2011年 4月 26日 9:59 JST　
　チャールズ・シューマー米上院議員（民主、ニューヨーク州）は、中国人民銀行（中央銀行）の周小川総裁と先週会談したが、同総裁ら中国当局者が人民元をもっと急速に上昇させることに消極姿勢をみせたと述べ、不満を表明した。同議員は中国の通貨政策に対する批判の急先鋒。

　9人の米上院議員とともに中国を訪問したシューマー議員は周総裁と2人だけで朝食を共にした。同議員は中国からの帰途、電話インタビューに応じ、人民銀行側が中国は現在の人民元上昇ペースを速めるつもりはないとの姿勢をみせたと述べた。中国側のこの姿勢は米議会で批判を浴びそうだ。

　シューマー議員は「わたしは周総裁から、行動する用意があることは既にそうしていると考えているとの明確な印象を得た」と述べた。
イメージ AFP/Getty Images

　同議員によると、周総裁は、人民元を緩やかに上昇させるなどでインフレとの戦いに成功するだろうと自信を示した。そして、中国が人民元上昇を加速化させるつもりはあるのかと質問したところ、同総裁は中国当局は通貨について、インフレに対処するために使う幾つかの道具の一つと認識していると語るにとどまったという。3月の中国のインフレ率は前年同月比で5.4％で、1年前の2倍のペース。

　中国政府が昨年6月、金融危機の間ストップしていた人民元の上昇をドルに対して再び上昇させると発表。それ以来、対ドルで4％以上上昇した。しかしシューマー議員は、人民元の上昇ペースは受け入れられないと批判している。同議員は「中国では、（人民元大幅上昇という）中国自身の小さな犠牲は、世界の他の諸国の被る大きな犠牲よりも、さらに大きいとみなされているようだ」と述べた。

　シューマー議員は周総裁と2人だけで朝食会に臨んだが、上院議員グループは他の中国政府当局者と別個に会見した。

　シューマー議員は今年に入って、通貨を管理しているとみられる諸国に対して、懲罰的な関税導入など強硬な措置をとるようオバマ政権に義務付ける法案を支持した。しかし、同法案などは議会を通過しなかった。下院は昨年、独自の中国通貨関連法案を可決した。しかし、その後の中間選挙で下院の過半数議席を牛耳った共和党はこれまでのところ、同法案をさらに推し進めようとしていない。

　シューマー議員は、周総裁ら中国当局者の発言を受けて、中国の為替政策を是正させるため、対中法案を通過させる努力が議会で強まるだろうと述べた。同議員は「（議会での）気運は盛り上がっている。なぜなら、いったん予算で合意すれば、雇用問題がナンバーワンの課題になり、中国（の為替政策がもたらす米雇用への影響）は最大の懸念事項になるからだ」と語った。

　同議員はまた、中国当局者がインフレをコントロールしていると主張したと述べ、「わたしは、彼らがインフレを制御できているとの見方に賛成しない。政府は単にマネーを供給しているだけで、よく微調整されているとは思わない」と語った。

記者: Michael R. Crittenden

2011年 4月 26日 13:37 JST　

　中国の年間10％の成長率はいつまで続き、低迷する世界経済のけん引役でいられるのかという古くから疑問が、中国ウォッチャーのエコノミストの間で最近再浮上している。モルガン・スタンレーの3人のアナリストもこの議論に加わっている。

画像を拡大する
イメージ
Associated Press

　3人はいずれもそれほど長くは続かないだろうが、減速は小幅にとどまり、年間経済成長率は2020年まで8％程度になると予想している。これでも、ほとんどの国がうらやむような記録的な成長率だ。

　中国経済に対する楽観・悲観のいずれの論者も、以下のような一定の前提があることには同意している。

　−どの国も永久に年間10％の成長率を維持できない。中国は30年間でそれを達成したため世界史的には特異な存在。

　−中国は投資戦略からのリターンが減少するようになるため、投資にあまり頼らずに消費支出に依存するよう国内経済を「再均衡化」する必要がある。

　−中国は急速に高齢化している。これは通常、経済活力にとって悪い兆候。

　−中国は技術革新を行う必要がある。継続した生産性の向上は継続した高度成長にとって重要になる。

　論者の見解が分かれるところは、中国はマイナス要因をプラスに転じることができるかどうだ。

　モルガン・スタンレーのQing Wang、Steven Zhang、Ernest Hoの3アナリストによると、購買力平価で算出した1人当たり国内総生産（ＧＤＰ）が約7000ドルに達すると、経済成長率はしばしば急速に減速する傾向がある。中国は08年にその水準を超えた。「歴史が指針となり、重力の法則が中国でも通用すれば、中国の経済成長は鈍化する見通し」という。1人当たりＧＤＰが7000ドルに達した時、日本と韓国でも同様な事態が起きた。

　しかしWang氏によると、この2国とは違い、中国は切り札的な存在だという。中国は大陸規模の経済であるため、その地理的な規模と人口の多さを駆使して、高度成長を続けることが可能という。中国は、一般普及品の製造拠点を低賃金の西部に移転させるほか（この傾向はすでに起きている）、地方の若者をプラントや工場のある都市部に移すことを奨励することができる。農村部から都市部への移住は、中国人口の高齢化というマイナス効果を減殺できるという。

　Wang氏は「（世界の工場として）インド以外に中国の替わりになるような国の候補はない」と指摘。「しかし、インドはインフラが未整備ですぐにはその準備ができていない」。

　最も議論を呼ぶかもしれない点は、この3アナリストが予測する「消費の黄金時代」。中国の消費がＧＤＰに占める比率は世界では有数の低さを示している。これについて多くのアナリストは、用心深い中国人は老後の蓄えや医療費の備えとして貯蓄していることを理由に挙げている。しかし、Wang氏によると、中国人は政府が新たな社会セーフティネットを構築するのを待ってから支出するようなことはしないという。彼らは金を稼げるのを待っている。今後10年で労働需要の増加で賃金が上昇するに従い、金を稼ぐだろう。韓国と日本では1人当たりＧＤＰが7000ドルに達した時、消費が拡大した。

　Wang氏によると、大きなマイナス要因は、中国の物価が上昇する可能性があることだ。過去10年のインフレ率は平均1.9％だったが、2020年までにそれが倍かそれ以上に上昇する可能性があるという。それは中国が世界のインフレ源になることを意味する。世界は中国製品を輸入するが、かつての中国製品は物価安定源でデフレでさえあったわけだが。

　確かにモルガン・スタンレーのこの3アナリストは楽観的すぎる結論に達している可能性がある。カリフォルニア大学バークレー校のバリー・アイケングリーン、アジア開発銀行のDonghyun Park、高麗大学のKwanho Shinの3氏は先月発表したリポートで、中国経済が減速する時期を独自に算出し、同国は2015年前後でスピードバンプに激突すると指摘。

　このリポートでは、他の急成長国でも一定の所得水準に達すれば、少なくとも2％ポイント成長率が低下するとしている。中国の場合は8％となり、実質的にはモルガン・スタンレーのアナリストらの予測と同じだ。しかしDonghyun Park氏らのリポートによると、他のエコノミストらが中国の経済成長率は現在から2030年の間に5〜6％に減速する可能性があると推定していることが指摘されている。これはモルガン・スタンレーのアナリストらの減速予測をかなり下回っている。

記者: Bob Davis

　［マドリード／ロンドン　２６日　ロイター］　ギリシャの債務再編をめぐる懸念やユーロ圏周辺国への危機波及不安がくすぶる中、スペインが２６日実施した短期証券（Ｔビル）入札では、落札利回りが前回から大幅に上昇した。

　スペイン財務省が実施した１９億７０００万ユーロ（２８億６０００万ドル）のＴビル入札では、３カ月物Ｔビルの平均利回りが１．３７１％と、３月入札時の０．８９９％から上昇。６カ月物Ｔビルも１．３６１％から１．８６７％に上昇した。

　入札は旺盛な需要を集めたが、同国がギリシャやアイルランド、ポルトガルなどの周辺国危機から無縁でいられるとの確証は乏しい。

　一方、欧州連合（ＥＵ）統計局が同日発表した２０１０年のユーロ圏財政収支統計によると、ギリシャの２０１０年の公的債務は、０９年の対国内総生産（ＧＤＰ）比

１２７．１％から１４２．８％に拡大した。

　これを受け、ギリシャ１０年債の対独連邦債利回りスプレッドは、１２０５ベーシスポイント（ｂｐ）に、ギリシャの期間５年のクレジット・デフォルト・スワップ（ＣＤＳ）スプレッドは１３４０ｂｐにそれぞれ拡大し、ユーロ導入後の最高水準を更新した。

　またギリシャの２０１０年の財政赤字の対ＧＤＰ比率は１０．５％と、直近の政府予想９．６％を上回った。ギリシャ財務省はこれについて、予想以上に深刻な景気後退で税収や社会保障費に影響が出たためと説明。欧州連合（ＥＵ）と国際通貨基金（ＩＭＦ）の経済プログラムに基づき定められた財政赤字目標を達成するため、必要なあらゆる措置を講じる方針をあらためて表明した。

　コメルツ銀行のエコノミスト、ラルフ・ソルビーン氏は「ギリシャの２０１０年財政赤字が２けたの規模になったことで、ギリシャの債務再編の議論がさらに熱を帯びるだろう」と述べた。

　ＥＵ統計局が発表した財政収支データでは、加盟国の多くが昨年、財政赤字削減に着手したことが示されたが、公的債務は域内ほぼすべての国で膨らんだ。

　スペインの公的債務はＧＤＰ比６０．１％と、仏独の水準を２０％ポイント超下回っており、比較的低水準にとどまっているが、市場の懸念は貯蓄銀行の資本増強コストに加え、同国が長引く不動産市場崩壊による影響に耐えられるのかどうかに集中している。

　一方、関係筋によると、ポルトガル支援策をめぐる同国と、欧州委員会、欧州中央銀行（ＥＣＢ）、ＩＭＦとの協議は、復活祭の週末も引き続き行われた。

　双方とも６月５日のポルトガル総選挙実施前の合意実現に向け、５月１６日に開催される欧州の財務相会合に間に合わせるよう、条件に関する詰めの協議を急いでいるという。

　［ダブリン　２６日　ロイター］　アイルランドが欧州連合（ＥＵ）に対して提出した文書によると、同国政府は２０１１年の純借入金は１４８億ユーロ（２１６億ドル）、国内総生産（ＧＤＰ）は１５６０億ユーロになると試算していることが分かった。

2010年ユーロ圏財政赤字が縮小、ギリシャは政府予想上回る
2011年 04月 26日 23:47 JST

　［ブリュッセル　２６日　ロイター］　欧州連合（ＥＵ）統計局によると、２０１０年のユーロ圏の財政赤字は対国内総生産（ＧＤＰ）比６．０％となり、０９年の同６．３％から縮小した。ただ、ギリシャやポルトガルの財政赤字は政府の予想上回った。

　２０１０年はドイツ、アイルランド、ルクセンブルク、オーストリア以外全てのユーロ圏諸国で財政収支が改善した。一方、公的債務の対ＧＤＰ比は２００９年の７９．３％から８５．１％に上昇。エストニア以外全てのユーロ圏諸国で拡大した。

　ギリシャの財政赤字の対ＧＤＰ比は２００９年の１５．４％から１０．５％に低下した。欧州委員会とギリシャ政府の予想は９．６％だった。債務は２００９年の対ＧＤＰ比１２７．１％から１４２．８％に拡大した。

　ギリシャ財務省は財政赤字が対ＧＤＰ比で政府予想を上回ったことについて、景気後退が予想以上に深刻だったことが理由とし、ＥＵおよび国際通貨基金（ＩＭＦ）による金融支援プログラムの下での目標を達成するためあらゆる措置を講じる考えを示した。

　２０１０年のポルトガルの財政赤字は対ＧＤＰ比９．１％となり、これも政府予想の８．６％を上回った。

　アイルランドの財政赤字の対ＧＤＰ比は２００９年の１４．３％から３２．４％に上昇、債務も６５．６％から９６．２％に上昇した。
　同文書はアイルランド政府が３月３１日にＥＵに提出した。アイルランド財務省関係者によると、２９日に発表される総合的な成長見通しが政府の最新の見通しとなる。

　２０１１年のＧＤＰが１５６０億ユーロとなった場合、同年の経済成長率は１．３４％となる。

　政府が２０１０年１２月に発表した２０１１年の成長率見通しは１．７％だった。ただ、国際通貨基金（ＩＭＦ）はこのほど同国の成長率見通しを０．５％に引き下げている。

ギリシャが財政健全化への取り組み再表明、赤字拡大修正受け
2011年 04月 27日 04:29 JST

ワールド
ロシア大統領「情報開示が国の義務」、チェルノブイリ事故25年で
ＲＪ／ＣＲＢ指数、0.10％安の366.19
アイルランドの2011年純借入金は148億ユーロ＝政府文書
カーター氏が平壌入り、金正日総書記らとの会談に意欲

　［アテネ　２６日　ロイター］　ギリシャ財務省は２６日、同国は欧州連合（ＥＵ）と国際通貨基金（ＩＭＦ）の経済プログラムに基づき定められた財政赤字目標を達成するため、必要なあらゆる措置を講じるとの考えを示した。

　財務省は、昨年の財政赤字の対国内総生産（ＧＤＰ）比率が１０．５％に拡大修正されたことを受け、「ギリシャ政府は引き続き赤字目標の達成にコミットしていく。赤字幅がかい離した主因は、予想以上に深刻な景気後退で税収や社会保障費に影響が出たためだ」との声明を発表した。

　その上で医療関連支出や税・社会保障費の徴収強化などでさらなる取り組みが必要との考えを示し、「目標達成に向け必要な措置はすべて、５月１５日までに議会に提出される中期の財政戦略に盛り込まれる」と言明した。

　ただ投資家は引き続きギリシャ国債に売りを浴びせており、ギリシャ１０年債の対独連邦債利回りスプレッドはこの日、１２０５ベーシスポイント（ｂｐ）に、ギリシャの期間５年のクレジット・デフォルト・スワップ（ＣＤＳ）スプレッドは１３４０ｂｐにそれぞれ拡大し、ユーロ導入後の最高水準を更新した。

　ＥＦＧユーロバンクのエコノミスト、プラトン・モノクロソス氏は「２０１０年の財政赤字規模および２０１１年第１・四半期の歳入の低迷を踏まえると、２０１１年の財政赤字削減目標を果実に達成するには、３０億ユーロ以上の追加措置が必要になる」と指摘した。

　ギリシャは今年、財政赤字の規模をＧＤＰ比７．４％まで圧縮することを目指している。

　［ニューヨーク　２６日　ロイター］　２６日のニューヨーク外国為替市場では、ユーロが対ドルで一時１年４カ月ぶりの高値をつけた。米連邦準備理事会（ＦＲＢ）の利上げは他の主要中央銀行よりも遅れるとの見方が強く、ユーロは一段高となる可能性がある。

　米連邦公開市場委員会（ＦＯＭＣ）はきょうから２日間の会合を開始した。バーナンキＦＲＢ議長は２７日の会合終了後に初めての定例記者会見を予定しており、ＦＲＢの超低金利政策からの出口戦略を見通す上で非常に注目されている。

　ユーロは一時下落していたものの、スペインが実施した約２０億ユーロの短期証券（Ｔビル）入札が順調に消化されたことで安心感が広がり、値を戻した。

　終盤時点でユーロ／ドルは０．４％高の１．４６３８ドル。

　ユーロ／ドルは今後、１．４８ドルを試す見通し。外為テクニカルアナリストによると、バーナンキ議長が緩和的金融政策を当面続ける可能性を示唆した場合、ユーロ／ドルはさらに１．５０ドルに向かう可能性があると指摘する。

　欧州中央銀行（ＥＣＢ）は今月、２００８年７月以来となる利上げに踏み切った。ＦＲＢはＥＣＢよりも政策引き締めにはるかに消極的で、その違いがユーロの年初来９．４％上昇の主な要因となっている。

　ドル／円は０．４％安の８１．５２円。一時は４週間ぶりの安値をつけた。

　ドル指数は０．３％安の７３．７９３。トレーダーは、先週つけた３年ぶり安値７３．７３５を試す可能性があると指摘する。

　豪ドルは終盤で０．５％高の１．０７７７米ドル。一時、変動相場制移行後の高値となる１．０７９２米ドルをつけた。

© Thomson Reuters 2011 All rights reserved.

　　４月26日（ブルームバーグ）：お祝いのケーキもパーティーもないが、ギリシャは先週末に救済申請１周年を迎えた。１周年のムードは最悪と言ってよい状態だ。

　　国債利回りは最高水準を繰り返し更新。クレジット・デフォルト・スワップ（ＣＤＳ）を使って国債を保証するコストも過去にない水準に跳ね上がった。財務省は債務再編が近いとのうわさの出所を突きとめようと、銀行の従業員の刑事捜査を開始した。

　　ギリシャは救済後の１年を、債務再編の発表で締めくくるべきだった。ギリシャがデフォルト（債務不履行）に伴う「ヘアカット（債務減免）」を発表するのは早い方が、誰にとってもよい結果になる。遅れればヘアカットによる減免幅が拡大するし、経済学者の研究は、減免幅が大きいほど後遺症も重いことを示している。現状ではギリシャ経済の傷みが激し過ぎるし、ユーロ圏諸国がデフォルトするという概念が受け入れられれば、共通通貨の問題の解決に向けた賢明な意見のやりとりを始められるようになる。

　　過去１週間の取引は、市場参加者の大半がギリシャのデフォルトを必至と考えていることを示した。先週末のギリシャ２年債利回りは22％超。よほど悪質な高利貸しでなければこんな金利は課さない。10年債利回りも15％付近。ギリシャ国債のＣＤＳスプレッドはデフォルトの確率が67％とみられていることを示唆した。

　　現実問題として、ギリシャ債の保証を購入することにあまり意味はない。向こう１年の間にロンドンに雨が降ることに対する保証を買うようなものだ。起こるか起こらないかではなく、いつ起こるかの問題だ。

　　　　　　　　　　お芝居はやめにしよう

　　ギリシャ政府と欧州連合（ＥＵ）の当局者らはいまだに、デフォルトという選択肢は存在しないと言い続けている。そんな芝居はもうやめるべきだ。

　　市場はデフォルトを最悪の破局として描きがちだ。これはコメントするバンカーや運用者の大半がデフォルトで損失を被る立場にあるためだ。実際には国家の債務不履行は今までにも起こっている。問題なのはへアカットの幅だ。

　　英国経済学会の会議で今月発表されたフアン・クルシズ、クリストフ・トレベシュ両氏の論文は、1970年以降に起こった国家と外国銀行、債券保有者の間での債務再編を研究したもの。68カ国での202件を検証した結果、「債務再編でのヘアカットの幅が大きい（債権回収率が低い）ほど、その後の市場でのスプレッド（調達のコスト）も大きく、資本市場から締め出される期間も長くなる」ことが分かった。つまり、債券保有者に負わせた損失が大きいほど、市場からは厳しく罰せられ、資本市場での調達が可能になるまでに時間もかかる。

　　　　　　　　　　2013年まで待てない

　　ギリシャの債務状況は悪化しつつある。再編を遅らせることは選択すべきではない。40−50％の債務減免を今年、債券保有者に求める方が、2013年になって70−80％の債権放棄を迫るよりましだ。

　　また、現在のギリシャ経済への打撃は余りにもひどい。失業率は15％を超え、中央銀行は今年さらに３％のマイナス成長を予想している。ギリシャ以外の世界経済が緩やかながらも持続的に回復しているにもかかわらずだ。

　　政府は歳出削減を続けている。今月はさらに220億ユーロ（約２兆6300億円）の削減計画を発表したが、これは経済を一段と落ち込ませるばかりだ。輸出が内需の落ち込みを補う兆候は見られない。歳出削減だけで債務危機を脱することは不可能だ。ある時点で、経済が再び成長し始めなければならない。今までのところ誰も、その処方せんを示していない。

　　　　　　　　　　　　　第二案へ

　　さらに、ギリシャがデフォルトとなれば、ユーロ圏を組み立て直す方法についての真剣な議論が始まる。第一案はギリシャを救済し経済を軌道に乗せ、危機感染を防ぐことだった。しかしこの案が機能しなかったことは明々白々だ。ギリシャには回復の兆しが見られず、アイルランドとポルトガルも救済を要請せざるを得なかった。当局がそれでも第二案に乗り換えようとしないなら、どうやって彼らを説得したらいいのか分からない。

　　ギリシャは避けられない運命を受け入れるべきだ。50％のヘアカットを宣言し、残りの債務については３年間、利払いを停止する。債券保有者には経済成長に連動した追加の支払いを約束しよう。

　　債務返済に充てるはずだった資金で、ギリシャは経済再建を開始できる。需要を回復し、競争力のある輸出産業を育成する。これがユーロ圏で生き残る唯一の道だ。他の欧州諸国はギリシャのデフォルトを回避するという負け戦をやめ、ユーロ圏のより良い仕組みの構築に集中的に取り組むことができる。

　　苦しみをこれ以上引き伸ばすことには意味がない。救済２周年まで待つなど論外だ。遅くとも５月末までに決断すれば、その後は皆、前を向いて歩き始めることができる。（マシュー・リン）

（リン氏は、ブルームバーグ・ニュースのコラムニストです。このコラムの内容は同氏自身の見解です）

原題：Greece Haircut for Bondholders AlreadyOverdue: Matthew Lynn（抜粋）
更新日時: 2011/04/27 03:14 JST

日本の高級品市場、震災前の水準にいずれ回復＝仏ディオール
4月上旬は1689億円の輸入超、月間でも赤字の可能性高まる
日本のアウトルック「ネガティブ」に変更、格付け据え置き＝Ｓ＆Ｐ
3月小売業販売額は13年ぶり減少幅、落ち込みは全国に拡大

　［コルドバ（スペイン）　２６日　ロイター］　欧州中央銀行（ＥＣＢ）のゴンサレスパラモ専務理事は２６日、ユーロ圏の政策金利は依然として非常に緩和的な水準にあると指摘し、低金利政策の終了を想定すべき、との考えを明らかにした。　

　同専務理事は「低金利の時代が終了する兆しがある。金利が異例に緩和的な時代が終了するというメッセージは民間セクター、企業、政府に届いているだろう」と述べた。ただ、終了の時期についての詳細は語らなかった。

　ＥＣＢが今月、主要政策金利であるリファイナンス金利を２５ベーシスポイント（ｂｐ）引き上げて１．２５％としたことについては、一連の利上げの開始を意味するわけではないとし、金利は毎月、その時点のデータを踏まえて決めると述べた。

　同専務理事はまた、ＥＣＢはユーロ圏１７カ国それぞれが抱える事情により利上げを見送ることはできず、ユーロ圏全体の平均に政策を合わせる必要があったとの考えを示した。

　ＥＣＢが年内に２度の追加利上げを行うとの市場の観測については「われわれは市場の考えにコメントしない」と述べた。

© Thomson Reuters 2011 All rights reserved.

A second stylized fact is the large variability in haircut size across space and time. The
simple scatter plot (Figure 1) is based on our estimates for 􀜪􀯌􀯓 (eq. 2) and illustrates the
large dispersion in haircut size, which has increased since the 1970s. Recent years have
seen a particularly large variation, with some deals involving haircuts over 90% and
others involving haircuts as low as 5%.
The figure also plots the relative size of each restructuring in nominal US dollars,
reflected by the size of the circles. One can see that there have been a series of very
13 For the 1980s and 1990s, the most important sources where a series of reports by the IMF and a
detailed restructuring survey collected by the Institute of International Finance (see Appendix). For the
more recent period, the most important source was Sturzenegger and Zettelmeyer (2006, 2007, 2008),
who kindly shared their database of bond-by-bond haircut calculations.
12
sizable deals from the mid 1980s on. Particularly the Brady deals restructured large
volumes of debt. The cases of Mexico 1990 ($54 bn) and of Brazil 1994 ($43 bn)
involved debt volumes in the range of Argentina restructuring of 2005 ($44 bn).
Interestingly, there have been several deals in the last decade with both very large
volumes of debt affected and exceptionally high haircuts. In particular, we find that the
three largest restructurings of recent years (Argentina 2005, Russia 2000 or Iraq 2006)
all implied haircuts of more than 50%.

Table 1 provides summary statistics. We find that across all restructurings between 1970
and 2007, 􀜪􀯌􀯓 is estimated at 36%, which means that investors were able to preserve
almost two-thirds of their asset value in each exchange. The volume-weighted average
haircut is even lower, amounting to about 31%. The table also shows notable differences in
haircut estimates depending on the formula applied. As predicted, the market haircut at
39% tends to be larger than the SZ haircut, although the differences are surprisingly small,
amounting to only 3 percentage points on average. We find larger differences when
comparing the 􀜪􀯌􀯓 estimates to the set of naïve haircut estimates, which were computed
using a uniform 10% discount rate (hence neglecting country and global credit market
conditions) and no forward interest rates. The average naïve haircut is a mere 20%, or
nearly half the average 􀜪􀯌􀯓 haircut. When using the face value reduction type of haircuts
the average is lowest (16%). These large discrepancies with 􀜪􀯌􀯓 suggest that these two
latter measures will miss the actual creditor value loss implied in restructurings.
Overall, creditor losses turn out to be surprisingly small, at least compared to corporate
restructurings. The most comprehensive dataset on US corporate bond and banking debt
restructurings for the period 1982-2005 by Moody’s (2006) estimates the average value
weighted market haircut to be 64%.14 According to these figures, US corporate debt
restructurings implied haircuts that were twice as high as those we find for sovereign debt
restructurings, during a roughly corresponding sample period.
Looking at different decades, we find that average haircuts were significantly lower in the
1970s and 1980s as compared to recent years (25% on average). Haircuts of restructurings
in the early and mid 1990s and those implemented after 1998 were nearly twice as high on
average (of 52% and 48% respectively). However, we do not find an increase in haircuts
when comparing the subsample of 16 Brady deals (between 1990 and 1997) with the
subsample of 13 sovereign bond restructurings since 1998. On the contrary, Brady deals
implied an average haircut of 52%, which is higher than that of recent bond debt exchanges
(of 49%).
14 Haircuts of senior secured corporate debt were 47% while those on senior secured banking debt were 36%
--very similar to those on sovereign. Altman and Kishore (1996) find similar results.
13

Figure 2 plots the distribution of haircuts with and without some degree of face value
reduction. It is evident that the 1990s and 2000s feature many more restructurings with
outright debt reduction. These cases also tend to imply higher creditor losses than debt
reschedulings that only involve a lengthening of maturities (with haircuts of 65% vs. 24%
respectively).
The type of debtor also matters. In particular, we find exceptionally high haircuts in
restructurings of highly indebted poor countries (HIPCs). To show this, we categorize a
subsample of restructurings as donor supported, i.e. deals that were largely financed by the
World Bank’s Debt Reduction Facility that grants funds to governments to buy back their
debts to external commercial creditors at a deep discount (see World Bank 2007). The
average haircut in the 23 donor supported restructurings since the early 1990s is estimated
at 87%. This is more than three times as large as for restructurings in middle income
countries, but is in the range of debt relief accepted by official creditors in these same
countries (see World Bank 2010).

To compare our haircut estimates to those of others, we summarize results of 16 recent
restructurings in Table 2. For the overlapping sample, our estimates are very similar to
those of SZ. When comparing their average haircut (reported in SZ 2006) to our preferred
haircut of (eq. 2) there is a mean absolute deviation of 6 percentage points. Only two
estimates differ significantly (by more than 10 percentage points), namely Pakistan 1999
and Ukraine 2000, and this is mostly because our methodology yields significantly lower
discount rates for these two cases. We also find our results to be roughly in line with the
net present value estimates by Bank of Spain and Bank of England staff (Bedford et al.
2005 and Diaz-Cassou et al. 2008a, 2008b), with an average deviation of 8.6 and 9.1
percentage points, respectively. Our results differ more markedly from Finger and Mecagni
(2007), who often apply a 10% discount rate, and from those reported by Benjamin and
Wright (2009), who do not calculate haircuts in present value terms but base their
estimates on World Bank data on debt stock reduction and interest and principal forgiven
(including buy backs).
14
4. Theory and Identification
This section briefly discusses the theoretical framework, the key hypotheses and our
empirical strategy to test them. Our main aim is to assess the link between restructuring
outcomes (haircuts) and subsequent borrowing conditions. More specifically, we focus on
the government’s post-restructuring access to finance by analyzing (i) monthly secondary
market bond spreads and (ii) different yearly measures of exclusion from capital markets.
Theoretically, we build on Asonuma (2010) and Yue (2010) who provide clear predictions
of how haircut size will affect country access to foreign credit. Yue’s (2010) dynamic
stochastic general equilibrium model generates endogenous exclusion from financial
markets after default, where the duration of exclusion increases with the amount of debt
reduced. A bad credit record due to high arrears and/or a low recovery rate of the defaulted
debt implies longer exclusion. Asonuma (2010) extends Yue’s model by incorporating the
rate of return offered on newly-issued debt after default. In his model, forward-looking
creditors and debtors bargain not only over the size of the recovery rate, but also on the
risk premium paid on debt issues after re-entry into capital markets. His quantitative
analysis reveals that the yield spread on new debt will be higher, the lower the implied
recovery rate of the restructuring, i.e. the higher the haircut. From these models, we can
derive two testable hypotheses: Hypothesis 1: The larger the size of H, the higher the yield
spreads after restructurings; and hypothesis 2: The larger H, the longer the period of
exclusion from capital markets.
In both Yue (2010) and Asonuma (2010) the implicit mechanism is the classic reputational
one suggested by Eaton and Gersovitz (1981): A good repayment record assures access to
credit in the future. However, there could be other channels through which haircut size
affects market access conditions. First, high haircuts could be seen as a signal of bad
fundamentals and government unwillingness to pay, with adverse consequences for
country spreads and capital access (in analogy with Sandleris, 2008). Second, investors
could assess whether the size of H is “excusable”, i.e. justified by bad macroeconomic
conditions. This channel is linked to Grossmann and van Huyck (1988) who suggest a
model in which debt-servicing obligations are implicitly contingent on the realized state of
the world. Adverse reputational effects would then only occur in case of excessive or
“inexcusable” haircuts. Third, there is the countervailing effect of debt relief. Sovereigns
imposing high haircuts will reduce their indebtedness more significantly, which makes
them more solvent thereafter, at least in the short run. In an atomistic bond market without
tacit creditor collusion a la Wright (2002), new lenders may ultimately reward the resulting
lower debt to GDP ratio. Higher haircuts would then imply lower post-restructuring
spreads and quicker reaccess.
Empirically, it is challenging to test how much each of these potential channels matters.
Our key contribution is to use our new dataset to document the relationship between H and
15
the terms of subsequent capital market access. However, one extension of our analysis
dissects the size of H into its “expected” and “abnormal” or “inexcusable” components
resulting from a rather parsimonious model. We then use these two variables in our
estimations on bond spread determinants. Of course, this is but a first step. Future versions
of the paper will attempt to disentangle the mechanisms at play in more detail.
We should mention an important challenge to identifying the impact of past haircuts on
subsequent borrowing. Intuitively, it can be that countries imposing higher haircuts are also
in a worse shape at the time of the restructuring and in the periods thereafter. However, if
time and country characteristics are correlated with the lagged values of 􀜪, the estimation
may be biased and the relationships documented below may not be causal. To address this
concern, we include country and time fixed effects and control for a set of observable
fundamentals suggested by theory and the previous international finance and asset pricing
literature. This mitigates, but not necessarily completely eliminates, the possibility that our
coefficients are picking up the correlation between H and an omitted variable that
represents the ultimate cause at stake. In this version of the paper, we are replicating the
models used in a literature that claims that defaults do not imply substantial increases of
future financial costs. Our refinement allows identifying a higher cost of defaults under the
maintained hypothesis that the received empirical models are an adequate testing tool.
Future versions of this paper will address this identification concern in more depth.
5. Haircuts and Post-Restructuring Spreads: Data and Results
In order to identify post-crisis episodes, we focus on “final” restructurings only, which we
define as those that were not followed by another restructuring (vis à vis private creditors)
within the subsequent four years.15 We interpret these as being the restructurings that
successfully resolved a debt crisis. We therefore disregard intermediate restructurings,
including most deals of the early and mid 1980s that only implied short term debt relief.
This definition yields a sample of 66 final restructurings in 62 countries during the period
1980-2006.16
5.1. Dependent Variable: EMBIG Spreads
To assess the role of haircut size for post-restructuring borrowing costs, we use monthly
data on secondary market bond spreads from J.P. Morgan’s EMBI Global (EMBIG) in the
period 1993 to 2006. EMBIG spreads have been used extensively in the academic literature
to proxy country borrowing costs in international financial markets. A main advantage of
using EMBIG data is that it allows constructing a monthly panel dataset for a large number
15 In the analysis, we also exclude years with ongoing defaults (including selective default) on bank and bond
debt using Standard & Poor’s (2007) data.
16 An overview of “final” deals is provided in Table 8 below. Note that there are only four countries in the
sample with more than one final deal. The cases are Argentina (1993 and 2005), Dominican Republic (1994
and 2005), Ecuador (1995 and 2000) and Uruguay (1991 and 2003).
16
of emerging market countries whose bonds satisfy certain minimum liquidity and global
visibility benchmark, so that one would expect informationally efficient pricing. The
EMBIG is composed of U.S.-dollar denominated sovereign or quasi-sovereign Eurobonds
and Brady Bonds that are actively traded in secondary markets, as well as a small number
of traded loans.17 While the EMBIG was only introduced in January 1998, historical data
for major emerging market countries is available back to 1993. Table 3 lists the 36
countries included in our pricing analysis. Of these 36 countries, 24 countries restructured
sovereign debt with private creditors at some point since 1990, while 12 never restructured,
at least not since the mid 1980s.

5.2.Preliminary Data Analysis
We begin with a preliminary analysis of bond spreads. Figure 3 plots monthly average
spreads during post-restructuring years measured in event time. The plot reveals a notable
difference in average spreads following deals involving haircuts larger and smaller than
30%, which is roughly the sample mean for these 24 defaulters. The deviation is small in
year one and two after the restructuring, but increases from year three onward, with
differences surpassing 200 basis points (bp) on average. Given the defaulter average spread
level of about 600 bp in the overall sample, these are very sizable differences. The figure
also shows that the unconditional sample mean spread for the 12 non-defaulters in the
sample was 223 bp, a lower bound that is never pierced by any of the defaulter groups in
the 100 months after restructurings.

5.2. Estimated Model on Post-Restructuring Spreads
Since asset markets are forward looking, understanding bond yields at a given point in time
requires controlling for current and expected future conditions affecting both the prevailing
price of credit risk and expected collection. To assess the role of credit history for
sovereign borrowing costs and to facilitate comparison with the received literature, we take
a bond spread equation in the vein of those in Dell’Arriccia et al. (2006), Panizza et al.
(2009) or Eichengreen and Mody (2000). Our innovation is that we use a continuous
measure of investor outcomes instead of a default dummy variable. The empirical model
is:
17 The spread is simply the difference between the weighted average yield to maturity of a given country’s
bonds included in the index and the yield of a U.S. Treasury bond of similar maturity. In line with most other
researchers, we use stripped spreads which focus on the non collateralized portion of the emerging country
bonds (see J.P. Morgan 2004 for details).
17
          
i N t T
X u
S I i t I i t I i t I i t I i t H
i t i t it
it i
1,..., 1,...,
, , , , ,
, 1
1 1 2 2 3 3 4 5 4 5 6 7 6 7
 
   
     

   
  
    
(3)
where I(i,t) is an indicator variable that equals 1 when month t belongs to year  after
country i finalized its last restructuring ( =1,2,3,4-5,6-7) and zero otherwise, Hi is the
haircut arising from that restructuring, Xi,t-1 is a vector of macroeconomic control variables
known during month t, 
i is a country fixed effect, 
t is a time fixed effect and uit is an
error term. The key parameters of interest are  which measure the partial effect of
haircuts on spreads  years into the future controlling for other simultaneous determinants
of spreads.
Specifically, we control for the debtor country’s level of total external debt to GDP, the
ratio of reserves to imports, the ratio of exports to GDP, the country’s annual rate of
inflation and (real) GDP growth, the level of the current account to GDP and the budget
balance. To capture political risk we include the standard aggregate risk index by ICRG.
These control variables are lagged by one year. International credit market conditions are
controlled for by including the 10 year US Treasury yield rate and the Lehman Brothers
low grade US corporate spread index. Following the received literature we also take into
account credit ratings, by including the residual of a regression of S&P and Moody’s
country credit ratings on the set of other fundamentals and variables in each specification.
The definition and sources of all variables are listed in Table 4 below.

Generically speaking, the previous literature estimates the  coefficients without
controlling for the magnitude of haircuts. Following Asonuma (2010) our contribution is to
allow for the possibility that the effect of restructuring on spreads could be a function of a
continuous measure of investor outcomes, not just of the mere existence of a restructuring.
It is natural to think of that specification in light of the wide variation of investor losses
documented in section 3.
Moreover, in one of the specifications below, we follow Grossman and van Huyck (1988)
in dissecting haircuts into a component that we call excusable (HEX) and another one that
we call inexcusable (HINEX) as estimated from a parsimonious auxiliary model. In that
specification, the estimating equation is:
          
          
i N t T
X u
I i t I i t I i t I i t I i t H
S I i t I i t I i t I i t I i t H
i t i t it
INEX
i
EX
it i
1,..., 1,...,
, , , , ,
, , , , ,
, 1
1 1 2 2 3 3 4 5 4 5 6 7 6 7
1 1 2 2 3 3 4 5 4 5 6 7 6 7
 
   
    
     

   
   
  
    
    
(4)
18
The idea is that, to the extent that sovereign contracts are implicitly contingent on the
realized state of the world, only haircuts that are abnormally high given the exogenous
shocks affecting an economy will be subsequently priced in the market. Hence, in equation
(4) we expect the  -coefficients to be positive while the  ‐coefficients should be zero.
5.3.Results: Haircuts and Subsequent Bond Spreads
Table 5 shows the results, both with and without the set of control variables suggested by
Eichengreen and Mody (2000) and Dell’Arriccia et al. (2006). The key result is that the
lagged values of 􀜪􀯌􀯓 are positive and significant for up to five or even seven years after the
restructuring event.18 This result holds for different model specifications and also when
including year fixed effects. The strictest model to test our hypotheses is that in column (6)
which includes year effects and controls for the potential endogeneity of the timing of
restructuring (e.g. as in countries hurrying to settle with creditors when they anticipate
favorable future borrowing conditions). It indicates that a one percentage point increase in
haircuts is associated with EMBIG spreads that are about 8.8 bp higher in year one after
the restructuring and still about 2.3 bp higher in years four and five. Put differently, a one
standard deviation increase in 􀜪􀯌􀯓 (about 20 percentage points in this sample of emerging
country final deals) is associated with spreads that are 1.7 percentage points higher in year
one and almost one half a percentage point higher in years four and five. Alternatively,
after controlling for a host of fundamentals affecting bond spreads, a jump in H from the
first to the third quartile of its distribution carries an expected increase in spreads of 252 bp
in year one and 65 bp in years four and five after the restructuring. With just two
exceptions, these coefficients are uniformly lower than those in the specifications of
columns (1) through (5) which replicate different specifications from Dell’Arriccia et al.
(2006) and Eichengreen and Moody (2000). Therefore the reported coefficients should be
taken as a lower bound of the estimated effects. Note also that most of the other variables
have the expected sign and are in line with Dell’Arriccia et al. (2006) and others.19

The estimated coefficients are much larger, and longer lived than those presented in
previous work on the impact of country credit history for borrowing costs.20 To make our
results directly comparable to the previous literature, we substitute the lagged values of
􀜪􀯌􀯓 with lagged values of a simple default dummy and obtain coefficients that are nearly
identical to Borenzstein and Panizza (2008) (see column (2) of Table 6). The coefficient
18 H is measured in percentage points throughout.
19 We also find that the coefficient of the U.S. Treasury 10-year yield is significant and negative, which is
surprising. A likely reason for this is that the U.S. 10-year yield is strongly correlated with the high-yield
bond spread (0.75), so that including both variables jointly may bias the estimated coefficient. We therefore
exclude the U.S. Treasury 10-year yield in the robustness analysis.
20 For example, the influential early studies by Lindert and Morton (1989) and Özler (1993) and a new,
rigorous paper by Benczur and Ilut (2009) all find that past default leads to an average increase in post-crisis
spreads of, at most, 50 basis points.
19
for past restructurings is in the range of 360 bp in year one after the restructuring, about
250 bp in year two, but much smaller and mostly insignificant thereafter. Thus, with the
binary dummy, effects appear strong but very short-lived. This stands in contrast to our key
finding, as the size of the haircut seems to matter both in the short and medium run.
To identify the channel behind this main finding we distinguish between “excusable” and
“inexcusable” haircuts. In line with Grossmann and van Huyck (1988), adverse
reputational effects would ensue only when haircuts are excessive or “inexcusable” given
exogenous shocks to the debtor. To our knowledge this refutable proposition has not been
taken to the data - so we next do a first attempt at it. We start with a parsimonious first
stage model in which 􀜪􀯌􀯓 of all restructurings in the sample (see Table 3) are regressed on
time effects (1990-1994, 1995-1999 and 2000-2007), regional effects (Europe and Central
Asia, Far Asia, Middle-East and Africa, and Latin America and Caribbean), a dummy for
Brady deals and the debt to GDP ratio in the year before the restructuring. 21 This
regression decomposes actual HSZ into its “predicted“ value and a residual which we
interpret as measuring the „inexcusable“ haircut.
Column (7) in Table 5 reports the results of replicating the model in column (6) but
dissecting HSZ into these two components as indicated in equation (4) above.22 The left
sub-column at the top reports the estimated coefficients on lagged predicted haircut while
the right sub-column shows the coefficients on the lagged inexcusable haircut. The results
are interesting: in the short run, markets penalize predicted haircuts. However, predicted
haircuts have no long run effects. Long run effects are associated only with inexcusable
haircuts and they are both economically and statistically significant. A one standard
deviation of inexcusable haircut causes ceteris paribus an increase in spreads of 150 basis
points in years four and five after the restructuring and of 207 basis points in years six and
seven after the restructuring. We conclude that, on first examination, the data seem
consistent with the reputational model of Grossman and van Huyck (1988), in which
investors differentiate between types of default.
5.4. Robustness
We next implement several further extensions and robustness checks, based on a
parsimonious specification that only includes explanatory variables that are significant
across the different models and weakly correlated among each other. We first assess
21 The first-stage results are not reported but are available from the authors on request. The specification
resulted from a “general to specific” modeling approach. Other variables considered, but not included in the
final first-stage regression, were natural disasters (share of population killed and affected by natural disasters
as moving sum between years t-4 and t-1), bank crisis (1 if there was a bank crisis between t-4 and t-1, 0
otherwise), terms of trade (cumulative percentage increase between t-4 and t-1), GDP growth (cumulative
percentage increase between t-4 and t-1) and the debt to exports ratio in year t-1. When the individually
significant variables were put alongside each other only debt to GDP remained significant.
22 This second stage regression has fewer observations than those of the previous columns since it is focused
on the sample of defaulting countries (e.g. those in the left-hand side of Table 3). It uses the abnormal
pressure exerted on creditors, conditional on a restructuring having taken place, and given conditions that are
exogenous to the debtor.
20
whether the results are sensitive to alternative definitions of haircuts as defined in section
2. Table 6 shows the results with alternative haircut measures. Column (3) shows results
when including lagged values of the face value reduction measure, column (4) for lagged
values of 􀜪􀯆, and column (5) for lags of the naïve haircut. In addition, column (6) also
shows the effects for lagged values of an “effective haircut” measure, which results from
multiplying HSZ by the fraction of total foreign debt owed to private international creditors
(in t-1) involved in the final deal (with data on debt to private creditors taken from the
GDF database). This last measure thus takes into account the percentage of debt affected
by the haircut. Finally, column (7) shows results when using a “decaying haircut” measure
which weights HSZ with linearly decreasing weights (by 0.1 per year). In this specification,
I in equation (4) is multiplied by (1-0.1(-1)).

Overall, we find that the results are rather stable across different haircut measures.
However, the effects seem to be more pronounced and longer-lived the closer the variables
measure the “true” loss suffered by creditors. To see this compare the results using the face
value reduction measure in column (3) with those relying on effective haircut in column
(6), which accounts for the volume of debt affected. The latter specification shows much
larger and significant coefficients in the medium run.
Further robustness checks are implemented in Table 7. In a first step, we restrict the time
frame to 1998-2006, thus dropping all Brady-era observations of 1993-1997 (column 1).
Next, we focus on the subsample of defaulters, defined here as countries that restructured
sovereign debt at least once after 1985 (those in the left column of Table 3). In both cases
we find the results for our benchmark equation to be robust. The same is true when
excluding two notable outlier countries, Argentina and Russia, which both defaulted
unilaterally, in 1998 and 2001 respectively, and imposed exceptionally high haircuts. As
can be seen in columns (3) and (4) the results are largely robust to these additional checks.
In a last step, we include a dummy variable for ongoing holdout and litigation events using
data from Trebesch (2008). We thereby take into account instances like in Argentina post
2005 or Peru post-1997 in which countries did come to a final restructuring but continued
in disputes with holdout creditors. As can be seen, the dummy variable for litigation is
insignificant and the effects of H are largely unchanged.

21
6. Haircuts and Duration of Exclusion: Data and Results
To assess the role of haircuts for exclusion duration we construct an annual dataset on
access to capital for the period of 1980 to 2006. The decision to use yearly data is in line
with related research and driven by data availability, as our duration analysis goes further
back in time and spans a larger number of defaulting countries, many of which do not have
reliable data at monthly frequency. We again focus on access conditions after “final”
restructurings, i.e. those not followed by another restructuring in the next four years. The
sample includes all 16 Brady deals and the set of recent sovereign bond restructurings, but
excludes the exceptional cases of Yugoslavia and Cuba due to missing data on explanatory
variables (Table 8 provides an overview on all 66 cases included).
6.1. Dependent Variable: Years of Exclusion
The dependent variable on exclusion duration is the number of years between a
restructuring event and the first year of reaccess to capital markets.23 In defining market
access, we build on Gelos et al. (2004) and Richmond and Dias (2009), the two key
empirical contributions on this issue in recent years. Gelos et al. define access as issuance
of a public or publicly guaranteed bond or syndicated loan on international markets that
leads to an increase in public indebtedness to private external creditors.24 Richmond and
Dias, in contrast, propose a broader definition based on aggregate capital flow data. They
define reaccess as the first year after settlement with a net debt transfer from private
foreign creditors to the public and/or private sector of the debtor country. Our aim here is
not to develop a novel measure of reaccess, nor do we plan to embark into lengthy
discussions on the benefits and drawbacks of alternative measures. By contrast, our
purpose is to analyze the findings of the received literature through the lenses of a more
precise definition of investor losses. As a baseline, we therefore construct a measure that is
as general as possible, simply by combining the definitions of Gelos et al. and Richmond
and Dias. Later on, we test the robustness of our results to alternative measures, including
the original ones proposed by the authors.
The main measure used here captures “partial” reaccess, defined as the first year with
primary market issuance and/or positive credit flows to the public sector. The measure
takes a value of one in case the government and/or public or publicly guaranteed
enterprises (i) places at least one syndicated loan or bond in international markets that
results in an increase in indebtedness or (ii) if the public sector receives net transfers from
private foreign creditors so that new borrowing minus debt service is positive. The first
criterion follows Gelos et al. and is measured using issuance data on individual syndicated
23 If a country restructures and regains market access in the same year, we consider the duration of market
exclusion to be one year.
24 This second criterion aims to exclude cases in which sovereigns issue debt only to roll over expiring
maturities.
22
loans and bonds provided by Dealogic. More specifically, we aggregate information of
3,688 public and publicly guaranteed bonds in 78 countries and 10,013 syndicated loans
that are public or publicly guaranteed from 143 countries between 1980 and mid 2008. In
line with Gelos et al. we only regard issuances that lead to an increase in the public sector
debt stock. Data on the stock of public sector debt owed to private creditors is taken from
the GDF dataset. The second criterion follows Richmond and Dias and is constructed from
aggregate data on bank and bond flows. The dummy takes a value of one in case bank or
bond transfers from foreign private creditors to the public and publicly guaranteed sector
are larger than 0. To check the robustness of our finding we also construct (i) measures of
“full reaccess” defined as the first year in which debt flows surpass 1% of GDP, (ii) a
measure that focuses on primary market issuance only (the original Gelos et al. definition),
and (iii) a measure that takes into account flows to the public and private sector of debtor
countries (the Richmond and Dias definition).
6.2. Preliminary Data Analysis
Next, we present descriptive findings on haircut size and the duration of exclusion. Table 8
lists restructuring events and the year of reaccess using various definitions. The average
duration from restructuring to partial reaccess is 5.5 years, while it takes an average of 8.2
years until full reaccess. We find that the duration of exclusion tends to increase with
haircut size. The average duration until partial reaccess is 2.9 years following
restructurings with haircuts lower than 30%, but 6.2 years for deals with H􀭗􀭞>30. Figure 4
plots the relationship between H􀭗􀭞 (in %) and years until partial reaccess for the full
sample, further pointing to a positive relationship between the two. The overall picture is
similar when using time until full reaccess, i.e. until loan or bond issuance or net inflows
surpass 1% of GDP.

Another way to illustrate the patterns of exclusion is to plot an empirical survival function.
We apply the non-parametric Kaplan-Meier estimator, which estimates an unconditional
survival function and is very popular in the survival analysis literature. This statistic
reports the compound probability of not having reaccessed the market for each year after
the restructuring. It can be defined as
􀜵 􁈘􁈺􀝐􁈻 􀵌 􀷑 􁉆
􀝊􀯝􀬿􀝀􀯝
􀝊􀯝
􁉇
􀯝|􀯧􀳕􀮸􀯧
􁈺5􁈻
where 􀝐􀯝, 􀝆 􀵌 1, …, denotes the times at which failure occurs, 􀝀􀯝 are the number of failures,
or “exits” at time 􀝐􀯝 and 􀝊􀯝 is the total number at risk of failure at time 􀝐􀯝 (see Kalbfleisch
and Prentice 2002). Here, the number of failures 􀝀􀯝 is simply the sum of countries that
23
successfully reaccess capital markets in a given year, while 􀝊􀯝 counts the number of
country cases that were excluded at 􀝐􀯝􀬿􀬵.

Figure 5 shows the result for the estimated survival function for partial reaccess. Unlike
previous research, we estimate survival functions depending on haircut size of the
restructuring. More specifically, we group cases with H􀭗􀭞60% and those
in-between. The graph shows that the estimated functions are markedly different for cases
with higher haircuts. Only 10% of countries with H􀭗􀭞>60% regain partial access within
two years, compared to 50% for cases with H􀭗􀭞 smaller than 30%. The figure also shows
that exceptionally high haircuts are often followed by exceptionally long periods of
exclusion. More than 70% of countries imposing H􀭗􀭞 􀵐 60 did not regain partial access
after 15 years.
6.3. Estimated Model on Exclusion Duration
In a second step, we estimate a semi-parametric Cox proportional hazard model that
includes our haircut measure as the key explanatory variable. The advantage of the Cox
model is that it allows including time-varying covariates and that it can deal with the
problems of censored observations and multiple events. For this model, the hazard rate for
the ith individual (or ith exclusion episode) can be written as
( ) ( ) exp( ), 0 h t h t z i    (6)
where ( ) 0 h t is the baseline hazard function, z a set of covariates and  a vector of
regression coefficients.
The key advantage of the Cox model vis-à-vis other duration models such as the
parametric Weibull model or the log logistic model, is that it is not necessary to specify a
functional form of the baseline hazard rate ( ) 0 h t . Instead, the shape of ( ) 0 h t is assumed to
be unknown and is left unparameterized. Accordingly, we estimate reduced form models
allowing the functional form of the hazard function to be explained by the data. The model
is estimated via a partial likelihood function of the following form:
,
exp( )
( ) exp( )
1 ( )
i
i
n
i j R t j
i
z
L z



    
 


 




 (7)
where ( ) ( : ) i j i R t  j t  t denotes the risk set (i.e. the number of cases that are at risk of
failure) at time i t . The model can be extended in a simple manner once time varying
covariates are included (see Lancaster 1990 for a detailed presentation).
24
In estimating the model we rely on the variance correction method proposed by Lin and
Wei (1989).25 This avoids misleading inference in the case of repeated events and is
relevant because some countries in our dataset had multiple restructurings and reaccess
episodes since 1980. Thereby potential learning effects are also taken into account.
In the choice of control variables we again build on previous literature, in particular on
Dell’Arriccia et al. (2006), Gelos et al. (2004) and Richmond and Dias (2009). One
difference compared to the above is that we now use country ratings by Institutional
Investor magazine instead of commercial rating agency ratings, simply because we cover a
much larger sample of countries and years than in the monthly EMBIG dataset. We also
include dummy variables for world regions following the World Bank classification as
well as period fixed effects (by decade).26
6.4. Estimation Results: Haircuts and the Duration of Market Exclusion
Table 9 shows the results for various specifications of the Cox proportional hazard model.
A positively signed coefficient means that higher values of a covariate increase the hazard
rate, i.e. the likelihood of failure in a given period. Here, a positive coefficient indicates
that higher values of that variable are associated with quicker reaccess relative to the
baseline, while negative coefficient indicates longer exclusion duration.

The main result of Table 9 is that the coefficient of 􀜪􀯌􀯓 is positive and robustly significant
in all specifications. It also has a sizable quantitative effect. To illustrate this and to allow
for a more intuitive interpretation, it is necessary to exponentiate the coefficients shown in
the table. The coefficient of 0.0304 in column (7) indicates that a one unit (percentage
point) increase in H􀭗􀭞 lowers the likelihood of reaccessing capital markets in a given year
by 3%.27 Accordingly, a 30 percentage point increase in 􀜪􀯌􀯓 decreases the likelihood of
reaccess by 60% in any given year, and after controlling for country fundamentals.28 This
indicates that restructuring outcomes play a crucial role for the speed of reaccess after
settlement.
Regarding the other variables included, we can report only few significant coefficients. In
line with Richmond and Dias (2009), we find population size to be positively associated
with the speed of reaccess, although this finding is not very robust to specification changes.
Higher debt to GNI levels and higher annual inflation have significant negative
coefficients, and are thus associated with longer exclusion duration. We also find country
25 For a survey on variance-correction methods for repeated events in survival analysis see e.g. Kelly and Lim
(2000).
26 Note that the proportional hazard survival models produce biased estimates with country fixed effects
(Allison 2002, Andersen et al. 1999).
27 The calculation is 100*(􀝁􀬿.􀬴􀬷􀬴􀬸-1) = -3
28 The calculation is 100*(􀝁􁈾􀬷􀬴∗􀬿.􀬴􀬷􀬴􀬸􁈿 -1) = -59.9
25
credit ratings to matter, with more favorable ratings increasing the likelihood of reaccess.
All other variables included turn out not to be significant. Like before we therefore strike a
balance between parsimony and performance of the model, and settle on a baseline
specification with a reduced set of covariates. Other than regional and decade fixed effects
this baseline specification includes the high-yield bond spread, debt to GNI, GDP per
capita, the growth rate, reserves to imports and the rating residual. Both the log-likelihood
statistics and the B.I.C. criterion indicate a superior model fit for this specification.
6.5.Robustness
As an important robustness check, we assess to what extent our main results depend on the
definition of market access. Table 10 reports results for three alternative dependent
variables. Columns (1) and (2) show results for duration until “full reaccess” defined above
as the first year with issuance volumes or net transfers to the public sector exceeding 1% of
GDP. The dummy for “full reaccess” uses these same data sources, but imposes a higher
threshold. It is coded as one (i) if bond or loan issuances in international markets exceed
1% of GDP or (ii) if net bank and bond transfers to the public sector exceed 1% to GDP.
The 1% threshold is chosen in accordance with Richmond and Dias and represents less
than one-half of the annual public sector borrowing requirement over the entire sample of
years and developing countries.29

Columns (3) and (4) extend the definition of access to include net transfers to the private
sector as well, following Richmond and Dias (2009). This broader definition translates into
significantly shorter periods of exclusion, as illustrated in Table 8.30 We next follow the
original definition of Gelos et al. (2004) in columns (5) and (6). Under this specification,
market access is measured by primary market loan or bond issuance only, thus
disregarding overall net transfers. When comparing the results for these very different
market access definitions, we find our main result to be surprisingly robust. H􀭗􀭞 remains
significant with a sizable coefficient of between -0.02 and -0.04. This gives strong support
that haircut size may play a crucial role for country access to capital markets postrestructuring.

29 GDP data is taken from the World Development Indicator dataset. The annual volume of loan and bond
placements is again aggregated from individual issuance data from Dealogic, while net transfers are again
taken from the GDF dataset.
30 Take the example of Chile, which restructured in 1990. According to our baseline definition, which focuses
on public sector borrowing only, the country regained partial access in 1994 and full access in 1998. Yet,
once debt flows to the private sector are taken into account partial access is gained as early as 1991, i.e. after
only one instead of 4 years.
26
To further assess the validity of our results, we restrict the sample to middle income
countries only (see columns 7 and 8 in Table 10). Specifically, we exclude all episodes
linked to donor funded debt restructurings, which usually take place in highly indebted
poor countries (HIPCs). The reason why this robustness check is important is that donor
funded debt restructurings tend to be of a different nature. First, haircuts in donor funded
deals tend to be significantly higher compared to all other restructurings. Additionally,
most HIPCs have no or only limited access to international capital markets, a fact that may
distort the estimated coefficients if these countries are included. As can be seen, however,
we find the results to be robust in the subsample of non-HIPC countries. The same is true
when restricting the sample to the group of emerging market countries included in our
analysis on EMBIG spreads above (results available upon request). Finally, we also show
results when using other haircut measures. Table 11 shows that the results are only little
affected when using the “market haircut” of eq. (2), a measure for plain face value
reduction, the set of “naïve haircuts” using a uniform discount rate, or the “effective
haircut” accounting for the share of outstanding debt restructured.
7. Conclusion
This paper computes haircuts implicit in debt restructurings between sovereigns and
private international creditors during 1970-2007. We use the variation in haircut size as a
novel measure of credit history and investigate the link between haircuts and subsequent
borrowing conditions of defaulting countries. Our main finding is that higher haircuts are
associated with significantly higher post-restructuring spreads and much longer periods of
market exclusion. This indicates punishment effects in credit markets, as predicted by
theory. Interestingly, however, the finding stands in contrast to much of the existing
literature, which finds only small or short-lived effects of defaults and restructurings on
access conditions.
Overall, we regard the analysis in this paper as a first step to reassess post-default
borrowing conditions of sovereigns. In particular, we see the need for further analysis on
the mechanism behind our finding.
27
Figure 1: Haircuts and Deal Volume over Time
Note: The Figure plots the size of haircuts in % (􀜪􀭗􀭞 from eq. 2) across countries and time.
The circle size reflects the volume of debt restructured (in current USD)
Figure 2: Restructurings with and without Debt Reduction
Note: The Figure plots the size of haircuts in % (H􀭗􀭞 from eq. 2) across countries and time.
ALB
DZA
DZA
ARG
ARG
ARG
ARG
BLZ
BOL
BOL
BOL
BIH
BRA
BRA
BRA
BRA
BRA
BRA
BGR
CMRCMR
CHL
CHL
CHL
CHL
CHL
COG
CRI
CRI
CRI
CIV
CRO
CUB
CUBCUB
DOM
DOM
DOM
DOM
ECU
ECU
ECU
ECU
ECU
ETH
GAB
GAB
GMB
GRD
GIN
GIN
GUY
GUY
HND
HND
IRQ
JAM
JAM
JAM
JAM
JAM
JAM
JAM
JOR
KEN
LBR MKD
MDG
MDG
MDG
MDG
MWI
MWI
MRT
MEX
MEX
MEX
MEX
MEX
MEX
MDA
MDA
MAR
MAR
MAR
MOZ
NIC MOZ
NIC
NIC
NIC
NIC
NER
NER
NER
NGA
NNGGAA
NGA
NGA
NGA
NGA
PAK
PAN PAK
PAN
PAN
PRY
PER
PER
PER
PER
PHL
PHL
PHL
PHL
POL
POL
POL
POL
POL
POL
POL
POL
ROM
ROM
ROM
RUS
RUS
RUS
RUS
STP
SEN
SEN
SEN
SEN
YUG
SLE
SVN
ZAF
ZAF
ZAF
SDN
TZA
TGO
TGO
TTO
TUR
TUR TUR
TUR
UGA
UKR
UKR
URY
URY
URY
URY
URY
VEN
VEN
VNM
YEM
YUG
YUG
YUG
YUG
ZAR
ZARZAR
ZAR
ZAR
ZAR
ZAR
ZMB
0 20 40 60 80 100
Haircut (preferred) in %
1975m1 1980m1 1985m1 1990m1 1995m1 2000m1 2005m1
SDN
PAN
CHL
ZAR
YUG
NIC
BRA
MDG TTO
VEN
ZAF POL
NIC
JAM ECU
NER
DOM
NER
TUR ECU
GMB
JAM
JAM
PER
MWI
CHL
NGA
MOZ
ROM
CUB
ZAR
PER
NGA
PHL
NIC ZAR
CUB
YUG
URY
TUR
ZAR MDG
POL
MEX
JAM
PER
ZAR
YUG
NGA
GAB
TUR
LBR
BRA
POL
URY
ZAR
YUG
POL
CUB
ECU
TGO
BRA
BOL
ROM
ZAR
POLARG MAR
CHL ZAF
ROM JAM
MDG
NBGRAA
CRI
POL
HND
JAM
PHL
POL
CHL
MWI
MEX
GIN
MEX
SEN
CRI
TUR ARG
NIC
MEX
MAR
SEN
URY
NGA
NGA
CHL
BRA
DZA
MKD
MAR
ZAF
PAN
SEN
MDG
SVN
RUS
SEN
CRO
GAB
DZA
JAM
PRY
UKR URY
MDA
PAK
RUS
DOM
BLZ
DOM
PAK
GRD
BOL
PHL
MEX
SLE
MEX
TGO
VEN
JOR
ZMB
CRI
CIV
ALB
STP
NGA
MOZ
ARG
BGR VNM
NER
DOM
GUY
BRA
BOL
NIC
PER
MRT
PHL
PAN
URY
ECU
POL
UGA
ETH BIH GUY
MDA
IRQ
YUG
GIN CMR
ARG
ECU
CTMZRA
KEN
UKR
HND
RUS
YEM
COG
RUS
0 20 40 60 80 100
Haircut (preferred) in %
1980m1 1985m1 1990m1 1995m1 2000m1 2005m1
Rescheduling Only With Face Value Reduction
28
Figure 3: Post-Restructurings Spreads (average, by haircut size)
Note: The figure plots average post-restructuring EMBIG spreads across restructuring
episodes with H􀭗􀭞 􀵏 30% and those with H􀭗􀭞 􀵐 30%, respectively. Years with ongoing default
as of S&P (e.g. Argentina 2001-2005) are excluded. The country cases with H􀭗􀭞 􀵏 30% are Algeria,
Chile, Dom.Rep.(post-2005), Pakistan, Phillipines, South Africa, Ukraine and Uruguay (post-1991/post-
2003). The country cases with H􀭗􀭞 􀵐 30% are Argentina (post-1993 and post-2005), Brazil, Bulgaria,
Cote d'Ivoire, Dom.Rep. (post-1994), Ecuador (post-1995/ post-2000), Mexico, Morocco, Nigeria,
Panama, Peru, Poland, Russia,Venezuela,Vietnam,Serbia and Montenegro. The Figure also shows the
average EMBIG spread 1993-2006 for countries that did not restructure (non-defaulters).
Figure 4: Duration of Exclusion and Haircut Size
Note: The figure plots the relationship between H􀭗􀭞 and years until reaccess to
capital markets after the related restructuring. The time period covered is 1980-
2006. Reaccess here is defined as the first of the following two events: (i)
issuance of a syndicated loan or bond on international markets OR (ii) a
positive net transfer of bond or bank credit to the public sector.
0
200
400
600
800
1000
12 24 36 48 60 72 84 96 108 120
Average EMBIG Spread (in bp)
Haircut >30%
Haircut 60%
0.00 0.25 0.50 0.75 1.00
0 5 10 15
Years of Exclusion (partial)
eaccess
30
Table 1: Haircut Estimates by Type of Restructuring and Era
Note: The Table shows summary statistics for different estimates and subsamples.
The “type of estimates” refers to different haircut computation formula (section 3.1.).
All other estimates in the table are based on the “preferred” haircuts (􀜪􀭗􀭞 from eq. 2).
“HIPC or Donor Funded” restructurings are those implemented in the poorest and
highly indebted countries supported by the IDA debt reduction facility (World Bank 2007).
Obs. Mean Std. Dev. Min Max
Market Haircut (eq. 1) 178 39.17 26.99 -8.60 97.00
SZ Haircut ("preferred", eq. 2) 177 36.05 27.22 -8.60 97.00
Naive Haircut (DR 10%) 177 19.61 34.99 -47.70 97.00
Write Off (of Face Value) 180 15.49 29.75 0.00 97.00
Pre-Brady (1970-1989) 100 24.69 18.04 -6.40 92.70
Brady Era (1990-1997) 48 52.08 28.24 3.60 92.30
Post-Brady (since 1998) 29 48.68 33.20 -8.60 97.00
HIPC or Donor-Funded 23 87.02 7.00 62.60 97.00
All Other Restr. 154 28.44 19.92 -8.60 92.70
Bank Debt Restructuring 163 36.52 27.75 -8.60 97.00
Bond Debt Restructuring 14 30.64 20.09 5.20 76.80
Rescheduling vs. Debt Reduction
Rescheduling Only 125 23.99 16.83 -6.40 92.00
With Reduction in Face Value 52 65.05 25.59 -8.60 97.00
By Type of Estimate
By Era
By Type of Debtor
By Type of Creditor
31
Table 2: Haircuts in Selected Recent Restructurings (1999-2007)
Note: The average haircuts by Sturzenegger and Zettelmeyer (2006, 2007, 2008) and those by the Bank of Spain and Bank of England staff (Benford et al 2005, Diaz-Cassou et al. 2008a,b)
are computed in present value terms using country-specific discount rates and can thus be compared to HSZ from equation (2). Finger and Mecagni (2007) mostly use a 10% discount rate,
while Benjamin and Wright’s (2009) estimates are based on nominal interest and principal forgiven, so that the results are not directly comparable.
Haircuts: Authors Estimates
Case Date of
Exchange
Anouncement
Default Debt
exchanged
(in m USD)
Particip
ation
Rate
SZ type
Haircut
(Preferred)
Market
Haircut
(eq. 1)
Underlying
Discount
Rate
Naïve
Haircut
(10% DR)
Face Value
Reduction
(in %)
SZ
average
haircut
SZ
Discount
Rate
SZ
haircut
10% DR
Benjamin
& Wright
(2009)
Finger &
Mecagni
(2007)
Bedford
et al.
(2005)
Diaz-Cassou
et al
(2008a,b)
Pakistan
(Bank debt) July 1999 Aug. 1998 preemptive 777 n.a. 11.2 11.2 0.130 6.6 0.0
Pakistan
(External bonds) Dec. 1999 Aug. 1999 preemptive 610 99% 13.9 13.2 0.142 1.8 0.0 31 0.214 0.3 29 9-27 35 30
Ukraine
(External bonds) April 2000 Dec. 1999 preemptive 420 97% 17.5 16.1 0.140 8.0 0.9 28.9 0.286 2.2 1 5 40 32
Ecuador (External
bonds) Aug. 2000 July 1998 Aug. 1999 6,700 98% 37.7 58.7 0.167 24.4 33.9 28.6 0.222 21 34 25 40 26
Russia
(London Club) Aug. 2000 Sept. 1998 Dec. 1998 31,943 99% 53.3 67.7 0.145 46.3 36.4 52.6 0.164 48.2 32 44 50 48
Moldova (External
bonds) Oct. 2002 June 2002 preemptive 40 100% 35.5 35.5 0.188 7.0 0.0 33.5 0.210 42 0-6
Uruguay
(External bonds) May 2003 March 2003 preemptive 3,127 90% 11.0 13.6 0.097 11.3 0.0 12.9 0.122 7.8 8-20 15 14
Moldova
(Gazprom debt) April 2004 Sept. 2002 mid 2001 115 n.a. 56.3 56.3 0.101 56.3 56.3 58
Serbia & Montenegro
July 2004 Dec. 2000 since 1990s 2700 n.a. 71.2 73.5 0.098 71.1 59.3 57 62
Dominica
(bonds and loans) Sept. 2004 June 2003 July 2003 144 72% n.a. 54.9 0.094 57.3 15.0
Argentina
(External bonds) April 2005 Oct. 2001 Jan. 2002 43,736 76% 76.8 78.8 0.104 76.0 29.4 75 0.082 77.8 63 75 70 73
Dominican Rep.
(External bonds) May 2005 Apr. 2004 preemptive 1,100 94% 5.2 5.2 0.097 5.9 0.0 1.5 0.096 1.6 1 5 1
Dominican Rep.
(Bank debt) Oct. 2005 Apr. 2004 Febr. 2005 180 n.a. 11.7 16.3 0.099 12.3 0.0 2
Grenada
(bonds and loans) Nov. 2005 Oct. 2004 Dec. 2004 210 91-97% 34.2 41.6 0.098 34.4 0.0
Iraq
(bank/commercial) Jan. 2006 in 2004 n.a. 17,710 96% 88.9 88.9 0.117 87.4 81.5
Belize
(bonds and loans) Febr. 2007 Aug. 2006 Sept. 2006 516 97% 24.6 30.5 0.098 25.5 0.0 28
Restructuring Details Prior Estimates
32
Table 3: Countries included in EMBIG Analysis
Countries with Sovereign
Restructuring since 1990
(n=24), with agreement
dates
Countries without
Restructuring since
1990 (n=12)
Algeria (07/1996) Colombia
Argentina (04/1993, 04/2005) China
Brazil (04/1994) Egypt
Bulgaria (06/1994) El Salvador
Chile (12/1990) Hungary
Cote d'Ivoire (03/1998) Indonesia
Croatia (07/1996) Lebanon
Dominican Rep.(08/1994, 05/2005) Malaysia
Ecuador (02/1995, 08/2000) South Korea
Mexico (05/1990) Thailand
Morocco (09/1990) Tunisia
Nigeria (12/1991) Turkey
Pakistan (12/1999)
Panama (05/1996)
Peru (03/1997)
Philippines (12/1992)
Poland (10/1994)
Russia (08/2000)
South Africa (09/1993)
Ukraine (04/2000)
Uruguay (05/2003)
Venezuela (12/1990)
Vietnam (12/1997)
Serbia & Montenegro (07/2004)
33
Table 4: Description of Data and Variables used in Estimations
Variable Description Frequency Source
Main Dependent Variables
EMBIG Stripped Spread Monthly average EMBIG spread Monthly JP Morgan/Datastream
Reaccess (partial reaccess)
Dummy capturing the first of the
following two events: (i) foreign
syndicated loan or bond issuance
(public or publicly gurant.), or (ii) net
transfer from private foreign creditors
to public sector
Yearly
Dealogic (primary market data);
GDF (series DT.NTR.PNGB.CD
and DT.NTR.PNGC.CD)
Main Haircut Measures
Haircut (M)
Market haircut (comparing par value
of old with present value of new debt,
see eq. 1)
Monthly/Yearly Own Calculations
Haircut (SZ)
Haircuts computed in analogy to
Sturzenegger and Zettelmyer
(comparing present value of old and
new debt, see eq. 2)
Monthly/Yearly Own Calculations
Control Variables
High-yield bond spread Lehman Brothers US Corporate
High Yield spread Monthly/Yearly Lehman Brothers/Bloomberg
US 10-year Treasury Yield Yield on 10-year US Treasury bonds Monthly/Yearly Datastream
Political Risk (ICRG) Political Risk Index (lagged) Monthly/Yearly ICRG (Political Risk Group)
Rating Rating average of available ratings or
only available rating.
Monthly
(S&P, Moody's),
Yearly (IIR)
S&P, Moody's (in EMBI analysis), and
Institutional Investor Magazine
(in yearly analysis / duration )
Credit Rating Residual Residual from regression of ratings on
fundamentals (cf. ratings) Monthly/Yearly Own calculations,
based on ratings data
External Debt / GNI (in %) Total external debt to GNI (in %,
lagged) Yearly WDI
GDP real growth (in %) GDP real growth (yoy in %, lagged) Yearly WDI
Current Account to GDP (in %) Current account to GDP, four-year
moving average (in%, lagged) Yearly WDI
Budget Balance to GDP (in %) Central government Fiscal Balance to
GDP (in %, lagged) Yearly Economist Intelligence Unit
Reserves to Imports (in %) Reserves (incl. gold) to Imports
(in %, lagged) Yearly WDI
Exports to GDP (in %) Exports to GDP (in%, lagged) Yearly WDI
Inflation (in %) Consumer price inflation (yoy in %,
lagged) Yearly WDI
GDP per capita (PPP, log) log of per capita GDP in purchasing
power parity Yearly WDI
Population (log) log of population size Yearly WDI
Note: GDF stands for Globald Development Finance, WDI stands for the World Development Indicators (both World Bank databases).
34
Table 5: Baseline Results for Haircuts and Bond Spreads
Note: The table shows coefficients of a fixed effects panel data regression with robust, country-clustered standard
errors. The dependent variable is the monthly average country spread to US treasury bonds (EMBIG stripped
spread), while the key explanatory variables are lagged values of HSZ up to 7 years after each final restructuring.
Column 7 contains residuals of a first-stage regression of H on fundamentals, thus dissecting HSZ into a predicted
component and a residual or “inexcusable” part. See equation 4 and section 5.3.
Fixed
Effects, No
Controls
Eichengreen -
Mody
Dell'Arriccia
et al. spec.
Dell'Arriccia
et al. spec.
Dell'Arriccia
et al. spec.
With Year
Fixed Effects
"Predicted"
Haircut
"Inexcusable"
Haircut
(1) (2) (3) (4) (5) (6)
10.32** 11.90*** 10.29*** 10.73*** 10.76*** 8.84*** 6.14** -7.71
(4.45) (1.85) (1.75) (1.93) (1.78) (2.78) (2.49) (9.50)
6.93** 9.07*** 8.69*** 8.86*** 8.97*** 7.79*** 4.96* 0.40
(2.90) (2.16) (2.23) (2.23) (2.03) (2.63) (2.70) (6.08)
6.08** 5.95*** 5.32** 5.27** 5.14*** 4.22* 1.56 9.00
(2.47) (2.03) (2.11) (2.07) (1.77) (2.33) (2.86) (7.49)
4.64*** 3.99*** 3.06** 3.08** 3.05*** 2.30* 0.67 9.45*
(1.66) (1.11) (1.28) (1.22) (1.07) (1.37) (1.94) (5.60)
2.55 2.54** 2.23 2.17 2.29* 1.70 1.00 13.05**
(1.97) (1.23) (1.41) (1.41) (1.23) (1.36) (1.67) (5.11)
-74.54*** -99.78*** -97.90*** -82.70*** -74.54***
(17.90) (20.57) (20.23) (20.35) (19.50)
-10.18*** -7.59* -7.10* -9.66** -10.44**
(3.69) (4.14) (4.08) (4.06) (4.44)
2.58* 2.70 3.86** 3.05* 2.75
(1.33) (1.94) (1.87) (1.60) (1.73)
-9.22* -9.25* -7.07* -8.67
(4.74) (4.74) (4.25) (5.31)
-11.76** -10.95** -5.64 -4.94
(4.93) (5.21) (5.82) (5.81)
1.25 5.89 4.82
(3.69) (3.68) (4.15)
-2.77 -2.36
(2.36) (2.71)
-5.67*** -5.03***
(1.51) (1.62)
0.03 -0.00
(0.06) (0.07)
-58.12*** -51.29*** -58.05*** -70.62*** -74.16***
(18.01) (16.20) (15.02) (15.40) (17.10)
46.70*** 44.93*** 47.83*** 44.69*** 61.17***
(9.56) (9.31) (9.47) (8.94) (10.46)
381.43*** 91.29 16.67 -124.45 58.32 37.44
(26.25) (122.61) (130.98) (281.60) (319.15) (356.78)
Country Fixed Eff. Yes Yes Yes Yes Yes Yes
Year Fixed Eff. No No No No No Yes
Observations 3,384 2,714 2,557 2,533 2,533 2,533
R -squared 0.092 0.361 0.416 0.436 0.468 0.517
(1.43)
(6.61)
(12.09)
High-yield bond
spread
Constant
Inflation (annual,
in %)
GDP Real Growth
(in %)
External Debt to
GNI (in %)
Budget Balance to
GDP (in %)
Current Account to
GDP (in %)
Credit Rating
(Residual)
Exports to GDP
(in %)
Reserves to
Imports (in %)
Political Risk
(ICRG)
US 10-year
Treasury Yield
Haircut (SZ),
1 year lag
Haircut (SZ),
2 year lag
Haircut (SZ),
3 year lag
Haircut (SZ),
4 & 5 year lag
Haircut (SZ),
6 & 7 year lag
-106.21***
(17.12)
-7.69**
(3.11)
2.53*
410.64**
(18.17)
(3.52)
(2.18)
(0.06)
Yes
Yes
1,733
0.597
(7)
(22.09)
(19.04)
(189.80)
-10.12
-28.31**
8.43
-4.82
-7.92***
-0.04
-44.49**
56.54***
35
Table 6: Extended Results for Bond Spread Estimation – Other Haircut Measures
Note: The table shows coefficients of a fixed effects panel data regression with robust, country-clustered standard
errors. The dependent variable is the monthly average country spread to US treasury bonds (EMBIG stripped
spread), while the key explanatory variables are lagged values of various haircut measures up to 7 years after each
final restructuring.
The “Preferred Haircut” refers to HSZ, the “Market Haircut” refers to HM , “Naïve Haircut” is the same as HM, but
applies a uniform 10% discount rate and no forward interest rates, “Effective Haircut” takes into account the
volume of debt affected by the restructuring and results from multiplying HSZ by the fraction of total foreign debt
owed to private international creditors (in t-1) that is involved in the exchange. Column 7 shows results when using
a “decaying haircut” measure, which weights HSZ with linearly decreasing weights (by 0.1 per year).
Benchmark
Preferred
Haircut
Restructuring
Dummy
Face Value
Reduction
Market
Haircut
Naive
Haircut
Effective
Haircut
Decaying
Haircut
(1) (2) (3) (4) (5) (6) (7)
8.52*** 364.02*** 12.83*** 6.91*** 7.88*** 8.56*** 8.54***
(1.61) (98.62) (2.45) (1.17) (1.36) (1.64) (1.59)
6.33*** 252.39*** 10.21*** 5.09*** 6.05*** 6.96*** 7.06***
(1.75) (76.60) (3.90) (1.48) (1.80) (1.95) (1.92)
3.49** 123.21* 6.39* 3.01** 3.64** 4.13** 4.39**
(1.62) (64.60) (3.62) (1.28) (1.76) (1.83) (1.99)
2.76** 77.30 0.05** 2.36** 2.75** 3.45*** 4.22**
(1.09) (49.33) (0.02) (1.01) (1.08) (1.06) (1.65)
1.97 47.77 0.03 1.54 2.11* 2.69** 4.59*
(1.21) (54.67) (0.02) (1.09) (1.20) (1.27) (2.69)
-59.27*** -58.52*** -57.92*** -60.56*** -59.56*** -60.75*** -59.25***
(16.29) (17.31) (17.89) (16.31) (16.84) (16.49) (16.29)
2.96** 3.09** 3.06** 3.04** 3.03** 2.91** 2.97**
(1.41) (1.53) (1.53) (1.40) (1.45) (1.45) (1.40)
-5.92*** -5.74*** -5.70*** -6.01*** -5.92*** -6.02*** -5.90***
(1.89) (1.82) (1.91) (1.91) (1.91) (1.91) (1.88)
36.33*** 37.61*** 36.57*** 36.53*** 36.56*** 36.12*** 36.32***
(8.53) (8.82) (8.31) (8.57) (8.54) (8.50) (8.51)
95.13 72.09 91.94 92.63 93.93 105.81 93.69
(142.40) (137.00) (149.31) (143.26) (145.09) (145.85) (141.59)
Ctry Fixed Eff. Yes Yes Yes Yes Yes Yes Yes
Observations 2,658 2,658 2,658 2,658 2,658 2,658 2,658
R -squared 0.381 0.384 0.371 0.378 0.375 0.377 0.382
4 & 5 year lag
6 & 7 year lag
"Plain" Measures Other Haircut Measures
1 year lag
2 year lag
3 year lag
High-yield bond
spread
Constant
Credit Rating
(Residual)
External Debt
to GNI (in %)
Reserves to
Imports (in %)
36
Table 7: Robustness Checks for Bond Spread Estimation
Note: The table shows coefficients of a fixed effects panel data regression with
robust, country-clustered standard errors. The dependent variable is the monthly
average country spread to US treasury bonds (EMBIG stripped spread), while the
key explanatory variables are lagged values of HSZ up to 7 years after each final
restructuring.
From 1998
on only
Subsample of
Defaulters
Excluding
Argentina,
Russia
Controlling for
Litigation
(1) (2) (3) (5)
9.61*** 8.09*** 9.00*** 7.93***
(2.90) (1.54) (2.22) (1.93)
6.26*** 6.19*** 6.48*** 6.01***
(1.87) (1.75) (1.91) (2.00)
2.62** 3.32** 3.10* 3.26*
(1.22) (1.50) (1.65) (1.83)
3.08*** 2.66** 2.80*** 2.69**
(1.00) (1.17) (1.06) (1.11)
1.21 1.43 1.88 2.00*
(1.03) (1.27) (1.27) (1.20)
-66.88*** -71.99*** -59.32*** -59.57***
(21.95) (18.67) (16.83) (16.36)
1.84 3.79** 3.35** 2.99**
(1.52) (1.90) (1.48) (1.43)
-7.18*** -6.87*** -6.14*** -6.02***
(1.89) (1.55) (2.02) (1.93)
34.27*** 45.09*** 36.76*** 36.08***
(8.94) (10.40) (8.98) (8.49)
55.59
(71.29)
221.67 -2.16 76.74 100.73
(158.13) (166.33) (151.03) (143.66)
Ctry Fixed Eff. Yes Yes Yes Yes
Observations 2,203 1,919 2,514 2,658
R -squared 0.412 0.431 0.380 0.384
Haircut (SZ),
1 year lag
Haircut (SZ),
2 year lag
Haircut (SZ),
3 year lag
Haircut (SZ),
4&5 year lag
Haircut (SZ),
6&7 year lag
External Debt
to GNI (in %)
Reserves to
Imports (in %)
Litigation
Constant
Credit Rating
(Residual)
High-yield bond
spread
37
Table 8: Overview on Restructuring Cases and Reaccess Years (duration analysis)
Country Restructuring
Date
Robustess Check
(Flows to Public
OR Private)
Partial Reaccess
(Flows > 0)
Full Reaccess
(Flows > 1% of
GDP)
Partial ( > 0), but
incl. flows to
private sector
Albania 1995 2006 2004
Algeria 1996 2002 2002
Argentina 1993 1994 1994 1994
Argentina 2005 2006 2006
Bolivia 1993 1994 1994
Bosnia and Herzegovina 1997 2006 2006 2001
Brazil 1994 1995 1995 1995
Bulgaria 1994 2006 1996
Cameroon 2003
Chile 1990 1994 1998 1991
Congo, Dem. Rep. 1989 2003 2003
Costa Rica 1990 1997 1998 1992
Cote d'Ivoire 1998 2004 2004
Dominica 2004
Dominican Republic 1994 2000 2001 2000
Dominican Republic 2005 2006 2006
Ecuador 1995 1996 1997 1996
Ecuador 2000 2005 2005 2001
Ethiopia 1996
Gabon 1994 1999 1999
Gambia, The 1988
Grenada 2005
Guinea 1998
Guyana 1999
Honduras 2001 2004 2004
Jamaica 1990 1993 1998 1993
Jordan 1993 2005 2005 2005
Kenya 1998 2002 2003 2002
Liberia 1982 1983 1983
Macedonia, FYR 1997 1998 2003 1998
Madagascar 1990
Malawi 1988 1989
Mauritania 1996 2001 2001
Mexico 1990 1993 1993 1991
Moldova 2004 2005
Morocco 1990 1994 2003 1993
Mozambique 1991 1993 1992
Nicaragua 1995 2002 1999
Niger 1991
Nigeria 1991 1993 1993
Pakistan 1999 2004 2006 2004
Panama 1996 1998 1998 1997
Paraguay 1993 1995 1999 1994
Peru 1997 1999 1999 1998
Philippines 1992 1994 1996 1993
Poland 1994 1995 1995 1995
Romania 1986 1993 1999 1991
Russian Federation 2000 2002 2002 2002
Sao Tome and Principe 1994
Senegal 1996 2000 1997
Serbia and Montenegro 2004 2005 2005 2005
Sierra Leone 1995
South Africa 1993 1994 1994 1994
Sudan 1985 2000 2000
Tanzania, UR 2004 2005 2005
Togo 1997
Trinidad and Tobago 1989 1990 1992 1990
Turkey 1982 1983 1983 1983
Uganda 1993 2001 2001
Ukraine 2000 2002 2002 2001
Uruguay 1991 1992 1994 1992
Uruguay 2003 2005 2005 2005
Venezuela 1990 1992 1992 1992
Vietnam 1997 2004 2005 2004
Yemen, Rep. 2001 2002 2002
Zambia 1994
Main Definition
(Flows to PUBLIC sector)
38
Table 9: Main Results for Haircuts and Years of Exclusion
Note: The table shows coefficients (not hazard rates) of a Cox proportional hazard model.
The dependent variable measures years from a restructuring until partial reaccess to capital markets,
defined as the first year with (i) issuance of a bond or syndicated loan on international markets
and/or (ii) positive net debt flows to the public sector of the debtor country.
(1) (2) (3) (4) (5) (6) (7)
-0.037*** -0.025*** -0.033*** -0.049*** -0.050*** -0.034*** -0.030***
(0.007) (0.007) (0.006) (0.012) (0.010) (0.009) (0.011)
0.849** 0.430
(0.408) (0.440)
0.221
(0.171)
-0.240 -0.150
(0.160) (0.147)
0.397
(0.471)
2.716
(3.516)
-0.037
(0.061)
-0.034*** -0.032***
(0.011) (0.011)
0.016 -0.030
(0.047) (0.065)
-0.011*** 0.001
(0.003) (0.002)
-1.320
(0.914)
-0.674 0.870
(1.027) (1.117)
0.041
(0.032)
0.071* 0.063**
(0.036) (0.032)
Observations 350 301 350 229 264 242 193
Log-Likelihood -74.10 -83.12 -85.53 -57.17 -76.69 -63.16 -57.44
B.I.C. 259.499 217.596 223.778 163.238 214.720 170.234 188.560
Haircut (SZ, in
%)
GDP per capita
(PPP, log)
High-yield bond
spread
US 10-year
Treasury Yield
Current Account
to GDP
GDP Real Growth
Reserves to
Imports
Credit Rating
(Residual)
Exports to GDP
Political Risk
(ICRG)
Inflation
Time Fixed
Effects (Decades)
Yes (Year
Dummies) Yes Yes Yes Yes Yes Yes
Country
Fundamentals
II
With
Sovereign
Rating
Budget Balance to
GDP
External Debt to
GNI
Full
Model
Region Fixed
Effects Yes Yes Yes Yes Yes Yes Yes
Population (log)
Year
Fixed
Effects
Country
Size and
Wealth
External
Financing
Conditions
Country
Fundamentals
I
39
Table 10: Robustness Analysis of Exclusion Duration
Note: The table shows coefficients (not hazard rates) of a Cox proportional hazard model. The dependent variable
measures years from a restructuring until reaccess to capital markets. In columns 1 and 2 “full reaccess” is
defined as the first year in which (i) the volume of bond issuances or new syndicated loans on international
markets and/or (ii) net debt flows to the public sector of the debtor country, surpass 1% of GDP. The dependent
variable in columns 3 and 4 is the same as in the baseline definition of “partial access” but also takes into account
capital flows to the private sector. It thus extends the baseline definition (see also Table 9 above) by a third
criterion, namely (iii) positive net debt flows to the private sector of the debtor country. The dependent variable
in columns 5 and 6 focuses only on primary market placements using Dealogic data. It measures years from the
restructuring until the first international bond issuance or syndicated loan by the government or a publicly
guaranteed entity. Columns 7 and 8 exclude highly indebted and poor countries.
(1) (2) (3) (4) (5) (6) (7) (8)
-0.03*** -0.02* -0.02*** -0.04*** -0.03*** -0.02** -0.02*** -0.03**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
-0.22** -0.17*** -0.17 -0.17
(0.09) (0.06) (0.13) (0.11)
1.05* -0.79* 0.94* -0.01
(0.58) (0.44) (0.54) (0.07)
-0.06 -0.12** -0.03 -0.36
(0.04) (0.05) (0.06) (0.46)
-0.43 1.47* 1.32 0.69
(1.33) (0.78) (0.99) (0.86)
0.07 0.15*** 0.11*** 0.06*
(0.05) (0.04) (0.04) (0.04)
Observations 505 316 285 161 463 279 143 121
Log-Likelihood -87.40 -68.06 -115.11 -84.51 -83.82 -64.52 -78.24 -65.45
B.I.C. 218.36 205.20 269.79 229.99 210.61 196.62 191.22 188.46
Yes Yes
Partial Acces,
Full Access Without HIPCs
Richmond-Dias
incl. Access by
Private Sector
GDP Real
Growth
Haircut (in %)
High-yield bond
spread
Gelos et al.
Primary Market
Access only
Time Fixed
Effects
Reserves to
Imports
Credit Rating
(Residual)
Yes
GDP p.c.
(PPP, log)
Region Fixed
Effects Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes
40
Table 11: Extended Results for Exclusion Duration – Other Haircut Measures
Note: The table shows coefficients (not hazard rates) of a Cox
proportional hazard model. The dependent variable measures years
from a restructuring until partial reaccess to capital markets, defined
as the first year with (i) issuance of a bond or syndicated loan on
international markets and/or (ii) positive net debt flows to the public
sector of the debtor country.
(1) (2) (3) (4)
-0.03***
(0.01)
-0.04***
(0.01)
-0.03**
(0.01)
-0.03**
(0.01)
-0.11 -0.11 -0.13 -0.08
(0.13) (0.14) (0.13) (0.14)
-0.03*** -0.03*** -0.03*** -0.03***
(0.01) (0.01) (0.01) (0.01)
0.48 0.56 0.58 0.73*
(0.46) (0.46) (0.43) (0.40)
-0.03 -0.04 -0.04 -0.02
(0.07) (0.07) (0.07) (0.07)
1.02 0.65 0.87 0.88
(1.09) (1.17) (1.20) (1.18)
0.07** 0.05 0.07** 0.06*
(0.03) (0.03) (0.03) (0.03)
Observations 200 200 200 200
Log-Likelihood -55.47 -54.20 -55.17 -55.55
B.I.C. 179.81 177.28 179.22 179.97
Region Fixed
Effects Yes Yes Yes Yes
GDP per capita
(PPP, log)
GDP Real Growth
Reserves to
Imports
Credit Rating
(Residual)
Market
Haircut
(eq. 2)
Market Haircut
Face Value
Reduction
Naive Haircut
Time Fixed Effects
(Decades) Yes Yes Yes Yes
Effective Haircut
High-yield bond
spread
External Debt to
GNI
Naive
Haircut
Effective
Haircut
Face
Value
Reduction
41
References
Aguiar, Mark, and Gita Gopinath. 2006. “Defaultable Debt, Interest Rates and the
Current Account." Journal of International Economics, 69(1): 64-83.
Allison, Paul. 2002. “Bias in fixed-effects Cox regression with dummy variables.”
Unpublished Paper, University of Pennsylvania.
Altman, Edward I.. 1987. “The Anatomy of the High-Yield Bond Market.” Financial
Analysts Journal, 43: 12-25.
Altman, Edward I.. 1989. “Measuring Corporate Bond Mortality and Performance.”
Journal of Finance, 44: 909-922.
Altman, Edward I., and Vellore M. Kishore. 1996. “Almost Everything You Wanted to
Know about Recoveries on Defaulted Bonds.” Financial Analysts Journal, 52(6): 57-
64.
Amador, Manuel. 2010. “Sovereign Debt and the Tragedy of the Commons.”
Unpublished Paper, Stanford University.
Andersen, Per Kragh, John P. Klein, and Mei-Jie Zhang. 1999. “Testing for Centre
Effects in Multi- Centre Survival Studies: A Monte Carlo Comparison of Fixed and
Random Effects Tests." Statistics in Medicine, 18(12): 1489-1500.
Andritzky, Jochen. 2006. Sovereign Default Risk Valuation: Implications of Debt Crises
and Bond Restructurings. New York: Springer.
Arellano, Cristina. 2008. “Default Risk and Income Fluctuations in Emerging
Economies." American Economic Review, 98(3): 690-712.
Asonuma, Tamon. 2010. “Serial Default and Debt Renegotiation.” Unpublished Paper,
Boston University.
Bedford, Paul, Adrian Penalver and Chris Salmon. 2005. “Resolving Sovereign Debt
Crises: The Market based Approach and the Role of the IMF.” Bank of England
Financial Stability Review, (18): 91 - 100.
Benczur, Peter and Cosmin Ilut. 2009. “Evidence for Relational Contracts in Sovereign
Bank Lending.” Unpublished Paper, Duke University.
Benjamin, David, and Mark L.J. Wright. 2009. “Recovery Before Redemption? A
Theory of Delays in Sovereign Debt Renegotiations.” Unpublished Paper, University of
California at Los Angeles.
Borensztein, Eduardo, and Ugo Panizza. 2008. “The Costs of Sovereign Default.”
International Monetary Fund Working Paper 08/238.
Buchheit, Lee, and G. Mitu Gulati. 2002. “Sovereign Bonds and the Collective Will.”
Emory Law Journal, 51:1317.
42
Bulow, Jeremy, and Kenneth S. Rogoff. 1989. “Sovereign Debt: Is to Forgive to
Forget?” American Economic Review, 79(1): 43–50.
Cavallo, Eduardo, Andrew Powell, and Roberto Rigobon. 2008. “Do Credit Rating
Agencies Add Value? Evidence from the Sovereign Rating Business Institutions.” Inter-
American Development Bank Research Department Working Paper 4601.
Clark, John. 1990. “The Evaluation of Debt Exchanges.” International Monetary Fund
Working Paper 90/9.
Claessens, Stijn, Ishac Diwan and Eduardo Fernandez-Arias. 1992. “Recent Experience
with Commercial Bank Debt Reduction.” World Bank Policy Research Working Paper
995.
Cline, William R. 1995. International Debt Reexamined. Washington, D.C.: Institute for
International Economics.
D’Erasmo, Pablo. 2010. “Government Reputation and Debt Repayment.” Unpublished
Paper, University of Maryland.
Dell’Ariccia, Giovanni, Isabel Schnabel, and Jeromin Zettelmeyer. 2006. “How Do
Official Bailouts Affect the Risk of Investing in Emerging Markets?.” Journal of
Money, Credit, and Banking, 38(7): 1689–1714.
Díaz-Cassou, Javier, Aitor Erce-Domínguez and Juan J. Vázquez-Zamora. 2008a.
"Recent Episodes of Sovereign Debt Restructurings. A Case-Study Approach." Bank of
Spain Occasional Paper 0804.
Díaz-Cassou, Javier, Aitor Erce-Domínguez, and Juan J. Vázquez-Zamora. 2008b.
"The role of the IMF in recent sovereign debt restructurings: Implications for the policy
of lending into arrears." Bank of Spain Occasional Paper 0805.
Dikhanov, Yuri. 2004. “Historical Present Value of Debt in Developing Countries.”
Unpublished Paper, The World Bank.
Eaton, Jonathan, and Mark Gersovitz. 1981. “Debt with Potential Repudiation:
Theoretical and Empirical Analysis.” Review of Economic Studies, 48(2): 289-309.
Eaton, Jonathan, and Raquel Fernandez. 1995. “Sovereign Debt.” In Handbook of
International Economics, ed. Grossman G.M., and Kenneth Rogoff, 3: 2031-2077.
Amsterdam: Elsevier, North-Holland.
Edwards, Sebastian. 1986. “The Pricing of Bonds and Bank Loans in International
Markets: An Empirical Analysis of Developing Countries' Foreign Borrowing.”
European Economic Review, 30(3): 565-589.
Eichengreen, Barry. 1989. “The U.S. Capital Market and Foreign Lending, 1920-1955.”
In Developing country debt and economic performance. Volume 1: The international
financial system, ed. Jeffrey Sachs, 107-155. Chicago and London: University of
Chicago Press.
43
Eichengreen, Barry, and Ashoka Mody. 2000. “What Explains Changing Spreads on
Emerging-Market Debt? ” In Capital Flows and the Emerging Economies: Theory,
Evidence and Controversies, ed. Sebastian Edwards, 107–134. Chicago and London:
The University of Chicago Press.
Eichengreen, B., and R. Portes. 1986. “Debt and default in the 1930s: Causes and
consequences.” European Economic Review, 30(3): 599-640.
Eichengreen, B., and R. Portes. 1989. “After the deluge: default, negotiation, and
readjustment during the interwar years.” In The International Debt Crisis in Historical
Perspective, ed. Eichengreen, B. and Lindert, P.H., 12-47. Cambridge, MA: MIT Press.
Esteves, Rui P. 2007. “Quis custodiet quem? Sovereign Debt and Bondholders'
Protection Before 1914.” University of Oxford Economics Series Working Papers 323.
Finger, Harald, and Mauro Mecagni. 2007. “Sovereign Debt Restructuring and Debt
Sustainability. An Analysis of Recent Cross-country Experience.” International
Monetary Fund Occasional Paper 255.
Folkerts-Landau, David. 1985. “The Changing Role of International Bank Lending in
Development Finance.” IM F Staff Papers, 32(2): 317-363.
Gelos, Gaston R., Ratna Sahay, and Guido Sandleris. 2004. "Sovereign Borrowing by
Developing Countries: What Determines Market Access?" International Monetary Fund
Working Paper 04/221.
Grossman, Herschel I., and John B. Van Huyck. 1988. “Sovereign Debt as a Contingent
Claim: Excusable Default, Repudiation, and Reputation.” American Economic Review,
78(5): 1088-1097.
Jorgensen, Erika, and Jeffrey Sachs. 1989. “Default and Renegotiation of Latin
American Foreign Bonds in the Interwar Period.” In The International Debt Crisis in
Historical Perspective, ed. Barry Eichengreen and Peter H. Lindert, 48–85. Cambridge,
MA. and London: MIT Press.
Kalbfleisch, Jack D., and Ross L. Prentice. 2002. The Statistical Analysis of Failure
Time Data. 2nd ed. New York: Wiley-Interscience.
Kaminsky, Graciela, and Sergio L. Schmukler. 2002. “Emerging Market Instability: Do
Sovereign Ratings Affect Country Risk and Stock Returns?” The World Bank Economic
Review, 16(2): 171-195.
Kelly, Patrick J., and Lynette L-Y. Lim. 2000. “Survival Analysis for Recurrent Event
Data: An Application to Childhood Infectious Diseases.” Statistics in Medicine, 19(1):
13–33.
Kletzer, Kenneth M., and Brian D. Wright. 2000. "Sovereign Debt as Intertemporal
Barter." American Economic Review, 90(3): 621-639.
44
Klingen, Christoph, Beatrice Weder, and Jeromin Zettelmeyer. 2004. “How Private
Creditors Fared in Emerging Debt Markets, 1970-2000.” International Monetary Fund
Working Paper 04/13.
Kozack, Julie. 2005. “Considerations in the Choice of the Appropriate Discount Rate
for Evaluating Sovereign Debt Restructurings.” IMF Policy Discussion Paper 05/09.
Lancaster, Tony. 1990. The econometric analysis of transition data. Cambridge; New
York: Cambridge University Press.
Lee, Suk Hun. 1991. “Using Terms of Rescheduling as Proxy for Partial Reneging on
LDC's Debt in a Test of Willingness-to-Pay Model.” Journal of International Money
and Finance, 10(3): 457-477.
Levy Yeyati, Eduardo. 2009. “Optimal Debt? On the Insurance Value of International
Debt Flows to Developing Countries.” Open Economies Review, 20(4): 489-507.
Lin, D.Y. and Wei L.J. 1989. “The Robust Inference for the Cox Proportional Hazards
Model.” Journal of the American Statistical Association, 84(408): 1074-1078.
Lindert, Peter H., and Peter J. Morton. 1989. “How Sovereign Debt Has Worked.” In
Developing Country Debt and Economic Performance, Volume 1: The International
Financial System, ed. Jeffrey Sachs, 39–106. Chicago and London: University of
Chicago Press.
Moody’s. 2006. “Default and Recovery Rates of Corporate Bond Issuers, 1920-2005.”
Special Comment. http://www.moodys.com.br/brasil/pdf/Default2006.pdf
Özler, Sule. 1993. “Have Commercial Banks Ignored History?” American Economic
Review, 83(3): 608–620.
Panizza, Ugo, Federico Sturzenegger and Jeromin Zettelmeyer. 2009. “The Law and
Economics of Sovereign Debt and Default.” Journal of Economic Literature, 47(3):
651-98.
Pitchford, R., and Mark L.J. Wright. 2007. “Restructuring the Sovereign Debt
Restructuring Mechanism.” University of California at Los Angeles Mimeo.
Reinhart, Carmen and Kenneth Rogoff. 2009. This Time is Different. Eight Centuries of
Financial Folly. Princeton: Princeton University Press.
Reinhart, Carmen M., Kenneth Rogoff, and Miguel A. Savastano. 2003. “Debt
Intolerance.” Brookings Papers on Economic Activity.
Richmond, Christine, and Daniel A. Dias. 2009. “Duration of Capital Market Exclusion:
An Empirical Investigation.” Unpublished Paper, Universidad of California at Los
Angeles.
Rieffel, Lex. 2003. Restructuring Sovereign Debt: The Case for Ad Hoc Machinery.
Washington, D.C.: Brookings Institution Press.
45
Sandleris, Guido. 2008. “Sovereign Defaults: Information, Investment and Credit.”
Journal of International Economics, 76(2): 267–275.
Standard & Poor’s. 2007. Sovereign Defaults and Rating Transition Data: 2006 Update.
Standard & Poor’s, RatingsDirect.
Sturzenegger, Federico and Jeromin Zettelmeyer. 2007a. Debt Defaults and Lessons
form a Decade of Crises. Cambridge, MA: MIT Press.
Sturzenegger, Federico and Jeromin Zettelmeyer. 2007b. “Creditors Losses versus Debt
Relief: Results from a Decade of Sovereign Debt Crises.” Journal of the European
Economic Association, 5(2-3): 343-351.
Sturzenegger, Federico and Jeromin Zettelmeyer. 2008. “Haircuts: Estimating Investor
Losses in Sovereign Debt Restructurings, 1998-2005.” Journal of International Money
and Finance, 27(5): 780-805.
Sy, Amadou N.R. 2004. “Rating the rating agencies: Anticipating currency crises or
debt crises?” Journal of Banking and Finance, 28(11): 2845-2867.
Tomz, Michael. 2007. Reputation and International Cooperation: Sovereign Debt
across Three Centuries. Princeton and Oxford: Princeton University Press.
Tomz, Mike and Mark L.J. Wright. 2007. “Do Countries Default in “Bad Times”?”
Journal of the European Economic Association, 5(2-3): 352-360.
Trebesch, Christoph. 2008. “Delays in Sovereign Debt Restructuring.” Unpublished
Paper, Free University of Berlin.
World Bank. 2007. “Debt Reduction Facility for IDA-Only Countries: Progress Update
and Proposed Extension,” Board Report 39310, Washington, D.C.
World Bank. 2010 “Heavily Indebted Poor Countries (HIPC) Initiative and Multilateral
Debt Relief Initiative (MDRI)―Status of Implementation”, Washington, D.C
Wright, Mark L. J. 2002. “Reputations and Sovereign Debt.” Working paper, University
of California at Los Angeles.
Yue, Vivian Z. 2010. “Sovereign Default and Debt Renegotiation.” Journal of
International Economics, 80(2): 176-187.