証はどうやって行いますか？
Example 1: Income taxation
Marginal income tax rates (%) in 2010 in the US
Marginal Tax Rate (%)
35
30
25
20
15
10
5
0
Taxable Income
16
最良の解決法：RCT（ランダム化比較試験）
l  医学研究では、新薬が開発された際、効果の検証はどうやって行いますか？
トリートメント
グループ
医薬品の投与を行うグループ
グルーピングは必ず無作為
（random）に行う
コントロール
グループ
医薬品の投与を行わないグループ
•  ２つのグループの比較によって医薬品の効果が測定できる
17
最良の解決法であるRCTを政策評価に取り入れるのが世界的潮流
例：「電力料金の価格インセンティブが節電行動に及ぼす影響を調べたい」
電力料金の変化
トリートメント
グループ
一般世帯
グルーピングは必ず無作為
（random）に行う
何も行わない
コントロール
グループ
•  ２つのグループの比較によって電力料金の効果が測定できる
18
なぜRCTは因果関係を導けるのか
電力料金の変化
トリートメント
グループ
一般世帯
グルーピングは必ず無作為
（random）に行う
何も行わない
Introduction
Research Design
Estimation
Welfare
コントロール
グループ
Nonlinear pricing is widely
used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclu
Conclusion
ランダム・グルーピングによって、「電力料金以外の他の要因によって２つ
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
のグループの消費量が違ったのでは？という可能性は科学的に排除される
Marginal income tax rates (%) in 2010 in the US
所得、家族構成、家の広さ、気温、他の政策の影響など全ての他の要因は２
Example
1: Income taxation
つのグループで統計的に等しくなる
35
Marginal income tax rates (%) in 2010 in the US
35
30
19
具体例２：政府、企業、大学研究者が主導した日本でのRCTによる政策評価
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
次世代エネルギー・社会システム実証実験
Example 1: Income taxation
Conclusion
Conclusion
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
資源エネルギー庁省エネルギー・新エネルギー部新産業・社会システム推進室
Example
1: Income taxation
Introduction
Design
Estimation
Conclusion
Marginal
tax rates
(%) inWelfare
2010 in the US
Nonlinear pricing isResearch
widely
used income
in many important
economic
policies
地方自治体
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Nonlinear pricing is widely used in many important economic policies
Introduction
Research
Design
Estimation
Welfare
企業 1: Income
Example
taxation
35
Marginal
income tax rates (%) in 2010
in the US
35
Introduction
Conclu
Introduction
x Rate
Rate
Marginal
Tax
Rate
nal
ate
ax
Tax
(%)
Rate
(%)(%)
Marginal
(%)Marginal
Tax
Marginal
Rate
Tax
Marginal
Rate
Tax
(%) Rate
Tax
(%)
Rate
(%)(%)
(%)
大学研究者：依田高典（京大）、田中誠（GRIPS)、伊藤公一朗（シカゴ大）
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
35
30
30
Nonlinear pricing is widely used in many important economic policies
Marginal income tax rates (%) in 2010 in the US
30
25
35
Research Design
Estimation
Welfare
25
日本の４地域でスマート・グリッドに関する実証実験
Example 1: Income taxation
Nonlinear pricing is20
widely used in many important economic policies
20
15
30
25
Conclusion
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
25
20
30
Conclusion
Introduction
Conclusion
Introduction
Research Design
Estimation
Welfare
大学研究者がRCTに基づいた実験設計のお手伝いと統計分析を行った
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is25
widely
used income
in many important
economic
15
10
20
１）横浜、２）けいはんな、３）豊田、４）北九州
Example
1: Income
taxation
15
Marginal
income tax rates (%) in 2010 in the US
10
5
20
15
本日は、けいはんな地域で行われた実験結果を簡単に紹介
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
10
5
0
15
10
Taxable
Income
Marginal
income tax
rates (%)
in 2010 in the US
35
30
30
05
10
5
Taxable Income
30
25
35
2 / 86
5
0
25
Taxable Income
25
200
30
2 / 86
Taxable Income
0
Taxable Income
20
20
15
25
2 / 86
20
具体例２：政府、企業、大学研究者が主導した日本でのRCTによる政策評価
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
原発事故以降、電力が足りない日本
Example 1: Income taxation
Conclusion
Conclusion
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
経済学的には、「強制的に企業や世帯の電力を止める」政策はナンセンス
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
２つの政策があり得る。一つは、①価格インセンティブを明確に与えること
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Introduction
Research Design
Estimation
Welfare
もう一つは、②良心に訴えて、自発的な省エネをお願いすること
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Conclu
Introduction
Rate
Marginal
Tax
Rate
rginal
lTax
Tax Rate
Tax
Rate
(%)(%)
(%)
Marginal
Marginal
Tax
Marginal
Rate
Tax Rate
Tax
(%)
Rate
(%)(%)
(%)
Nonlinear pricing is35
widelyincome
used
many
important
Marginal
taxin
rates
(%) in 2010
in the US economic policies
30
30
Research Design
Estimation
Welfare
30
25
35
Research
Design
Estimation
Nonlinear pricing is25
widely
used in many important
economicWelfare
policies
京都けいはんな地域で行った実験
Example 1: Income
taxation
25
20
30
Nonlinear pricing is widely used in many important economic policies
Introduction
Researchtaxation
Design
Estimation
京都けいはんな地域で行った実験の紹介
Example
1: Income
20
20
15
25
Introduction
Conclusion
Conclusion
Welfare
Marginal income tax rates (%) in 2010 in the US
Conclu
700世帯にスマートメーターを入れ各世帯30分ごとの電力消費データを入手
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
15
10
Nonlinear pricing is20
widely used in many important economic policies
15
35
Marginal
income tax rates (%) in 2010 in the US
10
5
15
35
10
Nonlinear pricing is10
widely used in many important economic policies
5
0
30
35
30
700世帯を３つのグループへランダムに分ける
Example 1: Income taxation
Taxable Income
Nonlinear pricing is25
widely
used in many important economic policies
0
5
5
30
１）価格インセンティブ・グループ
Example 1: Income taxation Taxable Income
25
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is25
widely
used income
in many important
economic
0
20
２）自発的な省エネをお願いする節電要請グループ
Example
1: Income
Taxable
Income
0 taxation
Marginal
income tax
rates (%)
in 2010 in the US
Taxable Income
20
15
20
３）何も行わないコントロール・グループ
Example
1: Income
taxation
35
Marginal income tax rates (%) in 2010 in the US
35
Introduction
Introduction
Introduction
Research Design
Estimation
Welfare
Conclusion
Research Design
Estimation
Welfare
Conclusion
Research Design
Estimation
Welfare
Conclusion
2 / 86
2 / 86
2 / 86
10
15
21
RCTのもう一つのメリットは分析結果が非専門家にも理解しやすいこと
Figure 3: E↵ects of Intrinsic and Extrinsic Incentives on Electricity Consumption
Mean Consumption (kWh per 30−minute)
.4
Control group
Intrinsic group
Extrinsic group
.35
Treatment
Hours
.3
.25
.2
8
9
10
11
12
13
14
15 16
Hour
17
18
19
20
21
22
22
政策設計への重要な発見：価格効果は持続するが、モラルへ訴える効果は持続せず
Panel B: Winter Experiment
Moral%Suasion%
Economic%Incen6ve%
Reduc6ons%in%usage%in%log%points
0.25%
0.2%
0.15%
0.1%
0.05%
0%
1st%
2nd%
3rd%
4th%
5th%
6th%
7th%
Treatment%Cycle
!0.05%
s figure shows the treatment e↵ects of moral suasion and economic incentives by treatment cyc
23
ctions in electricity usage in log points. The estimates are obtained from the estimation resu
具体例３：データ提供の重要性：RCTができない場合でも科学的分析は可能
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclu
RCTを行うえない場合でも、既存の行政データを賢く使う分析手法もある
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclusion
Marginal income tax rates (%) in 2010 in the US
Estimation
Nonlinear pricing is widely used in many important
economicWelfare
policies
Introduction
Research Design
Conclusion
具体例として私がカリフォルニア州政府・電力会社と行った研究の紹介
Example 1: Income taxation
Nonlinear pricing is35
widely used in many important economic policies
協力：California
Energy Commission（政府）, SCE, SDG&E（電力会社）
Example
1: Income taxation
Marginal income tax rates (%) in 2010 in the US
Marginal
Marginal
Tax
Rate
Tax
(%)
Marginal
Marginal
Tax Rate
Tax
Rate
(%)
(%)Rate (%)
30
提供されたデータ：世帯ごとの月間電力消費データ１０年分（加州全土）
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
35
Marginal income tax rates (%) in 2010 in the US
35
25
35
30
30
20
30
25
25
15
25
20
20
20
15
10
15
10
15
5
10
5
10
5
00
05
TaxableTaxable
Income
Taxable Income
Income
24
RCTができない場合の手法の一つ：あたかも実験が起こった状況を利用する方法
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
VOL. 104 NO. 2
Estimation
Welfare
Conclusion
ITO: DO CONSUMERS RESPOND TO MARGINAL ORConclusion
AVERAGE PRICE?
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
２つの電力会社の境界線に着目
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
都市の北と南で電力会社が異なる
Example
1: Income taxation
Marginal income tax rates (%) in 2010 in the US
北と南の住人は異なった電力料金を経験
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
35
Marginal income tax rates (%) in 2010 in the US
35
35
30
Marginal
Tax
Rate
Marginal
Marginal
Tax Rate
Tax
Rate
(%)(%)
(%)
30Woods
Laguna
Rancho
Santa
Margarita
Mission
Viejo
30
25
25
25
20
20
20
15
Aliso
Viejo
15
10
15
Las
Flores
10
5
Laguna
10
5
0
Beach
05
Coto
de
Caza
Laguna
Hills
0
2
3
4
5 mile
Taxable Income
TaxableLaguna
Income
Niguel
0
1
Border of electric utility service areas
2 / 86
City limits
Taxable Income
2 / 86
25
e A.2: Changes in Consumption from 1999 to 2000 by Distance from th
まずは「電力料金がの変化が両地域で同様であった」時期のデータを見る
y Border
Panel A. Changes in Consumption from July 1999 to July 2000
Changes in Consumption (%)
5
境界線から
この距離に
住む世帯の
平均電力消費量
0
−5
−10
−15
−1.5
−1
−.5
0
.5
Distance from the utility border (mile)
1
1.5
Panel B. Changes
in Consumption from境界線から南への距離
August 1999 to August 2000
境界線から北への距離
5
26
−15
南の地域で電力料金が上がりました。消費量はどうなったでしょう？
−1.5
−1
−.5
0
.5
1
1.5
Distance from the utility border (mile)
Panel B. Changes in Consumption from August 1999 to August 2000
Changes in Consumption (%)
5
0
−5
−10
−15
−1.5
−1
−.5
0
.5
Distance from the utility border (mile)
1
1.5
s: The figure provides evidence of the validity of the spatial regression discontinuit
境界線から北への距離
境界線から南への距離
n and suggests that using samples closer to the utility border does not a↵ect my 27
est
−15
南の地域で電力料金が上がりました。消費量はどうなったでしょう？
−1.5
−1
−.5
0
.5
1
1.5
Distance from the utility border (mile)
Panel B. Changes in Consumption from August 1999 to August 2000
Changes in Consumption (%)
5
0
−5
−10
−15
−1.5
−1
−.5
0
.5
Distance from the utility border (mile)
1
1.5
s: The figure provides evidence of the validity of the spatial regression discontinuit
境界線から北への距離
境界線から南への距離
n and suggests that using samples closer to the utility border does not a↵ect my 28
est
具体例３：データ提供の重要性：RCTができない場合でも科学的分析は可能
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclusion
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
RCTを行うえない場合でも、既存の行政データを賢く使う分析手法もある
Example 1: Income taxation
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
政府統計データ
Example
1: Income taxation
Introduction
Design
Estimation
Marginal
tax rates
(%) inWelfare
2010 in
Nonlinear pricing isResearch
widely
used income
in many important
economic
policies
国税調査などの行政データ
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Research
Design
Estimation
Nonlinear pricing is widely used in many important
economicWelfare
policies
補助金データ
Example
1: Income
taxation
35
Marginal income tax rates (%) in 2010 in the US
35
Nonlinear pricing is widely used in many important economic policies
Introduction
Research
Design
Estimation
徴税データ
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010
in the US
35
30
30
ate
(%)
Marginal
Tax
Rate
%)
ate
ax
Rate
(%)Marginal
(%)Marginal
Tax
Marginal
Rate
Tax
Marginal
Rate
Tax
(%)
Marginal
Rate
Tax
(%) Rate
Tax
(%)
Rate
(%)(%)
(%)
Introduction
Conclusion
Conclusion
the US
Conclusion
Welfare
企業が持つデータで政策分析に有用なデータ
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
25
35
Nonlinear pricing is30
widely used in many important economic
Introduction
25
Marginal income tax rates (%) in 2010 in the US
Research Design
25
20
35
30
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is20
widely
used in many important
economicWelfare
policies
20
15
30
25
35
Introduction
Conclu
policies
Conclusion
Conclusion
いずれも、他の先進国に比べて日本はデータ提供の面では立ち遅れている
Example 1: Income taxation
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
15
10
25
20
結果的に、世界中の研究者が「日本の政策は分析が不可能」という認識を持つ
Example
1: Income
15 taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is25
widely
used income
in many important
economic
10
5
20
15
日本で「エビデンスに基づく政策形成」を行うためには克服する必要がある点
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
10
5
0
15
10
20
各国ではデータセキュリティを守りつつ分析者へ提供する仕組みが整っている
Example
1: Income
taxation
35
Taxable
Income
Marginal
income tax
rates (%)
in 2010 in the US
35
05
10
5
15
Taxable
Income
Marginal
income tax
rates (%)
in 2010 in the US
30
35
2 / 86
30
5
0
10
Taxable Income
250
35
30
2 / 86
Taxable
Income
0
5
25
Taxable Income
20
30
25
2 / 86
29
0
Introduction
本日お話したい内容
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
Conclusion
Evidence-based
policymaking
Example 1: Income
taxation （エビデンスに基づく政策形成）とは
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
具体例１：オバマ大統領が設置したエビデンスに基づく政策のための評議会
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
Introduction
Research
Design
Estimation
Welfare
よくある誤解：「データ＝エビデンス」ではない
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Conclu
エビデンスとは政策効果を因果関係として示したデータ分析結果
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is widely used in many important economic policies
Marginal income tax rates (%) in 2010 in the US
35
30
30
Research Design
l Tax
Rate
(%)
Marginal
Tax
Rate
nal
ax
arginal
Rate
Tax Rate
Tax
(%)
Rate
(%)Marginal
(%)Marginal
Tax
Marginal
Rate
Tax Rate
Tax
(%)
Rate
(%)(%)
(%)
Introduction
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
25
Introduction
政策効果を正しく測定する方法の紹介
Example 1: Income taxation
Conclusion
Conclusion
25
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
25
20
最良の解決法はRCT（ランダム化比較試験）
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is20
widely
used income
in many important
economic
20
15
25
Introduction
Research
Design
Estimation
Welfare
具体例２：経産省、企業、大学研究者が主導したRCTによる政策評価
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
15
10
20
具体例３：RCTが実施できない場合でも既存のデータを賢く使う手法がある
Example
1: Income
taxation
15
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is15
widely
used in many important economic policies
10
5
Introduction
Research
Design income tax rates
Estimation
Conclusion
Marginal
(%) in 2010 in the Welfare
US
30
35
10
30
5
0
10
Introduction
Research
Design
Estimation
Conclusion
Taxable
Income
Nonlinear pricing is35
widely
used in many
important
economicWelfare
policies
25
30
政府機関と研究者のパートナーシップが不可欠
Example 1: Income
taxation
05
5
25
Taxable
Income
Nonlinear pricing is30
widely used in many
important
economic policies
20
25
2 / 86
具体例４：シカゴ大学
Crime
Labの取り組み
Example
1: Income
taxation
0
Marginal
income
taxIncome
rates (%) in 2010 in the US
Taxable
0
20
15
25
20
2 / 86
具体例５：シカゴ大学
EPICの取り組み
Example
1: Income
taxation
Taxable
Marginal
income
tax rates (%)
in 2010 inIncome
the US
10
20
15
2 / 86
15
35
Marginal income tax rates (%) in 2010 in the US
35
Introduction
Conclu
30
政策担当者と研究者（データ分析の専門家）のパートナシップが不可欠
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
政策担当者
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
Introduction
Researchtaxation
Design
Estimation
政策を実際に運用し、課題が何かを深く理解している
Example
1: Income
Conclusion
Conclusion
Welfare
Marginal income tax rates (%) in 2010 in the US
Conclu
一方で、データ分析への専門知識は必ずしも持ち合わせない
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Nonlinear pricing is widely used in many important economic policies
35
Marginal income tax rates (%) in 2010 in the US
35
Introduction
Research Design
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
Introduction
ginal
%)
Tax
Rate
Marginal
(%)
Tax
Rate
arginal
(%)Marginal
Tax Rate
Tax
Rate
(%)Marginal
(%)Marginal
Tax Rate
Tax
Rate
(%)(%)
(%)
30
研究者（データ分析の専門家）
Example 1: Income
taxation
Conclusion
Conclusion
Nonlinear pricing is30
widely used in many important economic policies
25
Research
Design
Estimation
Welfare
データ分析の専門性を持ち合わせている
Example 1: Income taxation
25
Marginal income tax rates (%) in 2010 in the US
25
20
一方で、実際の政策課題については明るくない
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Nonlinear pricing is20
widely
used in many important economic policies
20
15
Introduction
Research
Design income tax rates
Estimation
Conclusion
35
Marginal
(%) in 2010 in the Welfare
US
35
15
10
Introduction
Research
Design
Estimation
Conclusion
15
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
30
アメリカではこの点が両者によって理解されている
Example 1: Income
taxation
10
5
Introduction
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
25
10
エビデンスに基づく政策形成へ向けては両者の協力が不可欠という認識
Example
1: Income
taxation
5
0
25
Marginal
taxIncome
rates
(%) in policies
2010 in the US
Taxable
Nonlinear pricing is25
widely used income
in many
important
economic
20
オバマ大統領が設置した評議委員会が一つの例
Example
1: Income
Marginal
income tax rates (%) in 2010 in the US
05 taxation
Taxable Income
20
20
15
2 / 86
以下では、さらに２つの例を紹介
Example
1: Income
taxation
35
Marginal income tax rates (%) in 2010 in the US
35
15
100
2 / 86
15
Taxable
Income
Marginal
income
tax
rates
(%)
in
2010
in
the
US
35
30
30
10
5
30
25
35
Introduction
Conclu
31
具体例３：シカゴ大学 Crime Labの取り組み
Introduction
Research Design
Estimation
Welfare
Conclu
Nonlinear pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Conclusion
Ludwig
（シカゴ大学教授・Crime
Lab所長）
Design
Estimation
Welfare
Conclusion
Nonlinear pricing isResearch
widely
used in many 右：Jens
important
economic
policies
Introduction
シカゴ大学Crime
Labtaxation
Example 1: Income
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Researchtaxation
Design
Estimation
Welfare
政策担当者：シカゴ市、シカゴ警察、シカゴ市内の学校
Example
1: Income
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
研究者：シカゴ大学の研究者を中心とする研究チーム
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Conclu
Nonlinear pricing is widely used in many important economic policies
協力者：犯罪防止を目指すNPO
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Marginal income tax rates (%) in 2010 in the US
ate
te
x) Rate
(%)(%)
(%)
35
30
30
この３者が協力し、RCTなどの科学的分析手法を用いて政策評価を行っている
Example 1: Income
taxation
30
25
35
25
Marginal income tax rates (%) in 2010 in the US
25
20
30
32
具体例３：シカゴ大学 Crime Labの取り組み
Introduction
Research Design
Estimation
Welfare
Conc
Nonlinear pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Becoming
Man Program
Example 1:a Income
taxation
Conclusion
Conclusion
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
政策目標：青少年の犯罪をいかに予防するか
Example
1: Income taxation
Introduction
Design
Estimation
Conclusion
Marginal
tax rates
(%) inWelfare
2010 in the US
Nonlinear pricing isResearch
widely
used income
in many important
economic
policies
政策介入：中高に通う学生、及び少年院にいる青少年への犯罪防止教育
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Nonlinear pricing is widely used in many important economic policies
効果測定法：RCT（ランダムに選ばれた子供だけがこの教育を受ける）
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Introduction
(%)
ate
) (%)
政策効果：犯罪を犯す確率が大幅に下がり、高校卒業率が飛躍的に向上
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
35
30
30
Marginal income tax rates (%) in 2010 in the US
30
25
35
33
具体例４：シカゴ大学 EPIC(Energy Policy Institute at Chicago)の取り組み
Introduction
Research Design
Estimation
Welfare
Conclu
Nonlinear
pricing is widely used in many important economic policies
Michael Greenstone（シカゴ大学教授・EPIC所長）
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Conclusion
Conclusion
シカゴ大学
Policy Institute at University of Chicago)
Example 1:EPIC(Energy
Income taxation
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
Researchtaxation
Design
Estimation
Welfare
政策担当者：各国の政府や地方公共団体、世界銀行などの国際機関
Example
1: Income
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
研究者：シカゴ大学の研究者を中心とする研究チーム
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
Conclu
Nonlinear pricing is widely used in many important economic policies
協力者：環境・エネルギー問題に取り組むNPO
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Marginal income tax rates (%) in 2010 in the US
(%)
ate
) (%)
35
30
30
この３者が協力し、RCTなどの科学的分析手法を用いて政策評価を行っている
Example 1: Income
taxation
30
25
35
34
具体例４：シカゴ大学 EPIC(Energy Policy Institute at Chicago)の取り組み
Introduction
Research Design
Estimation
Welfare
Conc
Nonlinear pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
インドにおける環境汚染抑止政策
Example 1: Income taxation
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
政策担当者：インド・グジャラート州政府
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in
Nonlinear pricing is widely
used income
in many important
economic
政策目標：工場からの環境汚染を減らすこと
Example
1: Income
taxation
Marginal
(%) in 2010 in the Welfare
US
Introduction
Research
Design income tax rates
Estimation
Introduction
Conclusion
Conclusion
Conclusion
the US
Conclusion
政策介入：機能していない汚染モニタリング・システムを見直し、経済学の
Example
1: Income
taxation
35
Marginal
income
tax rates
(%) in 2010
in the Welfare
US
Research
Design
Estimation
Conclusion
Nonlinear pricing is35
widely
used
in many
important
economic
policies
Introduction
理論に基づいた新たなモニタリング・システムを導入
ate
te
x) Rate
(%)(%)
(%)
Marginal
income
tax rates
(%) in 2010
in the US
30
Nonlinear pricing is35
widely used
in many
important
economic
policies
30
効果測定法：RCT（ランダムに選ばれた工場だけがこの政策を受ける）
Example
1: Income taxation
30
25
35
政策効果：工場から排出される汚染物質が大幅に低下
Example
1: Income
taxation
25
Marginal
income tax rates (%) in 2010 in the US
25
20
30
35
本日お話した内容、及びディスカッション
Nonlinear
pricing is widely used in many important economic policies
Introduction
Research Design
Estimation
Welfare
Design
Estimation
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Introduction
本日お話した３点
Example 1: Income taxation
Conclusion
Conclusion
Design
Estimation
Conclusion
Nonlinear pricing isResearch
widely
used in many important
economicWelfare
policies
Evidence-based
policymaking （エビデンスに基づく政策形成）とは
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is widely
used income
in many important
economic
Introduction
Research
Design
Estimation
Welfare
政策効果を正しく測定する方法の紹介
Example 1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
Introduction
政府機関と研究者のパートナーシップが不可欠
Example
1: Income
taxation
35
Marginal
income tax rates (%) in 2010 in the US
35
Nonlinear pricing is widely used in many important economic
Marginal income tax rates (%) in 2010 in the US
35
30
30
Research Design
l Tax
Rate
(%)
Marginal
Tax
Rate
nal
ax
arginal
Rate
Tax Rate
Tax
(%)
Rate
(%)Marginal
(%)Marginal
Tax
Marginal
Rate
Tax Rate
Tax
(%)
Rate
(%)(%)
(%)
Introduction
Estimation
Welfare
Research
Design
Estimation
Nonlinear pricing is35
widely
used in many important
economicWelfare
policies
30
25
Introduction
ディスカッション
Example 1: Income taxation
Conclu
policies
Conclusion
Conclusion
25
Research
Design
Estimation
Conclusion
Nonlinear pricing is30
widely
used in many important
economicWelfare
policies
25
20
日本ではどのような場でエビデンスに基づく政策形成が有用となり得るか
Example
1: Income taxation
Marginal
tax rates
(%) in policies
2010 in the US
Nonlinear pricing is20
widely
used income
in many important
economic
20
15
25
霞が関でエビデンスに基づく政策形成を実現させるには何が必要か
Example
1: Income
taxation
Marginal
income tax rates (%) in 2010 in the US
15
10
20
研究者とのパートナーシップで求める点は何か、など
Example
1: Income
taxation
15
35
Marginal
income tax rates (%) in 2010 in the US
35
10
5
15
Marginal
income tax rates (%) in 2010 in the US
30
35
10
30
5
0
10
Taxable Income
25
35
30
05
5
25
Taxable Income
20
30
25
2 / 86
0
Taxable Income
0
20
15
25
20
2 / 86
Introduction
Taxable Income
10
20
15
15
2 / 86
36
参考資料pricing is widely used in many important economic policies
Nonlinear
Introduction
Research Design
Introduction
Research Design
Estimation
Welfare
Estimation
Conclusion
Welfare
Conclu
Nonlinear pricing is widely used in many important economic policies
具体例１：「エビデンスに基づく政策のための評議会」ウェブサイト
Example 1: Income taxation
Nonlinear pricing is widely used in many important economic policies
Research Design
Estimation
Welfare
Conclusion
https://www.whitehouse.gov/omb/management/commission_evidence
Example
1: Income
taxation
Introduction
Marginal income tax rates (%) in 2010 in the US
Estimation
Nonlinear pricing is Marginal
widely used
in many
important
economic
policies
income
tax rates
(%) in 2010
in the Welfare
US
Introduction
Research Design
具体例２と３
Example 1: Income taxation
Conclusion
Nonlinear pricing is35
widely used in many important economic policies
35
http://home.uchicago.edu/ito/research.html
Example
1: Income taxation
Marginal income tax rates (%) in 2010 in the US
Marginal
Tax
Rate
Tax
(%)
Rate
(%)
rginalMarginal
TaxMarginal
Rate
Marginal
(%)
Tax Rate
Tax
Marginal
Rate
(%)
(%)
Tax
Rate
(%)
30
30
Research
Design
Estimation
Conclusion
Ito, Ida,1:and
Tanaka
(2015):
Moral
Suasion
Incentives:
Example
Income
taxation
Marginal
income
tax rates
(%) in 2010
in the Welfare
USand Economic
Introduction
25 Evidence from Energy Demand, NBER Working Paper
Experimental
Nonlinear pricing is35
widely used
in many
important
economic
policies
35
Marginal
income
tax rates
(%) in 2010
in the US
25
20910. 20
Introduction
Research Design
Estimation
Welfare
35
30
30
Ito (2014):
Do Consumers
Respond to Marginal or Average Price?
Example
1: Income
taxation
20
15
from
Nonlinear
Electricity
Pricing, American
Economic
25
NonlinearEvidence
pricing is30
widely
used
in
many
important
economic
policies
Marginal
income
tax
rates
(%)
in
2010
in
the
US
10
25
Introduction Review, 104(2):
Research Design
Estimation
Welfare
Conclusion
537-63, 2014.
15
25
20
Introduction
Research Design
Estimation
Welfare
5
35
Nonlinear pricing is20
widely used in many important economic policies
20
15
10
具体例４：University
of Chicago Crime Lab
Example 1: Income
taxation
0
30
Nonlinear pricing is15
widely usedTaxable
in many
important economic policies
Income
10
IntroductionExample
Research Design
Estimation
Welfare
Conclusion
https://crimelab.uchicago.edu/ 1: Income
taxation
15
25
5
Marginal income tax rates (%) in 2010 in the US
10
5
2 / 86
Nonlinear pricing is20
widely used
in many
important
economic
policies
Marginal
income
tax rates
(%) in 2010
in the US
10
具体例５：University
of Chicago EPIC
Example 1: Income
taxation
5
00
35
TaxableTaxable
Income Income
15
35
https://epic.uchicago.edu/
Example
1: Income
taxation
05
Marginal income
taxIncome
rates (%) in 2010 in the US
Taxable
10
30
Conclu
Conclu
37
以上です。質問・コメントはこちらへお寄せください。
シカゴ大学公共政策大学院ハリススクール助教授伊藤公一朗
Email: [email protected]
Web: www.koichiroito.com
38

Download Report