In lieu of an abstract, here is a brief excerpt of the content:

• Panel on Shrinking the Federal Reserve Balance Sheet
• Arvind Krishnamurthy, Sydney C. Ludvigson, and Jonathan H. Wright

• Lessons for Policy from Research
• Arvind Krishnamurthy

ABSTRACT

I review lessons from the research on central bank actions over the last decade and draw out implications for expanding the Federal Reserve balance sheet (quantitative easing) and shrinking the balance sheet (quantitative tightening). As I outline, there is already enough evidence in the research to indicate the manner in which the Federal Reserve could update its policy normalization principles and plans.

Former Federal Reserve chairman Ben Bernanke famously quipped, in a 2014 discussion at the Brookings Institution, that "the problem with QE is that it works in practice, but it doesn't work in theory." Academic and policy research on quantitative easing (QE) has come quite far over the last decade, and we are less in the dark about the workings of QE. In this paper, I review the lessons from this research and then draw out implications for expanding the Federal Reserve balance sheet (QE) and shrinking the balance sheet (quantitative tightening, or QT).

There are three principal lessons from the research: (1) QE works differently than conventional monetary policy in that the impacts are highest in the asset market targeted. (2) QE impacts are highest during periods of financial distress, market segmentation, and illiquidity. While this statement is likely also true of conventional policy, the effects are much more dramatic with QE. (3) QE alters the quantity of central bank reserves, and the post-2008 regulatory and economic regime implies substantially higher necessary reserve balances. I review each of these points and then turn to their implications for the formulation of rules governing QE/QT. The Fed [End Page 233]

Click for larger view
View full resolution Figure 1.

Yield Changes by Maturity from UK QE for UK Gilts and Gilt-OIS Spreads

Source: Joyce and others (2011); copyright Bank of England and the Association of the International Journal of Central Banking; adapted with permission.

currently uses QE in two ways: to provide liquidity to markets during financial illiquidity episodes ("crisis QE") and to lower financing costs for borrowers at a time when the zero lower bound binds ("easing QE"). I argue that rules for these two types of policies should differ, but that the Fed has blurred the lines between them which has led to policy errors.

I. Lessons from Research

I.A. QE Works through Narrow Channels

Joyce and others (2011) present data from an event study around two significant QE news dates in 2009 by the Bank of England. On February 11, 2009, the Inflation Report and the subsequent press conference gave a strong indication that the bank would do QE. Markets interpreted this to mean that the bank would purchase bonds out to around fifteen-year maturity. On March 5, 2009, the bank announced that purchases would be in the five- to twenty-five-year range. Figure 1, replicating figure 4 in Joyce and others (2011), shows the changes in gilt yields around the event dates and the changes in the spread between gilt and overnight index swap (OIS) yields around these dates. Panel A shows the market reaction to the [End Page 234] February announcements: yields fall across the board. The pattern is similar to a conventional policy response in that there are larger effects on short-term bonds than longer-term bonds. In the curve showing the yield-OIS spread change, we see unique QE effects. If the policy transmission is akin to conventional monetary policy, there should be no change in these spreads as we would expect that both gilt yields and OIS yields will move in lockstep so that their spread would not change. Panel B shows the market reaction to the March announcement, and here we can really see the unique QE effects. First note that the effect on gilt yields is concentrated in the five- to twenty-five-year range, which the bank indicated as the target of QE purchases, with yields in the fifteen- to twenty-five-year range falling dramatically on the news that these maturities would also be purchased. Second, note that the yield-OIS spread change reflects the...

中文翻译：

---

缩减美联储资产负债表小组

以下是内容的简短摘录，以代替摘要：

• 缩减美联储资产负债表小组
• 阿文德·克里希那穆西、西德尼·C·路德维森和乔纳森·H·赖特

• 研究中的政策教训
• 阿尔温德·克里希那穆蒂

抽象的

我回顾了过去十年央行行动研究的经验教训，并得出了扩大美联储资产负债表（量化宽松）和收缩资产负债表（量化紧缩）的影响。正如我所概述的，研究中已经有足够的证据表明美联储可以更新其政策正常化原则和计划的方式。

美联储前主席本·伯南克 (Ben Bernanke) 在布鲁金斯学会 2014 年的一次讨论中曾说过一句著名的俏皮话，“量化宽松的问题在于它在实践中有效，但在理论上不起作用。” 过去十年来，有关量化宽松 (QE) 的学术和政策研究取得了很大进展，我们对量化宽松的运作方式也不再一无所知。在本文中，我回顾了这项研究的教训，然后得出了扩大美联储资产负债表（QE）和收缩资产负债表（量化紧缩，或 QT）的影响。

研究得出三个主要教训：（1）量化宽松的运作方式与传统货币政策不同，因为它对目标资产市场的影响最大。(2) 在金融危机、市场分割和流动性不足的时期，量化宽松的影响最大。虽然这种说法也可能适用于传统政策，但量化宽松的效果要显着得多。(3)量化宽松改变了央行准备金的数量，2008年后的监管和经济体制意味着必要的准备金余额大幅增加。我回顾了每一点，然后转向它们对制定量化宽松/量化交易规则的影响。美联储【第233页完】

单击查看大图
查看完整分辨率图 1。

英国量化宽松政策对英国金边债券和金边债券-OIS 利差的收益率变化

资料来源：乔伊斯等人 (2011)；版权所有英格兰银行和国际中央银行杂志协会；经许可改编。

目前以两种方式使用量化宽松：在金融流动性不足的情况下向市场提供流动性（“危机量化宽松”），以及在零下限约束时降低借款人的融资成本（“宽松量化宽松”）。我认为这两类政策的规则应该有所不同，但美联储模糊了它们之间的界限，从而导致了政策错误。

一、研究的教训

IA 量化宽松通过狭窄渠道发挥作用

Joyce 等人（2011 年）提供了围绕英格兰银行 2009 年两个重要量化宽松新闻日期进行的事件研究的数据。2009年2月11日，通胀报告和随后的新闻发布会强烈暗示央行将实施量化宽松。市场将此解读为该银行将购买期限约为十五年的债券。2009 年 3 月 5 日，该银行宣布购买期限为 5 至 25 年。图 1 复制了 Joyce 等人（2011 年）中的图 4，显示了事件日期前后英国国债收益率的变化以及这些日期前后英国国债和隔夜指数掉期 (OIS) 收益率之间利差的变化。图 A 显示了市场对[完第 234 页]的反应2月份公告：收益率全线下跌。这种模式与传统的政策反应类似，对短期债券的影响大于对长期债券的影响。在显示收益率与 OIS 利差变化的曲线中，我们看到了独特的量化宽松效应。如果政策传导类似于传统货币政策，那么这些利差应该不会发生变化，因为我们预计英国国债收益率和 OIS 收益率将同步变化，因此它们的利差不会发生变化。B图显示了市场对3月份公告的反应，在这里我们可以真正看到独特的QE效应。首先要注意的是，对英国国债收益率的影响集中在五年至二十五年期范围内，该银行将其列为量化宽松购买的目标，随着这些期限债券也将被购买的消息传出，十五至二十五年期债券的收益率急剧下降。其次，请注意，收益率-OIS 利差变化反映了……