Workshops

The Applied Statistics Workshop 2016

※統計数理研究所リスク解析戦略研究センター金融・保険リスク研究プログラムとの共催ワークショップ

日時

2016年4月8日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

黒住英司 (一橋大学)

Monitoring Parameter Constancy with Endogenous Regressors

要旨(Abstract) This paper proposes monitoring tests for parameter change in linear regression models with endogenous regressors. We consider a CUSUM-type test based on the instrumental variable (IV) estimation, as the IV method is standard for models with endogenous regressors. In addition, we propose a test based on the residuals from the least squares (LS) estimation. We show that for a given boundary function, both tests have the same limiting distribution under the null hypothesis, whereas their powers are different. In particular, when a structural change occurs early in a monitoring period, the test based on the LS method tends to detect it more rapidly than that based on the IV method. We apply our methods to investigate the Japanese Phillips curve and show that the LS based test performs well to detect a change in 2007, while neither test finds evidence of a change after 2013.

 

日時

2016年4月15日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

松田安昌 (東北大学)

空間CARMAモデル

要旨(Abstract) 離散ARMAモデルは定常時系列を表現する標準的なモデルとしてよく知られている。 一方、連続時間モデルの研究も古くからあり、Doob(1944),Bartlett(1946),Durbin(1961)らによるContinuous time ARMA (CARMA)モデルの研究が知られている。近年、高頻度データの分析が一般化するに伴って連続時間モデルのへの関心が高まり、 CARMAモデルが再び注目されている。本発表では、CARMAモデルを時系列モデルか ら空間モデルへと拡張する試みを紹介し、n次元空間上で定義される空間CARMAモ デルを提案する。空間CARMAモデルは、CARMAカーネルと呼ばれるパラメトリック な関数と多次元空間上のLevy過程の畳込みによって定義され、その相関構造は1 または3次元では解析的に表現される。不規則に位置する標本点における観測値 をもとにパラメータの推定法および予測(kriging)法を提案し、2015年度東京都 23区1247標本地点における公示地価の対前年度収益率へ応用する。なお時間があ れば、同データの空間ARCHモデルによる空間ボラティリティの分析(佐藤宇樹氏 (東北大学D1)による)も紹介したい。 (P. J. Brockwell氏(Colorado state univ.)と共同研究)

 

日時

2016年4月22日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

Mike So (Hong Kong University of Science and Technology)

Bayesian Inference Under Quasi-likelihoods

要旨(Abstract) Bayesian approach can effectively deal with a wide range of complicated statistical problems like high-dimensional inference, latent variable filtering and statistical learning. In classical Bayesian analysis, we need to fully specify the likelihood of underlying models so as to carry out statistical computation for posterior inference. The requirement of likelihood limits the application of Bayesian approach in solving semi-parametric problems or problems whose full likelihood is computationally intractable. In this paper, we propose an approximate Bayesian inference framework which incorporates quasi-likelihoods. It is expected that without the need of full likelihood specification, we can extend the scope of problems which Bayesian inference can solve. Two examples, one on econometric analysis and the other on information security analysis, are taken to demonstrate our framework. Results in the examples show that the approximate Bayesian inference framework can provide both consistent estimates as well as good credible interval coverage.

 

日時

2016年5月6日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

本田敏雄 (一橋大学)

Forward variable selection for sparse ultra-high dimensional varying coefficient models

Efficient estimation in semivarying coefficient models for longitudina/clustered data
要旨(Abstract)

要旨(Forward variable selection for sparse ultra-high dimensional varying coefficient models)

Varying coefficient models have numerous applications in a wide scope of scientific areas. While enjoying nice interpretability, they also allow flexibility in modeling dynamic impacts of the covariates. But, in the new era of big data, it is challenging to select the relevant variables when there are a large number of candidates. Recently several works are focused on this important problem based on sparsity assumptions; they are subject to some limitations, however. We introduce an appealing forward variable selection procedure. It selects important variables sequentially according to a reduction in sum of squares criterion and it employs a BIC-based stopping rule. Clearly it is simple to implement and fast to compute, and it possesses many other desirable properties from both theoretical and numerical viewpoints. Notice that the BIC is a special case of the EBIC, when an extra tuning parameter in the latter vanishes. We establish rigorous screening consistency results when either BIC or EBIC is used as the stopping criterion, although the BIC is preferred to the EBIC on the bases of its superior numerical performance and simplicity. The theoretical results depend on some conditions on the eigenvalues related to the design matrices, and we consider the situation where we can relax the conditions on the eigenvalues. Results of an extensive simulation study and a real data example are also presented to show the efficacy and usefulness of our procedure. This is joint work with Ming-Yen Cheng and Jin-Ting Zhang.

 

要旨(Efficient estimation in semivarying coefficient models for longitudina/clustered data)

In semivarying coe!cient modeling of longitudinal/clustered data, of primary interest is usually the parametric component which involves unknown constant coe!cients. First we study semiparametric e!ciency bound for estimation of the constant coe!cients in a general setup. It can be achieved by spline regression using the true within-subject covariance matrices, which are often unavailable in reality. Thus we propose an estimator when the covariance matrices are unknown and depend only on the index variable. To achieve this goal, we estimate the covariance matrices using residuals obtained from a preliminary estimation based on working independence and both spline and local linear regression. Then, using the covariance matrix estimates, we employ spline regression again to obtain our final estimator. It achieves the semiparametric e!ciency bound under normality assumption and has the smallest asymptotic covariance matrix among a class of estimators even when normality is violated. Our theoretical results hold either when the number of within-subject observations diverges or when it is uniformly bounded. In addition, the local linear estimator of the nonparametric component is superior to the spline estimator in terms of numerical performance. The proposed method is compared with the working independence estimator and some existing method via simulations and application to a real data example. This is joint work with Ming-Yen Cheng and Jialiang Li.

 

日時

2016年5月20日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

國濱剛 (名古屋大学)

Nonparametric Bayes models for mixed-scale longitudinal surveys

要旨(Abstract) Modeling and computation for multivariate longitudinal surveys have proven challenging, particular when data are not all continuous and Gaussian but contain discrete measurements. In many social science surveys, study participants are selected via complex survey designs such as stratified random sampling, leading to discrepancies between the sample and population, which are further compounded by missing data and loss to follow up. Survey weights are typically constructed to address these issues, but it is not clear how to include them in models. Motivated by data on sexual development, we propose a novel nonparametric approach for mixed-scale longitudinal data in surveys. In the proposed approach, the mixed-scale multivariate response is expressed through an underlying continuous variable with dynamic latent factors inducing time-varying associations. Bias from the survey design is adjusted for in posterior computation relying on a Markov chain Monte Carlo algorithm. The approach is assessed in simulation studies, and applied to the National Longitudinal Study of Adolescent to Adult Health.

 

日時

2016年6月24日(金 Friday)16:50-18:35
※ マクロワークショップと共催

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

※2008SNA移行に関する説明会
長谷川秀司 (内閣府)

国民経済計算の新基準「2008SNA」への移行について

要旨(Abstract) 今年末、我が国の国民経済計算(JSNA)は、16年ぶりに準拠する国際基準を変更 し、「1993年SNA」から「2008SNA」に移行する予定である。  「2008SNA」では、「ニューエコノミー」の展開、グローバリゼーション、金融市 場の発展等近年の経済・金融環境の変化を織り込んだ各種の概念・範囲の変更を行っ ている。例えば、企業の研究・開発(R&D)は、従来、生産活動にて中間消費される 扱いだったが、「2008SNA」では知識ストックの蓄積(固定資産の「知的財産生産 物」)と捉え、最終需要の総固定資本形成に計上することになる。また、これまで捕 捉・計上していなかった雇用者ストックオプションが、新たに雇用者報酬や金融資産 に記録される。  ワークショップにおいては、このような「2008SNA」の特徴や基本的な考え方、ま たGDP等マクロ経済の各集計値への影響のイメージについて説明し、統計ユーザーの 利便性の向上に資することを目指したい。

 

日時

2016年7月8日(金 Friday)16:50-18:35
※ ミクロ実証分析ワークショップと共催

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

Tetsuya Kaji (Massachusetts Institute of Technology)

Post-Observation Sample Selection and Integrable Empirical Processes

要旨(Abstract) Many empirical researches entail estimators that come from sample selection after observation; one might give estimates without outliers together with estimates with full sample to claim that his results are not the spurious consequence of a handful of rare observations; one might contrast the treatment effects of many subgroups, say poverty levels, to derive effective policy implications; one might consider the joint variation of the expected return and the conditional Value-at-Risk (CVaR) of a portfolio to better decide investment. Not much is known, however, as to the joint statistical properties of such estimators, and hence, the discussion has been prone to heuristics. This paper provides a statistical framework to deal with these estimators jointly, using what I call the /integrable empirical processes/. The proposed theory allows one to conduct a formal statistical test of outlier sensitivity, of multigroup analyses, or derive the joint distribution of many conditional returns of a portfolio. From a statistical perspective, the theory bridges the gap between the multivariate central limit theorems and the uniform central limit theorems. As an empirical application, I revisit the outlier robustness analyses discussed in Alatas et al. (2016) and Acemoglu et al. (2016).

 

日時

2016年7月29日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

浦拓哉 (University of California, Davis)

Heterogeneous Treatment Effects with Mismeasured Endogenous Treatment

要旨(Abstract) This paper studies the identifying power of an instrumental variable in the nonparametric heterogeneous treatment effect framework when a binary treatment is mismeasured and endogenous. I characterize the sharp identified set for the local average treatment effect under the exclusion restriction of an instrument and the deterministic monotonicity of the true treatment in the instrument. Even allowing for general measurement error (e.g., the measurement error is nonclassical and endogenous), it is still possible to obtain finite bounds on the local average treatment effect. Notably, the Wald estimand is an upper bound on the local average treatment effect, but it is not the sharp bound in general. I also provide a confidence interval for the local average treatment effect with uniformly asymptotically valid size control. Furthermore, I demonstrate that the identification strategy of this paper offers a new use of repeated measurements for tightening the identified set.

 

日時

2016年9月16日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

Paul Embrechts (ETH Zurich)

An Extreme Value Approach for Modeling Operational Risk Losses Depending on Covariates

要旨(Abstract) In financial risk management, Operational Risk data typically appear as entries in a BLxRT-matrix where BL stands for the number of business lines, and RT corresponds to risk types. For instance (BL) Corporate Finance and (RT) Internal Fraud. Banks and insurance companies of ten, at least for internal purposes, model Operational Risk losses based on such a data matrix and use a particular risk measure to be statistically estimated. From a mathemat ical point of view the (internal) data available consists of BLxRT marked point processes. A typical example consists of a (BL=8, RT=7)-matrix, with historical data in each cell. As risk measure one often takes a high quantile of the total matrix loss distribution function over a one year horizon (referred to in the industry as a one-year Value-at-Risk). In order to analyze this problem we introduce a dynamic version of Extreme Value Theory (EVT) introducing as co-variables rows, columns from the data matrix as well as time. The Operational Risk example is just mentioned as a motivating example, the general EVT methodology discussed is applicable well beyond this example. This talk is based on joint work with Valerie Chavez-Demoulin (EPF Lausanne) and Marius Hofert (University of Waterloo).

 

日時

2016年9月26日(月 Monday)12:10-13:10  ※開催日時にご注意下さい。

主催:ミクロ実証分析ワークショップ

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

岡達志 (National University of Singapore)

Quantile Treatment Effects in Difference in Differences Models under Dependence Restrictions and with only Two Time Periods (with Brant Callaway and Tong Li)

要旨(Abstract) This paper shows that the Quantile Treatment Effect on the Treated (QTT) can be identified using a combination of (i) a Distributional Difference in Differences Assumption and (ii) an assumption on the dependence between the change in untreated potential outcomes and the initial level of untreated potential outcomes for the treated group. The second assumption recovers the unknown dependence from the observed dependence for the untreated group. This result extends previous research that required at least three periods of data for identifying the QTT under a similar setup. We also provide identification results when the assumptions hold only after conditioning on observed covariates. Under an additional assumption, we also show that the QTT is identified when only repeated cross sections are available. Finally, we consider estimation and inference--we develop uniform confidence intervals and show the validity of an exchangeable bootstrap procedure.

 

日時

2016年10月28日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

蛭川雅之(摂南大学)

Consistent Estimation of Linear Regression Models Using Matched Data

要旨(Abstract) Economists often use matched samples, especially when dealing with earnings data where a number of missing observations need to be imputed. In this paper, we demonstrate that the ordinary least squares estimator of the linear regression model using matched samples is inconsistent and has a nonstandard convergence rate to its probability limit. If only a few variables are used to impute the missing data, then it is possible to correct for the bias. We propose two semiparametric bias-corrected estimators and explore their asymptotic properties. The estimators have an indirect-inference interpretation and they attain the parametric convergence rate if the number of matching variables is no greater than three. Monte Carlo simulations confirm that the bias correction works very well in such cases.

 

日時

2016年11月18日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

Shih-Han Chang (Sailthru Co.)

Comparison of Bayesian and Frequentist Multiplicity Correction For Testing Mutually Exclusive Hypotheses Under Data Dependence

要旨(Abstract) The problem of testing mutually exclusive hypotheses with dependent test statistics is considered. Bayesian and frequentist approaches to multiplicity control are studied and compared to help gain understanding as to the effect of test statistic dependence on each approach.The Bayesian approach is shown to have excellent frequentist properties and is argued to be the most effective way of obtaining frequentist multiplicity control, without sacrificing power, when there is considerable test statistic dependence.

 

日時

2016年12月2日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

塚原英敦 (成城大学)

The empirical beta copula and its applications

要旨(Abstract) The empirical beta copula is introduced by a simple idea of rearranging uniform random variates in the order specified by the componentwise ranks of the original sample. It turns out to be a special case of the empirical Bernstein copula, the degrees of all Bernstein polynomials being equal to the sample size. Necessary and sufficient conditions are given for a Bernstein polynomial to be a copula, and they imply that the empirical beta copula is a genuine copula. Furthermore, the empirical process based on the empirical Bernstein copula is shown to be asymptotically the same as the ordinary empirical copula process under fairly weak assumptions. A Monte Carlo simulation study shows that the empirical beta copula outperforms the empirical copula and the empirical checkerboard copula in terms of both bias and variance. Compared with the empirical Bernstein copula with the smoothing rate suggested in the literature, its finite-sample performance is still significantly better in several cases, especially in terms of bias. Some resampling schemes using the empirical beta copula are explored to see if any beneficial effect on the accuracy of resampling schemes for the empirical copula process.

 

日時

2016年12月22日(木 Thrsday) 13:00-17:40  

場所

東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第2セミナー室 [地図]
in Seminar Room 2 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]

※日時・場所が通常と異なりますのでご注意下さい。

報告

科学研究プロジェクト 「経済リスクの統計学の新展開:稀な事象と再起的事象」
オーガナイザー:国友直人(明治大学)・大森裕浩(東京大学)

要旨(Abstract) <セッションI:金融市場の統計分析>
Chair: 大屋幸輔
13:00〜13:40「Dynamics of Integration in East Asian Equity Markets」Tatsuyoshi Okimoto (Australian National University)
13:40〜14:20「Simultaneous multivariate point process models with an
application to causality analysis of financial markets」国友直人・栗栖大輔・天野裕介・粟屋直

休憩

<セッションII:経済リスクの統計的基礎>
Chair: 大森裕浩
14:25〜15:05「On Greeks」楠岡成雄
15:05〜15:40「On rare events」Tomoyuki Ichiba  (University of California,
Santa-Barbara)
15:45〜16:15「Discretization of Self-Exciting Peaks Over Threshold Models」栗栖大輔

休憩

<セッションIII:保険リスクと統計学>
Chair:川崎能典
16:20〜17:00「Dynamic risk measures for stochastic asset processes from ruin theory」清水泰隆
17:00〜17:40「日本人の寿命 -過去・現在・未来-」田中周二・長谷川敏彦・伊藤憲祐

 

日時

2017年1月20日(金 Friday)16:50-18:35

場所 東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]
報告

小池祐太(統計数理研究所)

高頻度データにおけるリード・ラグ効果のウェーブレット解析

要旨(Abstract) 本報告では, 2つの金融資産間のリード・ラグ効果を調べるための新たなモデル を提案する. 提案モデルでは, Brown運動に基づく連続時間確率過程モデリング と既存の離散時間モデルにおけるウェーブレットによるリード・ラグ解析の間の ギャップを埋め, リード・ラグ効果の多時間尺度構造を分析することが可能にな る. 更に, 提案モデルにおいて時間尺度ごとにリード・ラグ効果を分析するため の統計的方法論を与え, 確率ボラティリティ・非正則サンプリングを含むような 状況に適用可能な漸近理論を構築する. 最後に, 簡単な数値実験およびデータ分 析例を示す。本研究は慶應義塾大学林高樹教授との共同研究である.

 

日時

2017年2月2日(木 Thursday) 13:00-18:05

場所

東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第2セミナー室 [地図]
in Seminar Room 2 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]

※日時・場所が通常と異なりますのでご注意下さい。

報告

科学研究プロジェクト「経済統計・政府統計の理論と応用」
オーガナイザー: 国友直人・山本拓(政府統計関係)

要旨(Abstract) <挨拶>
13:00〜13:05「研究プロジェクトの計画」山本拓(統計研究会)

<セッションI> 政府統計・経済統計の諸問題1
Chair: 山本拓(統計研究会)
13:05〜13:40 「世帯類型別の貯蓄率の動向」宇南山卓(一橋大学)・大野太郎 (信州大学)
13:40〜14:15 「景気指標としての個人消費関連統計の比較研究」川崎茂(日本大 学)
14:15〜14:50 「自記式調査における複数回答方式をめぐる諸問題」土屋隆裕(統 計数理研究所)

休憩

<セッションII>政府統計・経済統計の諸問題2
Chair: 星野伸明(金沢大学)
15:00〜15:35 「SNA統計における時系列データ作成上の課題−ベンチマーキング及 び季節調整を中心に−」長谷川秀司(内閣府)
15:35〜16:10 「内閣支持率と株価の因果関係」川崎能典(統計数理研究所)

休憩

<セッションIII>経済統計の理論と応用
Chair: 国友直人(明治大学)
16:20〜16:55 「小地域推定の現状と課題」久保川達也(東京大学)
16:55〜17:30 「真のデータに線形制約がある際の観測誤差の修正−国民経済計算 の場合」千木良弘朗(東北大学)・山本拓(統計研究会)
17:30〜18:05 「Double Filter Instrumental Variable Estimation of Panel Data Models with Weakly Exogenous Variables」 Breitung, J.(ケルン大学) ・早川和彦(広島大学)・ M. Qi(広島大学)

 

日時

2017年3月7日(火 Tuesday) 16:50-18:35

主催:ミクロ実証分析ワークショップ
共催:ミクロ経済学ワークショップ

場所

東京大学大学院経済学研究科 学術交流棟 (小島ホール)1階 第1セミナー室 [地図]
in Seminar Room 1 on the 1st floor of the Economics Research Annex (Kojima Hall) [Map]

※日時・場所が通常と異なりますのでご注意下さい。

報告

Shakeeb Khan (Duke University)
"Adaptive Inference in Semiparametric Multinomial Response Models" (joint with F. Ouyang and E. Tamer)

要旨(Abstract) We consider identification, estimation and inference on regression coefficients in semi parametric multinomial response models. Our identification result is constructive and estimation is based on a localized rank objective function, loosely analogous to that used in Abrevaya, Hausman, and Khan (2010). We we show this achieves sharp identification which is in contrast to existing procedures in the literature such as, for example, Ahn, Powell, Ichimura, and Ruud (2014). In that sense, our procedure is adaptive (Khan and Tamer (2009)) in the sense that it provides an estimator of the sharp set when point identification does not hold, and a consistent point estimator when it does. Furthermore, our rank procedure extends to panel data settings for inference in models with fixed effects, including dynamic panel models with lagged dependent variables as covariates. A simulation study establishes adequate finite sample properties of our new procedures.