NBER Reporter NATIONAL BUREAU OF ECONOMIC RESEARCH

A quarterly summary of NBER research No. 3, September 2018

The 2018 Martin Feldstein Lecture

The Two Faces of Liquidity

Raghuram Rajan*

It has been about 10 years since the global financial crisis. What have we learnt about it? The behavior of financial-sector participants was clearly aberrant, and needed to be rectified. We have had a tre- mendous amount of regulation since then, especially focused on banks. Whether this will solve the underlying problems is an issue that is much debated. Yet we have at least attempted to tackle the problems of poor incentives and distorted financial firm capital structures, includ- ing mismatched asset-liability structures. We have paid far less attention to the shadow financial sector — the Raghuram Rajan financial institutions other than the banks — or to the role of macro- economic conditions in precipitating the crisis. We do acknowledge ALSO IN THIS ISSUE the need to bring the shadow sector under the regulatory net, for we Environment, Energy, and did learn that institutions can infect one another through common Unintended Consequences 10 markets. We are less clear about what ought to be done. On macroeco- nomic conditions, even when we do acknowledge a role, when it comes Taxation and Innovation 14 to altering policy, we shrug and move on. Somehow, it seems that we The Economics of Drug Development 19 have agreed that macroeconomic conditions are outside the remit of Mortgage Lending and anyone tasked with addressing financial stability. I will argue in my Housing Markets 25 talk today that this is a mistake. NBER News 29 Conferences 31 Program and Working Group Meetings 38 * Raghuram Rajan is the Katherine Dusak Miller Distinguished Service Professor of Finance at the University of Chicago’s Booth School of Business. He served as the 23rd Governor of the Reserve Bank of India between September 2013 and September 2016, and was chief economist and director of research at the International Monetary Fund 2003–06. He delivered the 2018 Martin Feldstein Lecture, based on this article, at the NBER Summer Institute on July 10, 2018. The lecture is named in honor of the NBER’s president emeritus.

Reporter Online at: www.nber.org/reporter the United States, with major components The IMF Financial Conditions Index, 1991–2017 of financial conditions tightening with a lag, especially after Taylor rule residuals Median cross-country measure of financial conditions become positive. Put differently, part of the 2.0 Easing Easing reason there is so much of a gap between 1.5 when the Fed started raising interest rates and when financial conditions started tight- 1.0 ening may well be because monetary condi- 0.5 tions remained very accommodative until 2006. 0.0 These figures are obviously not proof

-0.5 that credit conditions remained easy before the crisis because monetary policy was easy. -1.0 All I want to establish is that it is not Tightening entirely ridiculous to argue that monetary -1.5 policy’s influence needs to be investigated Q1 1992 Q1 1997 Q1 2002 Q1 2007 Q1 2012 Q1 2017 in any post-mortem of the crisis. However, for the rest of the talk, it is sufficient for my The Financial Conditions Index is an internationally representative composite of various economic indicators purposes if you grant me that financial con- Source: The International Monetary Fund ditions were easy for a long period before the crisis. Figure 1 Financial Conditions Taylor rule, an empirical description of how The Consequences of Easy and Monetary Policy policy is ordinarily set). It also indicates Financing Conditions that lending standards for corporate loans Figure 1, computed by the IMF, is the and mortgage loans did not start tighten- What are the consequences of easy median at every point in time of the distri- ing until after Taylor rule residuals became financing conditions? The seminal work of bution of financial conditions across coun- positive. In other words, there seems to be a Claudio Borio and Philip Lowe suggests sus- tries, with higher being easier.1 lag between a tightening of monetary policy tained rapid credit growth combined with What we see is that leading up to the and a tightening of representative elements large increases in asset prices increases the financial crisis, we have a tremendous easing of financial conditions. probability of an episode of financial insta- of financial conditions, even though the Fed In Figure 3, we see a similar picture for bility.3 More recently, in a study of crises, started raising the federal Arvind Krishnamurthy funds rate in June 2004. As and Tyler Muir show the crisis spread in 2007, Euro Area Taylor Rule Residuals and Lending Standards that credit growth and we had an extremely sharp credit spreads are nega- tightening of financial con- Taylor Rule residuals (percentage points) Banks reporting a tightening of their lending standards (%) tively correlated before a ditions. By the middle of 1.5 80 crisis begins, with spreads Corporate loans 2009, you see financial con- 1.0 widening a little only just ditions easing once again, 60 before the onset of the and they have continued to 0.5 crisis.4 The change in become much easier. 40 credit spreads as the cri- Now consider two 0.0 sis occurs seems to pre- 20 figures from the work -0.5 dict the extent of the sub- of Angela Maddaloni sequent output decline. 2 0 and José-Luis Peydró. -1.0 Mortgage loans It seems the credit mar- Figure 2 suggests that kets do not really price monetary policy in the -1.5 -20 in the crisis until it is Eurozone was very accom- Q4 2002 Q4 2003 Q4 2004 Q4 2005 Q4 2006 Q4 2007 almost upon them; the modative before the cri- greater their compla- Taylor Rule residuals are a measure of expansionary or contractionary monetary policy. Each residual is the dierence sis, as measured by Taylor between the policy rate and the rate suggested by the Taylor Rule – an empirical description of how policy is ordinarily set cency, the greater seems rule residuals (the differ- Source: Maddaloni and Peydró, European Central Bank Working Paper Series, No. 1248 the gravity of the cri- ence between policy rates sis. David López-Salido, and rates suggested by the Figure 2 Jeremy Stein, and Egon

2 NBER Reporter • No. 3, September 2018 Zakrajšek5 and Atif Mian, Amir Sufi, and since everybody was doing it, Citi could tors get lulled by a series of good signals to Emil Verner6 find similar effects. What the- not be the only one to stop. Herd behavior overweight the probability that the state of ories might account for this? models of banking such as one I have ana- the world is good.8 A few bad signals do not Three come immediately to mind. One lyzed have this flavor; being the lone lender cause investors to worry that the bad state is herd behavior in banking markets. Perhaps to stop lending has costs.7 For instance, in a may be imminent, because they still think the most famous statement made before the credit boom, loan officers may believe they the good state is likely. Eventually, though, crisis was by Chuck Prince, the chairman of have to maintain the pretense that they have enough bad news leads to a radical change Citigroup, who responded thus to a ques- really good lending prospects, and credit in beliefs, and investor pessimism causes the financial crisis. So there is a neglect of the “tail” bad state initially and excessive credit U.S. Taylor Rule Residuals and Lending Standards extension, an initial underreaction to bad news, and eventually, overreaction. Taylor Rule residuals (percentage points) Banks reporting a tightening of their lending standards (%) A different behavioral model, in which 3 100 there is a distribution of optimists and pes- 2 80 simists in the economy, is advanced by John Geanakoplos.9 Relative to pessimists, opti- 1 60 mists think there is a higher probability of 0 40 the good state, where the price of the asset Corporate loans being traded will be higher still. They are -1 20 willing to buy the asset, and even borrow to -2 0 buy yet more. The pessimists sell at the price available in the market, and lend money -3 Mortgage loans -20 to the optimists. The arrival of good news -4 -40 therefore enhances the wealth of optimists, Q4 1992 Q4 1995 Q4 1998 Q4 2001 Q4 2004 Q4 2007 and their positive views have greater impact on the asset price. In contrast, bad news ren- Taylor Rule residuals are a measure of expansionary or contractionary monetary policy. Each residual is the dierence ders a few optimists bankrupt, and allows between the policy rate and the rate suggested by the Taylor Rule – an empirical description of how policy is ordinarily set the consequent preponderance of pessimists Source: Maddaloni and Peydró, European Central Bank Working Paper Series, No. 1248 to push down the asset price. Therefore there is overreaction to news in either direc- Figure 3 tion because it changes the wealth of players, tion from the Financial Times: “When the quality has not deteriorated, since no one and therefore effectively changes the mone- music stops, in terms of liquidity, things will else seems to be worried or shows signs of tary weight placed on beliefs. So behavioral be complicated. But as long as the music problems. Rather than be singled out as the explanations of various kinds could be use- is playing, you’ve got to get up and dance. incompetent lender who cannot find good ful in explaining the abrupt shift that we We’re still dancing.” prospects, the loan officer ensures credit saw from the complacency before the crisis I met Mr. Prince at a conference a few spreads do not reflect deterioration, and to the panic once it got underway. years later and I asked him, “So why did you neither does loan volume. Indeed, if any say that?” He responded that he had a whole borrower cannot repay, he lends him more A Liquidity-based Model of lot of investment bankers who were doing money to paper over default, thus “ever- the Consequences of Easy deals. And while he was aware the qual- greening” the loan. It is only when credit Financing Conditions ity of deals was falling, he knew that if he problems become overwhelming, marking stopped them from doing more deals, they the start of the crisis, that banks start declar- As if we do not already have enough would have left immediately for a job with ing their true losses and stop lending, at models, let me add one more. Unlike these the competition. Hiring and grooming peo- which point it is much easier for everyone other models, it will relate leverage increases ple is tough. So in an attempt to preserve the else to disclose their mistakes, since they and spread declines directly to the easy franchise, so to speak, he had to let them go will no longer be outliers. The longer the liquidity that prevailed before the crisis. a little more to the edge. period of hiding, of course, the greater the It will explain why spreads were flat until Now, of course, he was making a calcu- losses that have built up, and the greater the just before the crisis. We will see why sus- lation that retaining these people was more deterioration in credit quality when the cri- tained expectations of high future liquid- important than the cost to the balance sheet sis starts. ity can make the subsequent downturn lon- in terms of deteriorating credit quality. The Then there are true behavioral models ger and deeper, other than just through the statement came back to haunt him. But nev- such as that of Nicola Gennaioli, Andrei increase in leverage during the run-up to ertheless, there was a rationale, which is that Shleifer, and Robert Vishny in which inves- the crisis and the change in expectations

NBER Reporter • No. 3, September 2018 3 at its onset. This model is detailed in my work with Douglas Diamond and Yunzhi Hu.10 Essentially, I NBER Reporter will argue that high expectations of liquidity can be problematic. I will then propose another liquidity-based The National Bureau of Economic Research is a private, nonprofit research orga- model to explain why the downturn was so sharp and nization founded in 1920 and devoted to objective quantitative analysis of the the recovery so abrupt. Taken together, we will see American economy. Its officers and board of directors are: the two faces of financial liquidity, the title of this President and Chief Executive Officer — James M. Poterba talk, and why the authorities have the serious chal- Controller — Kelly Horak Corporate Secretary — Alterra Milone lenge of keeping liquidity at the right level — neither too high nor too low. BOARD OF DIRECTORS Consider an industry that requires special indus- Chair — Karen N. Horn try knowledge to produce. Within the industry, Vice Chair — John Lipsky Treasurer — Robert Mednick there are firms run by expert incumbents. There are DIRECTORS AT LARGE also industry insiders, who know the industry well enough to be able to run firms as efficiently as the Peter Aldrich Diana Farrell Karen Mills Elizabeth E. Bailey Jacob A. Frenkel Michael H. Moskow incumbents. Industry outsiders — such as financiers John H. Biggs Robert S. Hamada Alicia H. Munnell who don’t really know how to run industry firms but John S. Clarkeson Peter Blair Henry Robert T. Parry have general managerial/financial skills — are the Kathleen B. Cooper Karen N. Horn James M. Poterba Charles H. Dallara Lisa Jordan John S. Reed other agents in the model. George C. Eads John Lipsky Marina v. N. Whitman Financiers have two sorts of control rights; Jessica P. Einhorn Laurence H. Meyer Martin B. Zimmerman first, control through the right to repossess and sell Mohamed El-Erian the underlying asset being financed if payments are DIRECTORS BY UNIVERSITY APPOINTMENT missed, and second, control over cash flows generated Timothy Bresnahan, Stanford Samuel Kortum, Yale by the asset. The first right only requires the friction- Pierre-André Chiappori, Columbia George Mailath, Pennsylvania less enforcement of property rights in the economy, Alan V. Deardorff, Michigan Marjorie B. McElroy, Duke which we assume. It has especial value when there is Edward Foster, Minnesota Joel Mokyr, Northwestern John P. Gould, Chicago Cecilia Elena Rouse, Princeton a large number of capable potential buyers willing to Mark Grinblatt, California, Los Angeles Richard L. Schmalensee, MIT pay a high price for the firm’s assets. Greater wealth Bruce Hansen, Wisconsin-Madison Ing o Wa l t er, New York amongst industry insiders — which we term industry Benjamin Hermalin, California, Berkeley David B. Yoffie, Harvard liquidity — increases the availability of this asset-sale- DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS based financing. Because we analyze a single indus- Jean-Paul Chavas, Agricultural and Applied Economics Association try, high levels of this industry liquidity can be inter- Martin Gruber, American Finance Association Philip Hoffman, Economic History Association preted as an economy-wide boom. Easy monetary Arthur Kennickell, American Statistical Association policy, with lower than normal policy rates, would Jack Kleinhenz, National Association for Business Economics contribute positively to industry liquidity. Clearly, Robert Mednick, American Institute of Certified Public Accountants Peter L. Rousseau, American Economic Association this kind of control right is exogenous to the firm. Gregor W. Smith, Canadian Economics Association The second type of control right is more endoge- William Spriggs, American Federation of Labor and nous, and conferred on creditors by the firm’s incum- Congress of Industrial Organizations Bart van Ark, The Conference Board bent manager as she makes the firm’s cash flows more appropriable by, or pledgeable to, creditors over the The NBER depends on funding from individuals, corporations, and private medium term. She could do this, for example, by foundations to maintain its independence and its flexibility in choosing its research activities. Inquiries concerning contributions may be addressed to James improving accounting quality or setting up escrow M. Poterba, President & CEO, NBER, 1050 Massachusetts Avenue, Cambridge, accounts so that cash flows are hard to divert. We MA 02138-5398. All contributions to the NBER are tax-deductible. assume enhancing pledgeability takes time to set up but is also semi-durable; improving accounting qual- The Reporter is issued for informational purposes and has not been reviewed by the Board of Directors of the NBER. It is not copyrighted and can be freely repro- ity is not instantaneous, because it requires adopting duced with appropriate attribution of source. Please provide the NBER’s Public new systems and hiring reputable people, but equally, Information Department with copies of anything reproduced. firing a reputable accountant or changing accounting practices has to be done slowly — perhaps at the time Requests for subscriptions, changes of address, and cancellations should be sent to Reporter, National Bureau of Economic Research, Inc., 1050 Massachusetts the accountant’s term ends — if it is not to be noticed. Avenue, Cambridge, MA 02138-5398 (please include the current mailing label), So the incumbent manager sets pledgeability one or by email to [email protected]. Print copies of the Reporter are only mailed to period in advance, and it lasts a period. subscribers in the U.S. and Canada; those in other nations may request electronic subscriptions at www.nber.org/drsubscribe/. The two sources of control rights interact. When

4 NBER Reporter • No. 3, September 2018 anticipated industry liquidity is suffi- creditors to collect more if the incum- credit spreads that lenders charge are con- ciently high, increased pledgeability has bent stays in control, because the creditors tained, even as borrowing increases. no effect on how much industry insid- have the right to seize assets and sell them However, consider now a sustained ers will bid to pay for the firm; they have when not paid in full. In such situations, boom that is anticipated to continue with enough wealth to buy the firm at full the incumbent has to “buy” the firm from high probability, where industry insiders value without needing to borrow much creditors, by outbidding industry insid- will have plenty of wealth. Repayment of against the firm’s future cash flows. Higher ers or paying debt fully, which reduces her any corporate borrowing today is enforced pledgeability is not needed for enhanc- incentive to enhance pledgeability. entirely by the potential high resale value ing bids when the industry is very liq- The tradeoff in setting pledgeability of the firm — at the future date, wealthy uid. When anticipated liquidity is lower depends both on the probability she will industry insiders will bid full value for the so that industry insiders have to borrow need to sell and on the amount that she firm without needing high pledgeability substantially to bid for the firm, higher has promised to pay creditors — as well to make their bid, as described by Shleifer pledgeability does enhance their bids. as industry liquidity, as explained earlier. and Vishny.11 Let us now understand an incumbent A higher promised payment increases the Since pledgeability is not needed to firm manager’s incentives while choos- amount that she needs to pay to “buy” enforce repayment in a future highly liq- ing cash flow pledgeability for the next the firm from creditors but reduces the uid state, a high probability of such a period. Let us assume she may have some residual proceeds that she receives if she state encourages high borrowing up front, reason to sell some or all of the firm sells the firm. Therefore, higher prom- which crowds out the incumbent’s incen-

How Anticipated Market Liquidity Crowds Out Future Pledgeability

1 2 3 4 5

High probability More borrowing Higher-price Low incentive High capacity to of high future available up bid for firm, to raise repay debt liquidity front to buy firm higher debt pledgeability

Source: Illustrative diagram representing author’s theory

Figure 4 next period with some probability, either ised payments exacerbate the incumbent’s tive to enhance pledgeability, even if there because she is no longer capable of run- moral hazard associated with pledgeabil- is a possible low liquidity state where ning it or she needs to finance a new ity, and when they exceed a threshold, pledgeability is needed to enhance credi- investment. Higher pledgeability gener- the incumbent will set pledgeability low. tor rights. In other words, when prospec- ally increases the price at which she can Anticipating this, creditors will limit how tive liquidity exceeds a threshold, lenders sell the firm when she is no longer capa- much they will lend to the incumbent up stop imposing any constraint on leverage, ble of running it, because industry insid- front. and take their chances if that liquidity ers can borrow against future pledgeable In sum then, we have two outside does not materialize. Leverage increases, cash flows to finance their bids for the influences on pledgeability — the antici- but becomes riskier. Figure 4 summarizes firm. It increases the firm’s debt capacity pated liquidity of industry insiders and this sequence. up front — we assume the incumbent has the level of outstanding debt. In normal A crisis or downturn under these cir- bought the firm by borrowing as much as times, the need to provide the incumbent cumstances occurs when liquidity does she can to buy it, which also allows us to incentives for pledgeability keeps up- not materialize. If the low liquidity state examine leverage choices. front borrowing moderate. As prospec- is realized, the enforceability of the firm’s However, having borrowed up front, tive industry liquidity increases, though, debt, as well as its borrowing capacity, the incumbent faces moral hazard associ- the incumbent is able to borrow more to will fall significantly because pledgeabil- ated with pledgeability. A higher bid from finance the asset, while still retaining the ity has been set low. Industry insiders, industry insiders also enables the existing incentive to set pledgeability high. The also hit by the downturn, no longer have

NBER Reporter • No. 3, September 2018 5 much personal wealth, nor does the low asset price boom and high leverage are thus points to over 300 basis points. A steady cash flow pledgeability of the firm allow delayed, not just because debt has to be recovery to normal levels started around them to borrow against future cash flows to written down — and undoubtedly frictions mid-2009. pay for acquiring the firm. The firm’s debt in writing down debt would increase the Bankers alleged a “buyers” strike dur- capacity falls significantly, and it gets into length of the delay — but also because firms ing this period. What they meant, presum- financial distress. Credit spreads rise sub- have to restore the pledgeability of their ably, was that the market was clearing at stantially, and they will such a low price for finan- stay high until the firm cial assets that few wanted raises pledgeability, Measuring Pledgeability: Unqualified Audits to sell if they could hold on. which will take time, Arbitrage opportunities also or industry liquidity Percentage of firms submitting unqualified audits appeared, such as a large nega- comes back up, which 20 tive bond-CDS basis, which could take even longer. Other firms meant that a riskless position The neglect of pledge- paying a positive spread over ability because of high 2.0 3.5 Treasuries could be created, 15 4.5 leverage at the end of a 6.3 4.3 provided one could borrow. 6.1 5.3 4.4 sustained boom makes 4.1 And there was the nub. the recovery difficult No one, including well-capi- and drawn out. 10 Construction firms talized liquid financial institu- tions, seemed willing to lend. Testable There was a sharp rise in cen- Implications 5 tral bank reserves held by com- 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 mercial banks from around The testable impli- the time of the Lehman fail- cations of this model An “unqualified audit” is the best possible outcome from a financial audit ure, suggesting they were Source: Lisowsky, Minnis, and Sutherland, The University of Chicago Booth School of Business Working Paper No. 14-30 are that pledgeability holding cash instead of lend- falls as high liquidity Figure 5 ing. Why was there this flight persists.12 There is a to liquidity? greater fall in industries To explain the extreme that are undergoing booms. Conversely, it cash flows to cope with a world where tightness in credit and the decline in trad- rises over time when liquidity dries up. liquidity is scarcer. It is the need to raise ing, I will appeal to the idea of liquidity Petro Lisowsky, Michael Minnis, and pledgeability that may make the down- Andrew Sutherland examine the percent- turn more prolonged. Higher anticipated age of unqualified (that is, good) audits sub- liquidity in some future states can there- S&P/LSTA U.S. Leveraged Loan 100 Index mitted to bankers by firms in the booming fore induce more eventual misallocation in 13 construction industry before the crisis. As less-liquid states, a spillover effect between Average bid as a percent of par value Figure 5 shows, that percentage fell steadily states that operates through leverage and 100 before the crisis, only to stabilize when the pledgeability! crisis hit. Of course, since other industries 90 were booming, the percentage fell for them The Sharp Onset of the Crisis also. 80 So what we really want to see is the dif- As the crisis hit, illiquidity spread ference between the construction industry through the economy, not just among cor- and others. porate assets where issues like pledgeability 70 As is clear from Figure 5, the gap might be a concern. Liquidity dried up for increased until about 2006, around the time a large set of financial assets, with signifi- 60 monetary policy started getting tight by cant price drops and high bid/ask spreads. the Taylor rule measure or around the time Consider, for example, instruments issued financial conditions started tightening. The on the U.S. Leveraged Loan 100 Index. 50 gap then reversed. In sum, pledgeability did Figure 6 shows that the average bid col- 01/2008 07/2008 01/2009 07/2009 01/2010 fall during the construction industry boom, lapsed from around 90 cents on the dollar to only to reverse course, albeit slowly, as tight just above 60 cents around September 2008 Source: Data from Thomson Reuters liquidity set in. (the Lehman Brothers crisis), while the bid/ Recoveries following periods of an ask spread blew out from about 100 basis Figure 6

6 NBER Reporter • No. 3, September 2018 again, this time the liquidity — or illiquid- One consequence of the prospective in which the depressed asset value recovers ity — of financial institutions identified by fire sale is that it may depress asset values are precisely the states in which the bank Diamond and myself.14 Here is the idea: Let so much that the bank is insolvent. This survives, bank management would much a set of banks which we will term “fragile,” may precipitate a run on the bank, which prefer holding on to the illiquid assets and with substantial short-term liabilities, have a may cause more assets to be unloaded on risking a fire sale and insolvency to selling significant quantity of assets that have a lim- the market, further depressing the price. the asset and ensuring its own stability in ited set of potential buyers. One example of Importantly, the returns to those who have the future. Indeed, the bank would prefer such an asset is a mortgage-backed security excess liquid cash at such times can be to spend its cash to load up more on secu- which, in an environment in which some extraordinarily high. rities that are exposed to the liquidity risk mortgages have defaulted, can be valued Folding back to today, the prospect of a because its private valuation for those secu- accurately only by some specialized firms future fire sale of the bank’s asset can depress rities exceeds the market’s valuation. Fragile such as BlackRock or KKR. Furthermore, the asset’s current value — investors need to banks become “illiquidity seekers.” let us assume that, with some probability, be enticed through a discount to buy the The intuition here is clearly analogous the fragile banks will need to realize cash asset today, otherwise they have an incen- to the risk-shifting motive of Michael Jensen quickly in the future. Such a need for cash tive to hold back because of the prospect of and William Meckling16 and the underin- may stem from unusual demands of the buying the asset cheaper in the future. More vestment motive of Stewart Myers,17 though banks’ customers, who draw on commit- generally, the high returns potentially avail- the bank “shifts” risk or underinvests in our ted lines of credit or on their demandable able in the future to those who hold cash model by refusing to sell an illiquid asset deposits. It may also stem from panic, as can cause them to demand a high return for rather than by taking on, or not taking on, depositors and customers, fearing the bank parting with that cash today. a project. Also, illiquid institutions not only could fail, pull their deposits and accounts However, the elevated required rate act as an overhang over the market, elevat- from the bank. Regardless of where the of return now extends to the entire seg- ing required rates of return, but they also demand for liquidity comes from, it would ment of the financial market that has the risk future insolvency by holding on to the force banks to sell assets or, equivalently, expertise to trade the security. If this seg- assets, further elevating required returns. raise money quickly at a future date. Given ment also accounts for a significant frac- Thus, there is an inherent source of adverse that the potential buyers for the bank’s tion of the funding for potential new loans, feedback in any financial crisis, which is why assets, like BlackRock and KKR, have lim- the elevated required rate of return will be cleaning up the financial systemex post may ited resources and will drive a hard bargain, contagious and also will depress lending. be an important contributor to recovery. At the asset would have to be sold at fire sale Moreover, the overhang of fragile banks will the same time, ex ante regulation to prevent prices (Shleifer and Vishny, and Franklin affect lending not only by distressed banks an excessive buildup of exposure to liquidity Allen and Douglas Gale).15 but also by healthy potential lenders, a fea- risk may also be warranted. ture that distinguishes this explanation from In sum then, this model can explain those where the reluctance to lend is based both the trading freeze as well as the credit S&P/LSTA U.S. Leveraged Loan 100 Index on the poor health of either the bank or its freeze that occurred after the Lehman deba- potential borrowers. Note that the adverse cle. The market feared that there were many Average bid as a percent of par value effect of prospective future illiquidity on banks with hard-to-sell assets, and these 100 current lending is absent in models where assets would be dumped on the market if future asset values are low for other reasons, these fragile banks had to meet demands for 90 such as reduced future payoffs. In such cases, liquidity. Hoping no such liquidity demand low asset values do not lead to an elevated would materialize, banks held on to the rate of return to buyers. hard-to-sell assets, for these would gener- 80 More surprising though, the fragile ate high returns under those conditions. bank’s management, knowing that the bank Anticipating that such liquidity would be 70 could fail in some states in the future, does demanded with high probability, thus forc- not have strong incentives to sell the illiquid ing fire sales by fragile banks, healthy finan- 60 asset today, even though such sales could cial firms held on to cash so that they could save the bank — sales of the asset subject to put it to work at that time when returns potential fire sales dry up. The reason is sim- would be high. Credit therefore became 50 ple. By selling the asset today, the bank will scarce, thus opening up arbitrage opportu- 01/2008 07/2008 01/2009 07/2009 01/2010 raise cash that will bolster the value of its nities wherever borrowing was needed to outstanding debt by making it safer. But in close them. Source: Data from Thomson Reuters doing so, the bank will sacrifice the returns Finally, what explains the recovery in that it would get if the currently depressed asset prices, the decline in spreads, and Figure 6 value of the asset recovers. Since the states the disappearance of arbitrage opportuni-

NBER Reporter • No. 3, September 2018 7 U.S. Unemployment and Mortgage Delinquency Rates

Unemployment rate (%) Mortgage delinquency rate (%) 11 12 NBER dated NBER dated recession recession 10 10

9 8

8 6 7 4 6

2 5

4 0 01/2005 01/2006 01/2007 01/2008 01/2009 01/2010 01/2011 01/2012 01/2004 01/2006 01/2008 01/2010 01/2012 01/2014 01/2016

Source: Data from the U.S. Bureau of Labor Statistics and the Board of Governors of the Federal Reserve System

Figure 7 ties? Was it good news on fundamen- Moreover, capital was set to be raised so returns from purchasing in potential fire tals? While the NBER dated the recovery that bank assets would not shrink. Finally, sales or holding on to illiquid assets, thus from June 2009, that dating happens only the Treasury backstopped banks that allowing trading and lending to resume. much later. As seen in Figure 7, unem- could not raise capital with its Capital Bid/ask spreads narrowed, asset prices ployment stabilized only in 2010, and Assistance Program. recovered, and arbitrage opportunities the first month without job losses was By forcing banks to become healthy, dwindled. November 2009. It is hard to imagine that and implicitly guaranteeing they would In sum then, the lesson to take away it was well known that the recovery was be, the stress tests effectively removed from this model is that anticipated illi- underway from mid-2009. the overhang of potentially insolvent quidity can lead to frozen markets and Was it good news on mortgages? As or highly illiquid banks, thus reducing credit. The authorities may need to clean seen in Figure 7 on the up a system even in next page, delinquencies the midst of a crisis started declining steadily U.S. ‘Covenant-Lite’ Leveraged Loans, 2004–2017 in order to restore only in mid-2012. trading and lending. Why then did Total covenant-lite loans outstanding (billions of dollars) Percentage of leveraged loans that are covenant-lite So liquidity infusion markets start return- 300 60 into the markets and ing to normal around 250 50 capital infusion into mid-2009? Arguably, it specific institutions was a sequence of Fed 200 40 may be necessary to emergency programs stabilize the financial but especially the stress 150 30 system. Of course, if tests conducted by a little liquidity infu- the Federal Reserve in 100 20 sion is good, why not March 2009, with the do more, and more results announced in 50 10 permanently? This

May 2009, which were 0 0 seems to be the les- responsible. The regula- Q1 2005 Q1 2007 Q1 2009 Q1 2011 Q1 2013 Q1 2015 Q1 2017 son financial authori- tors examined 19 banks, ties have drawn. and 10 were asked to “Covenant-lite loans” do not contain typical protective covenants that benefit lenders If we go back to raise capital. The details Source: Data from S&P Global Market Intelligence Figure 1, financial of each bank’s examina- conditions across the tion were made public. Figure 8 world are now again

8 NBER Reporter • No. 3, September 2018 as easy as they have ever been. But as 3 C. Borio and P. Lowe, “Asset Prices, Moral Hazard, and Liquidity,” we saw from the first model, too much Financial and Monetary Stability: Journal of Finance, 66(1), 2011, liquidity can also be bad, for it induces Exploring the Nexus,” BIS Working Paper pp. 99–138; J. Dow, G. Gorton, and leverage and causes financial sector partic- No. 114, 2002. A. Krishnamurthy, “Equilibrium ipants to effectively neglect risks. Indeed, Return to Text Investment and Asset Prices Under if we look at the volume of covenant- 4 A. Krishnamurthy and T. Muir, “How Imperfect Corporate Control,” American lite loans today in Figure 8, it dwarfs the Credit Cycles Across a Financial Crisis,” Economic Review, 95(3), 2005, quantity before the global financial crisis. Stanford Graduate School of Business pp. 659–81; and A. Eisfeldt and A. The bottom line is that both too little, Working Paper, 2017. Rampini, “Managerial Incentives, as well as too much, anticipated liquidity Return to Text Capital Reallocation, and the Business can be problematic for financial stability. 5 D. López-Salido, J. Stein, and E. Cycle,” Journal of Financial Economics, To the extent that accommodative financ- Zakrajšek, “Credit-Market Sentiment 87(1), 2008, pp. 177–99. ing conditions (i.e., easy liquidity) are and the Business Cycle,” Quarterly Return to Text caused by accommodative monetary pol- Journal of Economics, 132(3), 2017, pp. 13 P. Lisowsky, M. Minnis, and A. icy, it suggests monetary policy and finan- 1373–1426. Sutherland, “Economic Growth and cial stability cannot be separated. Return to Text Financial Statement Verification,” How should monetary policy take 6 A. Mian, A. Sufi, and E. Verner, Journal of Accounting Research, financial stability into account? This, to “Household Debt and Business Cycles 55(4), 2017, pp. 745–94. my mind, is the huge unaddressed issue Worldwide,” Quarterly Journal of Return to Text since the crisis, though some papers — for Economics, 132(4), 2017, pp. 1755– 14 D. Diamond and R. Rajan, “Fear of example, one by Diamond and myself,18 1817. Fire Sales, Illiquidity Seeking, and the and another by Emmanuel Farhi and Jean Return to Text Credit Freeze,” Quarterly Journal of Tirole19 — have made beginnings. Today, 7 R. Rajan, “Why Bank Credit Economics, 126(2), 2011, pp. 557–91. liquidity is slowly in the process of being Policies Fluctuate: A Theory and Return to Text withdrawn. Will we have better or worse Some Evidence,” Quarterly Journal of 15 F. Allen and D. Gale, “Limited outcomes than the previous time liquid- Economics, 109(2), 1994, pp. 399–441. Market Participation and Volatility ity was withdrawn? I don’t know, though Return to Text of Asset Prices,” American Economic it is clear we will have some stress in pock- 8 N. Gennaioli, A. Shleifer, and R. Review, 84(4), 1994, pp. 933–55. ets where leverage has built up as easy Vishny, “Neglected Risks: The Psychology Return to Text financial conditions change. We have to of Financial Crises,” American Economic 16 M. Jensen and W. Meckling, see whether, this time around, it is indeed Review, 105(5), 2015, pp. 310–14. “Theory of the Firm: Managerial different. Return to Text Behavior, Agency Costs, and Ownership 9 J. Geanakoplos, “The Leverage Cycle,” Structure,” Journal of Financial NBER Macroeconomics Annual 2009, Economics, 3(4), 1976, pp. 305–60. 1 The Financial Conditions Index, 24(1), 2010, pp. 1–65, http://www.nber. Return to Text presented in the IMF’s Global Financial org/chapters/c11786.pdf. 17 S. Myers, “Determinants of Corporate Stability Report, is a composite of short- Return to Text Borrowing,” Journal of Financial term rates, long-term real rates, term 10 D. Diamond, Y. Hu, and R. Rajan, Economics, 5(2), 1977, pp. 147–75. spread, corporate spread, interbank “Pledgeability, Industry Liquidity, and Return to Text spread, equity price growth, equity return Financing Cycles,” NBER Working Paper 18 D. Diamond and R. Rajan, “Illiquid volatility, credit to GDP, credit growth, No. 23055, January 2017. Banks, Financial Stability, and Interest and house price growth. Return to Text Rate Policy,” Journal of Political Return to Text 11 A. Shleifer and R. Vishny, Economy, 120(3), pp. 552–9, https:// 2 A. Maddaloni and J. Peydró, “Bank “Liquidation Values and Debt Capacity: doi.org/10.1086/666669 Risk-Taking, Securitization, Supervision, A Market Equilibrium Approach,” Return to Text and Low Interest Rates: Evidence from Journal of Finance, 47(4), 1992, pp. 19 E. Farhi and J. Tirole, “Collective the Euro Area and the U.S. Lending 1343–66. Moral Hazard, Maturity Mismatch, Standards,” European Central Bank Return to Text and Systemic Bailouts,” American Working Paper No. 1248, 2010, pp. 12 For the interested reader, related Economic Review, 102(1), 2012, pp. 2121–65. recent papers include: V. Acharya 60–93. Return to Text and S. Viswanathan, “Leverage, Return to Text

NBER Reporter • No. 3, September 2018 9 Research Summaries

Environment, Energy, and Unintended Consequences

Matthew J. Kotchen

Economists are fascinated with based on joint production of a private unintended consequences. A policy good and a public environmental good.1 designed to accomplish a particular The purchase of electricity from renew- objective will sometimes have the oppo- able sources of energy provides an exam- site effect, or create new problems apart ple. While green electricity may cost from the one it originally sought to cor- more than electricity generated from Matthew Kotchen is a professor of rect. Well-intentioned individuals will fossil fuels, it produces the joint prod- economics at Yale University, with a pri- sometimes make choices that are coun- ucts of electricity (a private good) and mary appointment in the Yale School of terproductive to the very causes they lower emissions (a public good). Forestry & Environmental Studies and seek to support. Understanding the full Does this mean green products are secondary appointments in the School impact of policy interventions and indi- always beneficial for the environment? of Management and the Department of vidual choices is critical for the design, The answer turns out to depend on Economics. His research interests lie at the implementation, and improvement of whether there are opportunities to pro- intersection of environmental and public more effective and efficient policies. vide the public good separately. It is pos- economics and policy. Much of my research over the last sible, for example, that one’s purchase of Kotchen is a research associate in the 15 years has focused on unintended con- green electricity crowds out other activi- NBER Programs on Environmental and sequences in the field of environmental ties that reduce emissions. In such cases, Energy Economics and Public Economics. and energy economics and policy. The introducing a green good can counterin- He has held previous and visiting posi- starting point is often a simple question: tuitively increase pollution and reduce tions at Williams College, the University Does a specific policy or choice that economic welfare. of California, Santa Barbara and Berkeley, is driven by concern for environmen- In a subsequent theoretical paper, I Stanford University, and Resources for the tal protection or energy conservation consider joint production that is instead Future. He has also served as an associ- deliver on its promise, and if not, why based on the private provision of a pub- ate dean of academic affairs at Yale, as not? Research attempting to answer this lic “bad.”2 The setup more closely aligns deputy assistant secretary for environ- question has led to contributions to eco- with the way economists typically think ment and energy at the U.S. Department nomic theory on the private provision about particular goods and services gen- of the Treasury, and as a member of the of public goods and to empirical studies erating a negative externality. A novel Environmental Economics Advisory on a range of topics such as renewable feature of the model is the way that Committee of the U.S. Environmental energy, corporate social responsibility, consumers can make donations that are Protection Agency. daylight saving time, building codes, and motivated, in part, to offset the negative Kotchen leads a new NBER initia- electric cars. externality. In this context, I show how tive comprising an annual conference in donations and economic welfare differ Washington, DC, and an annual publi- Private Provision of from the standard model for privately cation, both titled Environmental and Environmental Public Goods provided public goods. Energy Policy and the Economy. Funded by One general result is that dona- the Alfred P. Sloan Foundation, this is an Many individuals are concerned tions continue to increase, rather effort to stimulate policy-relevant research with the environmental impact of their than decrease, as an economy grows. in the field and to disseminate this research consumption choices, and these con- Moreover, an unintended consequence directly to policymakers. It is modeled after cerns have driven the emergence of mar- of this market arrangement is that the the NBER initiatives Tax Policy and the kets for environmentally friendly goods opportunity to make offsetting dona- Economy and Innovation Policy and the and services. My first theoretical con- tions will typically stimulate demand for Economy. tribution was to model “green goods,” the externality-causing good. For exam-

10 NBER Reporter • No. 3, September 2018 ple, the ability to purchase a carbon offset intervening mechanism, whereby compa- applied energy policies. Implemented as a might help an individual justify the pur- nies might pursue CSR strategies to off- conservation measure in both world wars, chase of a less fuel-efficient car. Indeed, set corporate social irresponsibility (CSI). DST has a long and fascinating history. the theory provides a framework for Jon Jungbien Moon and I disaggre- Indeed, Benjamin Franklin produced an understanding markets for environmental gated one of the widely used indices for early economic analysis of DST, showing offsets, with those that promote carbon CSR into separate measures of CSR and how much tallow and candles could be neutrality in response to climate change CSI across seven dimensions, including saved if clocks were changed to encour- being an increasingly salient example. corporate governance, community rela- age early rising during long summer days, tions, human rights, and the environ- when people could take greater advantage Offsetting Goods and Bads ment.4 Analyzing data on more than of natural daylight. The same argument is 3,000 publicly traded companies over still used today to justify DST as energy Does giving consumers a way to pay 14 years, we found that CSI is a signifi- policy in the United States, yet surpris- for their “sins of emissions” help justify an cant predictor of CSR, both overall and ingly little analysis on the subject has increase in polluting activities? Along with within specific dimensions. For exam- occurred since Franklin’s day. collaborators Grant Jacobsen and Mike ple, when a company is responsible for Taking advantage of a natural experi- Vandenbergh, I set out ment that occurred in to investigate whether Indiana, Laura Grant such behavior occurs.3 The E ect of Daylight Saving Time on Electricity Use and I estimated the We obtained electric- effect of DST on elec- 5 ity billing data for resi- Percentage change in monthly residential electricity consumption due to daylight savings time tricity consumption. dential households in 4 In 2006, the state Tennessee before and Period of daylight saving time switched to DST while after a utility company 3 simultaneously shift- introduced a volun- 2 ing some of its coun- tary green-electricity ties to a different time program. A key feature 1 zone. The combination of the program was of these two policies that households could 0 provided treatment choose to participate and control groups at different levels in -1 that allowed us to support of new wind compare differences in -2 and solar generation Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec. residential electricity intended to offset the consumption before emissions associated Light-blue bars represent 95% confidence intervals and after the pol- with their own electric- Observations for April and November are excluded due to partial exposure to DST during those months icy change. We found ity consumption. Source: M. Kotchen and L. Grant, NBER Working Paper No. 14429 that — contrary to the We found that Figure 1 policy’s intent — DST households participat- increased electricity ing above the minimum threshold had an environmental accident, it compen- consumption [Figure 1]. While Franklin’s no change in electricity consumption, but sates by undertaking pro-environmen- conjecture about the demand for lighting those participating only at the minimum tal actions. When it comes to corpo- holds up, modern-day demand for heat- level — representing a “buy-in” mental- rate governance, however, the findings ing and cooling differs across hours of the ity — increased their electricity consump- are more nuanced: After an event that day, and the shift to DST increased both. tion by 2.5 percent. We thus identified reflects poorly on corporate governance, Building codes are another ubiqui- some of the first evidence on the behav- companies tend to compensate in nearly tous form of energy policy. The regula- ioral response to undertaking a pro-envi- all dimensions of CSR except for reform- tion of building practices first focused on ronmental action. ing governance itself. energy for purposes of national security I then became interested in knowing in the wake of the Arab oil embargo in whether a similar phenomenon was taking Saving Energy and Reducing the 1970s. Today, building energy codes place within corporations. The existing Pollution — Or Not across the United States and other coun- literature on corporate social responsibil- tries are motivated by concerns about ity (CSR) tends to focus on the relation- Many people are surprised to hear energy efficiency and climate change. ship between CSR and financial perfor- that daylight saving time (DST) is one Until recently, however, engineering sim- mance. I was curious about a potential of the more longstanding and universally ulations provided the only evidence on

NBER Reporter • No. 3, September 2018 11 their effectiveness. This and their environmen- led Jacobsen and me to Variation in Emissions Produced to Charge an Electric Car tal impacts depend on search for an opportu- a comparison of emis- Pounds of COƒ emissions produced by an electricity-generating power plant nity to provide an eval- needed to power a plug-in electric vehicle for 35 miles sions at tailpipes versus 35 uation that accounted 1–4 5–8 9–12 1–4 5–8 9–12 power plants. for actual construc- 30 am pm Joshua Graff- tion practices and the 25 Zivin, Erin Mansur, behavior of household and I developed a residents. 20 method for estimating We found one in 15 marginal emissions of

Florida, where the state 10 electricity generation increased the strin- at different locations gency of its building 5 and times of day across 9 energy code in 2002. 0 the United States. WECC ERCOT FRCC MRO NPCC RFC SERC SPP Average We obtained a unique Regions (see Figure 3 for reference) While previous studies dataset that included either relied on simula- detailed information “Regions” are various interconnected power grids overseen by the North American Electric Reliability Corporation tion estimates or aver- on the characteristics Source: J. Gra Zivin, M. Kotchen, and E. Mansur, NBER Working Paper No. 18462 age — rather than mar- of residential dwell- ginal — emissions, our ings and monthly bill- Figure 2 approach is based on ing data for electricity hourly load and emis- and natural gas. This enabled us to com- more about the effectiveness and effi- sions data across different interconnec- pare energy consumption of observation- ciency of building energy codes. tions of the electricity grid [Figures 2 and ally similar residences built just before We are also beginning to learn more 3]. and after the building code change. We about the potential of electric cars to The results can be used to estimate found significant decreases in both elec- lower demand for energy and reduce pol- the effect on CO2 emissions from any tricity and natural gas consumption, and lution. Generous subsidies at the state electricity-shifting policy; we focused on estimated the private payback period to and federal level, along with the extraor- increased demand to charge electric cars. be approximately six years.6 dinary market valuation of Tesla, signal We found considerable differences in the Yet subsequent research by Arik high confidence in the future benefits emissions based on geographic location Levinson, studying data from California, and scale of the electric car market. But and hours of the day. The heterogeneity is raised questions about whether the often missing from future visions is that driven by the fact that electricity is gener- energy saving effects would endure over charging electric cars also requires energy, ated in different ways, mostly from coal or the long run.7 This natural gas, at different spurred a reevaluation locations and at peak of our Florida find- Regional Powergrids versus off-peak times of ings over a longer time Interconnected power grids overseen by the North American Electric Reliability Corporation day. Notably, we found period when addi- MRO that in many Upper tional data were avail- RFC Midwestern states, an able. The results indi- NPCC electric car generates cated that after five or more CO2 emissions six years, electricity than the average econ- savings were no longer omy car. The research evident, while the nat- showed that the future ural gas savings persist- WECC environmental prom- ed.8 Questions about SERC ise of electric cars the underlying mecha- depends critically on nism and generalizabil- FRCC how electricity is gen- ity of these short-run SPP erated on the grid. and long-run effects ERCOT Subsequent research remain, but the num- Intended for illustrative purposes. Only represents the U.S. portion of North American interconnections. has also shown the ber of papers appearing Source: North American Electric Reliability Corporation importance of consid- on the subject suggests ering the health effects that we will soon learn Figure 3 of local pollution.10

12 NBER Reporter • No. 3, September 2018 Looking Ahead 16608, December 2010, and European 135–53. Economic Review, 56(5), 2012, pp. Return to Text To conclude, I must admit that the 946–60. 9 J. Graff-Zivin, M. J. Kotchen, and pattern in my research of identifying and Return to Text E. T. Mansur, “Spatial and Temporal estimating unintended consequences is 4 M. J. Kotchen and J. J. Moon, Heterogeneity of Marginal Emissions: itself an unintended consequence, the “Corporate Social Responsibility for Implications for Electric Cars and Other result of opportunistically pursuing Irresponsibility,” NBER Working Paper Electricity-Shifting Policies,” NBER research questions without preconceived No. 17254, July 2011, and The B.E. Working Paper No. 18462, October notions. While I think that uncovering Journal of Economic Analysis & Policy 2012, and Journal of Economic unexpected and sometimes counterintui- (Contributions), 12(1), 2012. Behavior & Organization, 107, 2014, tive findings is important, the growing set Return to Text pp. 248–68. of environmental and energy challenges 5 M. J. Kotchen and L. E. Grant, “Does Return to Text also requires economic research with a Daylight Saving Time Save Energy? 10 S. P. Holland, E. T. Mansur, N. Z. directly constructive agenda. Fortunately, Evidence from a Natural Experiment Muller, and A. J. Yates, “Environmental many in the field are doing precisely this. in Indiana,” NBER Working Paper No. Benefits from Driving Electric Vehicles?” A few recent and selected examples of my 14429, October 2008, and The Review NBER Working Paper No. 21291, own efforts with such a goal include using of Economics and Statistics, 93(4), June 2015, and published as “Are There revealed preferences to test among mod- 2011, pp. 1172–85. Economic Benefits From Driving Electric els for charitable giving to environmental Return to Text Vehicles? The Importance of Local causes,11 drawing insights about national 6 G. D. Jacobsen and M. J. Kotchen, Factors,” American Economic Review, and international climate policy from a “Are Building Codes Effective at Saving 106(12), 2016, pp. 3700–29. public goods framework,12 and develop- Energy? Evidence from Residential Return to Text ing new ways to think about long-term Billing Data in Florida,” NBER 11 R. Deb, R. S. Gazzale, and M. J. and intergenerational social discount Working Paper No. 16194, July 2010, Kotchen, “Testing Motives for Charitable rates.13 and The Review of Economics and Giving: A Revealed-Preference Statistics, 95(1), 2013, pp. 34–49. Methodology with Experimental Return to Text Evidence,” NBER Working Paper No. 1 M. J. Kotchen, “Green Markets and 7 A. Levinson, “How Much Energy Do 18029, May 2012, and Journal of Private Provision of Public Goods,” Building Energy Codes Really Save? Public Economics, 120, 2014, pp. 181– Journal of Political Economy, 114, Evidence from California,” NBER 92. 2006. pp. 816–34. Working Paper No. 20797, December Return to Text Return to Text 2014, and American Economic Review, 12 M. J. Kotchen, “Which Social Cost 2 M. J. Kotchen, “Voluntary Provision 106(10), 2016, pp. 2867–94. of Carbon? A Theoretical Perspective,” of Public Goods for Bads: A Theory of Return to Text NBER Working Paper No. 22246, May Environmental Offsets,” NBER Working 8 M. J. Kotchen, “Do Building Energy 2016, and Journal of the Association Paper No. 13643, November 2007, and Codes Have a Lasting Effect on Energy of Environmental and Resource The Economic Journal, 119(537), Consumption? New Evidence From Economists, 5(3), 2018, pp. 673–94. 2009, pp. 883–99. Residential Billing Data in Florida,” Return to Text Return to Text NBER Working Paper No. 21398, 13 E. P. Fenichel, M. J. Kotchen, and E. 3 G. D. Jacobsen, M. J. Kotchen, and July 2015, and published as “Longer- T. Addicott, “Even the Representative M. P. Vandenbergh, “The Behavioral Run Evidence on Whether Building Agent Must Die: Using Demographics Response to Voluntary Provision of Energy Codes Reduce Residential to Inform Long-Term Social Discount an Environmental Public Good: Energy Consumption,” Journal of the Rates,” NBER Working Paper No. Evidence from Residential Electricity Association of Environmental and 23591, July 2017. Demand,” NBER Working Paper No. Resource Economists, 4(1), 2017, pp. Return to Text

NBER Reporter • No. 3, September 2018 13 Taxation and Innovation

Ufuk Akcigit and Stefanie Stantcheva

Innovation is the source of tech- financial incentives and only strive for nological progress and, ultimately, intellectual achievement. the main driver of long-run eco- Related questions are whether nomic growth. In recent work with taxes impact the quality of innova- several co-authors, we have shown tion, where inventors decide to locate, Ufuk Akcigit is an associate pro- that the U.S. states that produced the and what firms they work for. In addi- fessor of economics with tenure at the most innovations also grew fastest tion, there is a question of whether University of Chicago and a research over the 100-year period from 1900 taxes influence where companies allo- associate at the NBER, where he is affil- to 2000.1 We also have documented cate R&D resources and how many iated with the Productivity, Innovation, that innovation is strongly associated researchers they employ. and Entrepreneurship program. He with social mobility. U.S. regions that Answers to these questions are also is affiliated with the Centre for experienced more innovation also still lacking, and there is a scarcity of Economic Policy Research. witnessed much stronger intergenera- empirical evidence. The gap in our Akcigit’s research focuses on eco- tional and social mobility, especially understanding is especially large when nomic growth, productivity, firm dynam- when innovations were attributable it comes to the effects of tax policy on ics, and the economics of innovation. His to new entrant firms. Innovation also technological development over the work aims to uncover the sources of tech- correlates strongly with top income long run. nological progress and innovation that inequality, but not so much with mea- serve as engines of long-run economic sures of inequality such as the Gini or Theory and Empirics of growth. He is also interested in the role the 90/10 ratio, and is associated with Taxation and Innovation of public policy in growth, with a focus greater well-being across the United on environmental regulations, public States.2,3 There are two complementary research and funding for universities, and Given all the important conse- dimensions along which to think industrial policies such as R&D tax cred- quences of innovation, it is essen- about the interplay between taxation its and corporate taxation. tial to understand how public poli- and innovation. First, taxation on per- Some of his current papers explore cies impact innovation in the United sonal or corporate income or wealth the impact of trade and foreign compe- States and across the world. Our joint may affect innovation. This may be tition on innovation, the role of politi- research agenda explores the interplay an unwelcome byproduct of taxes cal connections in reducing innovation, between taxation and innovation. that are set for completely unrelated and the social origins and family back- Major changes in U.S. tax policy, goals, such as to raise revenues. Thus, grounds of inventors. such as those in the Tax Cuts and Jobs reduced innovation could be one of His research has been published Act of 2017, raise questions about the efficiency costs of taxation; this in leading economics journals and has whether higher taxes stifle growth, could affect the assessment of optimal been supported by the Ewing Marion productivity, and innovation. taxes, since the elasticity of innova- Kauffman Foundation, the Alfred P. If innovation, like many other eco- tion with respect to taxes would influ- Sloan Foundation, and the National nomic outcomes, is the result of inten- ence the elasticities that enter into the Science Foundation. He is the recipi- tional effort and investment, then optimal tax formulas.4,5 This under- ent of a CAREER Award from the higher taxes will reduce the expected scores the importance of quantifying National Science Foundation. net return to these inputs and lead to the elasticity of innovation to taxa- Akcigit holds a BA in economics less innovation. Yet for at least some tion along all the relevant margins. from Koç University in Istanbul and path-breaking superstar inventors Second, tax policy could be designed a PhD in economics from MIT. Prior from history, such as Thomas Edison, intentionally so as not to hurt, or even to joining the University of Chicago Alexander Graham Bell, and Nikola to stimulate, innovation. in 2015, he was an assistant profes- Tesla, the picture that comes to mind Our research agenda on taxa- sor of economics at the University of is one of hard-working, enthusiastic tion and innovation seeks to under- Pennsylvania. scientists who are unconcerned with stand and quantify the effects of

14 NBER Reporter • No. 3, September 2018 taxation — of personal income, cor- the quantity of innovation, as cap- porate income, and wealth — on tured by the number of patents; the innovation by firms and individu- quality of innovation, as measured als. How do taxes shape all these by patent citations, and the share of agents’ choices leading up to innova- patents assigned to companies rather tions? Our empirical studies are based than individuals at both the state on modern-day data — European pat- level and individual-inventor level. ent office data since 1975, for exam- We also consider the location choices ple — and on long-run historical data, of firms and inventors, including such as the universe of U.S. inven- superstar inventors, as well as the cre- tors since 1836. Theoretically and ation of path-breaking, highly cited quantitatively, we study the design inventions. of decentralized innovation policies: We employ several identification combinations of taxes, tax credits, and strategies, and find consistent results Stefanie Stantcheva is a professor subsidies that can make agents inter- across the different approaches. First, of economics at Harvard University nalize the spillovers from innovations we control for state, year, and, at the and a research associate at the NBER, and foster innovation. We illustrate individual level, inventor-fixed effects, where she is affiliated with the Public our research approach by focusing on and include individual or state-level Economics and Political Economy pro- three distinct studies. time-varying controls in our specifi- grams. She is a research fellow at the cation. These go a long way toward Centre for Economic Policy Research. Taxation and Innovation absorbing unobserved heterogeneity. Stantcheva’s research focuses on the in the 20th Century In addition, we exploit tax schedule optimal design of tax systems and pub- differences across individuals within lic policies, taking into account labor Although the United States expe- a given state-year, due to tax progres- market features, complex social pref- rienced major changes in its tax code sivity, and compare individuals in dif- erences, and long-term effects such as throughout the 20th century, we cur- ferent tax brackets. Thus, we can also human capital acquisition and innova- rently do not know how these tax include state-times-year fixed effects tion by individuals and firms. Part of changes influenced innovation at to filter out other policy variations her work is centered around the study either the individual or corporate or confounding economic circum- of empirical effects of taxation of indi- level. This challenging question has stances. Second, at both the macro viduals and firms, on inequality, top largely gone unanswered because of and micro levels, we use an instru- incomes, migration, human capital, and a lack of long-run systematic data on mental variable strategy that consists innovation. innovation in the United States and of predicting the total tax burden fac- In recent work, she is exploring the the difficulty of identifying the effects ing a firm or inventor — a composite interplay between taxation and innova- of taxes. We leverage three new data- of state and federal taxes — with the tion using historical as well as modern- sets, which we constructed from his- changes in the federal tax rate only, day data and the optimal design of R&D torical data sources, to explore these holding the state taxes fixed at some policies. She is also studying how people issues.6 The datasets are a panel of the past level. This provides variation that form social preferences regarding redis- universe of U.S. inventors since 1920 is only driven by federal-level changes tribution, using large-scale online sur- and their associated patents, citations, and, thus, exogenous to any individual vey and experiment tools. and firms; a panel of all R&D labs in state. Third, we use a border county Stantcheva is the recipient of a the United States since 1921, matched strategy as a stand-alone and in com- CAREER Award from the National to their patents and with data on their bination with our instrumental vari- Science Foundation and is a Sloan research employment levels and loca- able. Finally, we study specific, sharp, Research Fellow. tions; and a historical state-level cor- tax-change episodes. She holds a BA in economics from porate and personal income tax data- We find that taxation of both the University of Cambridge, a Master’s base.7 This unique combination of corporate and personal income neg- degree in economics and finance from data allows us to systematically study atively affects the quantity, quality, Ecole Polytechnique, a Master’s in the effects of both personal and cor- and location of innovation at the economics from the Paris School of porate income taxation since 1920 on state level and the individual inven- Economics, and a PhD in economics the micro level of individual inventors tor and firm levels.8 The elasticities from MIT. She was a junior fellow at and individual firms that do R&D and of all these innovation outcomes with the Harvard Society of Fellows from on innovation at the macro state level. respect to taxes are relatively large, 2014–16, before joining the Harvard Our innovation outcomes include especially at the macro level, where faculty in 2016.

NBER Reporter • No. 3, September 2018 15 cross-state spillovers and extensive margin responses add to the micro U.S. State Marginal Tax Rate and Patent Production elasticities. Figure 1 illustrates the negative correlation between the per- Log patents sonal income tax at the 90th income 0.2 percentile and the log of patents in a state. 0.1 We also find that corporate inven- tors are more elastic with respect to 0 personal and corporate income tax- ation than non-corporate inventors. Agglomeration effects appear to mat- -0.1 ter as well: Inventors are less sensitive to taxation in places where there is -0.2 already more innovation done in their technological field. -0.3 -4 -2 0 2 4 The International Mobility 90th percentile-earner personal-income marginal tax rate (%) of Superstar Inventors in Both patents and tax rates are reported net of control variables Response to Taxation Source: U. Akcigit, J. Grigsby, T. Nicholas, and S. Stantcheva, NBER Working paper No. 24982 There is a long-standing debate about whether higher top tax rates Figure 1 will cause a “brain drain” of high- one major challenge that arises when quality are comparable enough to be income and high-skill economic studying migration responses to taxes, similarly affected by country-year level agents. In fact, many of the great namely, to model the counterfactual policies and economic developments; inventors were international immi- payoff that an inventor would get in but only those inventors in the top grants: Alexander Graham Bell, inven- each potential location, thanks to a set bracket are directly affected by top tor of the telephone and founder of of detailed controls that come from taxes. Hence, the lower-quality top the Bell Telephone Company; James the patent data, most notably, mea- 5–10 percent, top 10–25 percent, and Kraft, inventor of a pasteurization sures of an inventor’s quality based on below top 25 percent groups serve as technique and founder of Kraft Foods; past citations. control groups for the top 1 percent and Ralph Baer, creator of a TV gam- Our measure of the effects of the group. ing unit that launched the video game top tax rate filters out all country-year Figure 2 provides some prelimi- industry, are examples. level variation and exploits the differ- nary visual evidence of the effects of Inventors are frequently more ential impacts of the top tax rate on taxes. It shows how the number of mobile than other high-skill individu- inventors at different points in the superstar top 1 percent foreign inven- als, and they carry and transmit their income distribution within a country- tors in the U.S. increased after the Tax valuable knowledge and expertise to year cell. To implement this strategy, Reform Act of 1986 relative to a coun- others, which makes them important we define superstar inventors as those terfactual path estimated from a syn- for both new knowledge creation and in the top 1 percent of the quality dis- thetic control country. for its diffusion. Yet little is known tribution, and similarly construct the Overall, we find that superstar about the international mobility of top 1–5 percent, the top 5–10 per- inventors’ location choices are signif- labor in response to taxation. Rigorous cent, and subsequent quality brack- icantly affected by top tax rates. The evidence is lacking because of a scar- ets. We know from other research that elasticity to the net-of-tax rate of the city of international panel data. inventor quality is strongly correlated number of domestic superstar inven- We use a unique type of interna- with income and that top 1 percent tors is around 0.03, while that of for- tional panel data on inventors from the inventors rank very high in the top tax eign superstar inventors is around 1. European and U.S. patent offices and bracket. The probability of being in These elasticities are larger for inven- from the Patent Cooperation Treaty the top bracket and the fraction of an tors who work for multinational com- to study the international migration inventor’s income in the top bracket panies. Inventors are less sensitive to responses of superstar inventors to declines as one moves down the qual- taxes in a country if their company top income tax rates for the period ity distribution. Top 1 percent inven- performs a higher share of its research of 1977–2003.9 We are able to tackle tors and those of somewhat lower there, suggesting that the location

16 NBER Reporter • No. 3, September 2018 decision is influenced by the com- cult to predict a firm’s innovation suc- tion spillovers and the correction for pany and by career concerns that may cess, even based on many observables. the monopoly power induced by the dampen the effects of taxes. The government would like to encour- intellectual property rights system that age the best firms, but policies have to emerges from the method of distin- R&D Policy Design work despite the asymmetric informa- guishing good firms from bad ones. tion and unobservable inputs, and need The more complementary observable Countries enact many different, to distinguish productive firms from R&D investment is to firm research often very costly policies designed to less productive ones. productivity, as opposed to being com- foster research and development by To solve this problem, we build on plementary to the unobservable R&D firms. These are motivated by the view new dynamic mechanism design meth- effort, the more rents a firm can extract that there is underinvestment in R&D ods developed in several recent papers if R&D investment is subsidized. This because of the non-internalized spill- and offer a new approach to allow puts a brake on how well the govern- overs that the innovations of one firm for spillovers between agents (here, ment can correct for spillovers and can have on other firms and, ultimately, firms) with asymmetric information.11 monopoly distortions. On the other on society. Yet, there is no consensus on We then estimate the model and use hand, if R&D investments are more how such policies should be designed. it to simulate a range of policies for complementary to unobservable firm We therefore study the joint design firms of different ages, sizes, and pro- R&D effort, the optimal R&D sub- of R&D policies and corporate taxa- ductivities. We use U.S. Patent and sidy will be greater because subsidizing tion.10 The key new elements in our Trademark Office patent data matched the observed input will lead the firm to analysis are the assumptions that firms to Compustat data on publicly traded put in more of the unobservable input are heterogeneous in their ability to firms, as well as the Longitudinal as well. produce innovations, and that this abil- Business Database (LBD) for all firms. The policies that efficiently trade ity is known to the firm, but not to the This allows us to see the observable off these considerations are different government. In addition, while some inputs to innovation, that is, a firm’s from current policies as well as sim- of the inputs into the R&D process are R&D expenses, as well as the outputs of pler policies, such as linear R&D sub- observable (R&D investment), others innovation as captured by the patents sidies and taxes. Nonlinear policies, are unobservable (R&D effort). The and their citations. such as an R&D subsidy that depends returns to these inputs are also stochas- We show that the need to screen on the amount of R&D investment and tic, which makes innovation risky. firms can starkly influence the shape a profit tax that depends on the level These ingredients capture some of of R&D policies and firm taxation. of profits, can come closer to the con- the very real constraints facing policy- The central policy tradeoff is between strained-efficient outcome. makers. For instance, it is very diffi- the Pigouvian correction for innova- Our findings suggest that taxes significantly affect innovation and that they can thus have far-reaching The 1986 Tax Reform Act and Foreign Superstar Inventors in the U.S. consequences on technological prog- ress and growth. If designed properly, Share of foreign superstar inventors in the U.S., normalized to 1 in 1986 the tax system could help foster inno- 2.5 vation by better aligning the incen- Tax reform takes e ect Actual tives of private agents with the social value of innovation. 2 1 U. Akcigit, J. Grigsby, and T. Nicholas, 1.5 “The Rise of American Ingenuity: Innovation and Inventors of the Golden Estimate based on control (synthetic) country Age,” NBER Working Paper No. 23047, 1 January 2017. Return to Text 2 0.5 P. Aghion, U. Akcigit, A. Bergeaud, R. 1982 1984 1986 1988 1990 1992 Blundell, and D. Hémous, “Innovation and Top Income Inequality,” NBER Working Paper No. 21247, June 2015, Source: U. Akcigit, S. Baslandze, and S. Stantcheva, NBER Working Paper No. 21024 and forthcoming in The Review of Economic Studies. Figure 2 Return to Text

NBER Reporter • No. 3, September 2018 17 3 P. Aghion, U. Akcigit, A. Deaton, Paper No. 24466, March 2018. 10 U. Akcigit, D. Hanley, and S. and A. Roulet, “Creative Destruction Return to Text Stantcheva, “Optimal Taxation and and Subjective Well-Being,” NBER 7 The authors constructed the corpo- R&D Policies,” NBER Working Paper Working Paper No. 21069, April 2015, rate tax database; the personal income No. 22908, December 2016. and the American Economic Review, tax database was constructed by Jon Return to Text 106(12), 2016, pp. 3869–97. Bakija. [J. Bakija, “Documentation 11 A. Pavan, I. Segal, and J. Toikka, Return to Text for a Comprehensive Historical U.S. “Dynamic Mechanism Design: A 4 E. Saez, “Using Elasticities To Drive Federal and State Income Tax Calculator Myersonian Approach,” Econometrica, Optimal Income Tax Rates,” NBER Program,” Williams College Working 82(2), 2014, pp. 601–53; S. Working Paper No. 7628, March 2000, Paper, 2008.] Stantcheva, “Optimal Taxation and and The Review of Economic Studies, Return to Text Human Capital Policies over the Life 68(1), 2001, pp. 205–29. 8 U. Akcigit, J. Grigsby, T. Nicholas, and Cycle,” NBER Working Paper No. Return to Text S. Stantcheva, “Taxation and Innovation 21207, May 2015, and the Journal of 5 E. Saez and S. Stantcheva, “A in the 20th Century,” NBER Working Political Economy, 125(6), 2017, pp. Simpler Theory of Optimal Capital Paper No. 24982, September 2018. 1931–90; S. Stantcheva, “Learning Taxation,” NBER Working Paper Return to Text and (or) Doing: Human Capital No. 22664, September 2016, and the 9 U. Akcigit, S. Baslandze, and Investments and Optimal Taxation,” Journal of Public Economics, 162, S. Stantcheva,“Taxation and the NBER Working Paper No. 21381, 2018, pp. 120–42. International Migration of Inventors,” July 2015; S. Stantcheva, “Optimal Return to Text NBER Working Paper No. 21024, Income, Education, and Bequest 6 U. Akcigit, S. Caicedo, E. Miguelez, March 2015, and the American Taxes in an Intergenerational Model,” S. Stantcheva, and V. Sterzi , Economic Review, 106(10), 2016, pp. NBER Working Paper No. 21177, “Dancing with the Stars: Innovation 2930–81. May 2015. through Interactions,” NBER Working Return to Text Return to Text

18 NBER Reporter • No. 3, September 2018 The Economics of Drug Development Pricing and Innovation in a Changing Market

Craig Garthwaite

Pricing and competition in phar- innovation. As a result, these parame- maceutical markets is an area of great ters are inherently context-specific, and debate and controversy, much of which as the market for developing and sell- stems from the fact that patent protec- ing pharmaceuticals changes, policy- tion allows firms to charge high prices makers should reevaluate the fundamen- for potentially life-saving treatments. In tals of the tradeoff. For example, factors the absence of patents, other firms would that decrease the costs of developing Craig Garthwaite is an associate be attracted by the large profits earned products — such as less stringent clin- professor of strategy and the direc- by incumbent firms and enter the mar- ical research requirements — or those tor of the Program on Healthcare at ket. Such entry would likely raise current that meaningfully increase potential rev- Northwestern University’s Kellogg period welfare by reducing prices and enues — such as large-scale increases in School of Management. An applied increasing access to valuable medications. prescription drug coverage or the abil- microeconomist, he studies the effects However, society enacts regulations ity to develop products targeting partic- of government policies and social phe- that prohibit this entry in pharmaceu- ularly deadly diseases — could support nomena, focusing on the health and bio- ticals and other intellectual property- shorter or weaker patent protection. In pharmaceutical sectors. Recent work has dependent markets, allowing high price- contrast, factors that increase the dif- focused on private sector effects of the cost margins to exist for a period of time. ficulty and/or length of the develop- Affordable Care Act. In prior work, he Policymakers accept the reduced output ment process — such as targeting dis- has examined the impact of government from higher prices in order to provide eases where demonstrating efficacy is cash assistance programs on health. He appropriate incentives for firms to make more difficult — would support stronger is an NBER research associate and is large fixed-cost investments in new prod- or longer patent protections. affiliated with the bureau’s Health Care ucts. That is, there is an implicit trade- Given the dependence of the devel- Program. off in which some degree of current wel- opment of pharmaceutical products on Garthwaite also studies pricing and fare is sacrificed in order to ensure strong the existing body of scientific knowledge, innovation in the biopharmaceutical incentives for future innovation. scientific advancements likely will affect sector. In this area he has examined the A body of research supports this the optimality of the tradeoff between effect of expanded patent protection tradeoff, showing a robust relationship access today and innovation tomorrow. on pricing in the Indian pharmaceu- between expected profitability of phar- In partnership with various co-authors tical market, the innovation response maceutical products and investment in across a series of papers, I have investi- of United States pharmaceutical firms research and development.1 The trade- gated how changes in the development to increases in demand, and the rela- off is not intended to be permanent. process of pharmaceuticals impact the tionship between health insurance After a period of time, patented products economics of drug development, pricing, expansions and high drug prices. His are meant to face additional competition and innovation. research has appeared in journals such that decreases prices, either through the One strand of research examines as The Q uarterly Journal of Economics, introduction of therapeutic substitutes changes in the ability of firms to cre- American Economic Review, The Review that engage in “brand-brand” competi- ate products targeting small and specific of Economics and Statistics, and Health tion that has a moderate effect on prices, patient populations — products that are Affairs. In 2015, he was named one or from post-patent generic competi- often paired with diagnostic tests indi- of Poet and Quants 40 Best under 40 tion, which drives prices even lower. The cating the product’s likely efficacy in an Business School Professors. degree and nature of the eventual com- individual. Broadly speaking, these types Garthwaite received a BA and petition is dictated by a combination of of drugs are part of the evolving world of a Master’s in Public Policy from the policies and market forces. precision medicine. The ability of firms University of Michigan and his PhD The parameters of the complicated to develop such products is more than in economics from the University of tradeoff between static and dynamic effi- simply a scientific advancement or curi- Maryland. Prior to receiving his PhD, ciencies, such as the length and strength osity. A pharmaceutical market involving he served as director of research for the of patents, are intended to provide the products targeting small patient popula- Employment Policies Institute and in incentives for an optimal amount of tions has vastly different economic fun- other public policy positions.

NBER Reporter • No. 3, September 2018 19 damentals than those that prevailed marily interested in trials that employ geting conditions afflicting relatively when the parameters of our existing biomarkers for the purpose of identify- small numbers of patients. Since the intellectual property system were devel- ing patient populations that are more fixed costs of research and development oped. This mismatch between public (or less) likely to respond to particu- are broadly unrelated to the size of the policy and the current reality of drug lar medications. We therefore exploit potential pool of patients, firms gener- development has implications for both additional information on the role of ally find it difficult to invest profitably optimal policy and firm strategies. the biomarker in a trial to identify in products that create large amounts of In a recent paper, Amitabh those related to products that we define value per patient but treat relatively few Chandra, Ariel Dora Stern, and I exam- as “likely precision medicines” (LPMs). individuals. ine the degree to which the market is Figure 1 depicts the growth in trials Recognizing this fact, many devel- increasingly focusing on R&D activi- for these LPMs over time and shows an oped countries have implemented poli- ties related to precision medicines, and increase in their use across all phases cies that provide additional incentives discuss the economic implications of of clinical development. In particular, for products targeting small patient such a shift in the product mixture.2 there has been a marked increase in populations. In the United States, these We first use data policies took the from the Cortellis form of the Orphan Competitive Precision Medicine Development Trials, 1995–2016 Drug Act (ODA), Intelligence Clinical which provides both Trials Database Pharmaceutical development trials using precision biomarkers (%) research and devel- (Cortellis), which is 10 opment tax credits, Phase 2 compiled by Clarivate 8.6% and allows extended (and formerly by 8 periods of market 7.4% Thomson Reuters) Phase 1 exclusivity for firms 6 and contains all reg- 5.6% developing products istered clinical tri- aimed at conditions als from two dozen 4 afflicting fewer than international clin- Phase 3 200,000 patients. ical trial regis- 2 These two policies tries. Importantly are intended to shift for our purposes, 0 the optimal invest- these data contain 1995 2000 2005 2010 2015 ment threshold for detailed descriptions firms. Passed in 1983, of the trials includ- Source: A. Chandra, C. Garthwaite, and A. D. Stern, NBER Working Paper No. 24026 and forthcoming in E. Berndt, the ODA originally ing the use and spe- D. Goldman, and J. Rowe, eds., Economic Dimensions of Personalized and Precision Medicine, University of Chicago Press relied on firms dem- cific role of any bio- onstrating a lack of markers. At a high Figure 1 economic viability level, a biomarker is for a product, rather “a defined characteristic that is mea- Phase I trials for LPMs in recent years. than a strict population limit. The sured as an indicator of normal biolog- The increasing percentage of LPMs 200,000-patient limit was added in ical processes, pathogenic processes, or in pharmaceutical development has 1984 and, according to the Department responses to an exposure or interven- clear economic ramifications. In par- of Health and Human Services, was tion, including therapeutic interven- ticular, the ability to create these prod- arbitrarily based on the prevalence tions,” — that is, measurable features of ucts changes optimal pricing policies, of narcolepsy and multiple sclerosis. a patient.3 decisions about which drugs to prior- The decision to pick those conditions, Biomarkers can serve a variety of itize in the development process, and which established the patient popula- purposes in a clinical trial. Some, but the structure of existing government tion threshold, was influenced by the not all, of these purposes may relate to research and development incentives. then-existing technology and associ- precision medicine. For example, a bio- One area in which these scien- ated fixed costs for drug development. marker can be included to measure the tific developments affect the market In the 35 years since the passage toxicity of a product across an entire involves a firm’s investment decision of the ODA, advances in technology population; this is valuable, but isn’t for products targeting small patient related to biomarkers, as well as devel- especially relevant to targeting. When populations. Historically, there have opments in the understanding of the considering the economic evolution of been limited incentives for pharma- human genome, have changed the cost the pharmaceutical market, we are pri- ceutical firms to develop products tar- structure for firms developing products

20 NBER Reporter • No. 3, September 2018 targeting small patient populations. Benjamin Berger, Nicholas Bagley, ‘Orphan Drug’ Designations and Approvals, 1983–2017 Chandra, Stern, and I examine the mar- ket for orphan drugs and the implica- “Orphan drug” designations and approvals 500 tions of changes in the ability of firms Human Genome to develop these types of products.4 Project completed Designations Orphan designations are a formal regu- 400 latory acknowledgement that a firm is attempting to develop a drug for a rare 300 disease and is a necessary precursor to developing an approved orphan drug. 200 In recent years, as Figure 2 illustrates, there has been a marked increase in the 100 Approvals number of these designations — with a rapid increase in the United States 0 broadly following the completion of 1985 1990 1995 2000 2005 2010 2015 the Human Genome Project. “Orphan drugs” are products aimed at treating conditions a licting fewer than 200,000 patients. The rapid increase in clinical “Designations” are regulatory acknowledgments that are a necessary precursor to developing an approved orphan drug. trial activity for precision products Source: Forthcoming in J. Lerner and S. Stern, eds., Innovation Policy and the Economy, Volume 19, University of Chicago Press and the number of products receiv- ing orphan designations should change Figure 2 the nature of pricing and competition truly competitive post-patent market The attractiveness of any market in these markets. Precision medicines for complex biologic products (i.e. bio- from the perspective of a new entrant is and orphan products have, on average, similars), for small-molecule products a function of the expected profitability higher prices than other medications. the expiration of a patent is normally of entry. For products targeting excep- As we discuss in our joint work, and as followed by generic entry and large tionally small patient populations, the Stern, Chandra, and Brian Alexander price decreases. However, as technol- fixed costs of entry and the likelihood explain in an additional paper, these ogy allows for drug developers to tar- of intense post-entry price competi- high prices are the result of a selection get increasingly smaller patient popu- tion mean that a new entrant is unlikely effect of products brought to market lations, the future prospects for this to earn profits. This means that in rather than a special pricing rule for competitive system are limited. the markets for some drugs targeted orphan diseases.5 For example, given at small populations, a generic firm the small patient population, firms will will never emerge — regardless of how only bring to market products that gen- Generic Competition among high a price-cost margin the incum- erate large amounts of value for indi- Orphan and Non-Orphan Drugs bent firm is able to charge. Evidence viduals with those conditions. For such of this phenomenon can be seen in the products, the potential value created Drugs approved, 1984–2011 nature of generic competition across leads to an expected price and result- 1,500 orphan and non-orphan products in ing profits that justify the research Figure 3. Approximately 50 percent of 1,200 and development investments. Thus, small-molecule non-orphan drug prod- in equilibrium, prices are higher for ucts have faced generic competition in 900 51% orphan drugs that firms choose to bring the form of a competitor firm filing to market than for other drugs. Share of an abbreviated new drug application 600 drugs with While the high prices for orphan generic (ANDA) — a necessary regulatory step competitor(s) drugs may represent an equilibrium 300 for a firm to produce a generic product. based on the investment decisions of In contrast, only 33 percent of small- 33% firms, as for other drugs, these high 0 molecule orphan products have ever prices are only intended to exist while Non-orphan Orphan had an ANDA filed against them. the product is under patent protec- Figure 4, on the next page, demon- tion. For traditional, small-molecule Data pertain to small molecule drugs strates the role of market size in deter- Source: Forthcoming in J. Lerner and S. Stern, eds., drugs, the United States has long pro- Innovation Policy and the Economy, Volume 19, mining future competition. It shows vided the policy framework to support University of Chicago Press the average peak demand in phar- a robust system of generic competition. macy claims data for orphan and non- While the United States still lacks a Figure 3 orphan products based on the pres-

NBER Reporter • No. 3, September 2018 21 ence of an ANDA. These data come tition may quickly allow patients to created; this is often called indication- from OptumLabs and comprise all capture that value. For example, Vertex based pricing. Such a pricing system is retail and mail-order pharmacy claims (the manufacturer of Kalydeco) is promoted by many policy activists who and inpatient and outpatient medi- developing several additional products believe that charging prices based on a cal claims filed by United Healthcare that target cystic fibrosis patients with product’s indication-specific value will beneficiaries between 1992 and 2017. mutations — each of which will likely lower prices and reduce pharmaceutical The table demonstrates the importance face little competition from generics profits.6 However, a system allowing of market size in determining compe- or therapeutic substitutes. Future work firms to charge different prices based tition. Products that never receive an should examine the degree to which on a consumer’s value and willingness ANDA have meaningfully smaller peak firms are shifting research away from to pay for a product presents the ideal demand compared with those that face the larger market products that are conditions for price discrimination. some form of generic To understand competition. Demand for Pharmaceuticals with Generic Competition and Without the way in which The lack of indication-based generic competition pricing allows firms Average pharmacy claims in peak year between 1992 and 2017 (000s) emerging to com- to price-discriminate, pete with branded Non-orphan drugs Orphan drugs consider how phar- 35 products losing pat- maceutical firms set ent protection is cur- 30 prices for products rently limited to a 25 that can be used to relatively small num- 20 treat multiple condi- ber of drugs that tar- tions. For all drugs, get very small patient 15 prices are set based on markets. However, 10 negotiations between as more and more 5 pharmaceutical firms firms develop preci- and payers/phar- 0 sion medicines, the With potential Without generic With potential Without generic macy benefit manag- lack of generic entry generic competition competition generic competition competition ers. In most cases, a in small drug mar- firm’s optimal price is “Orphan drugs” are products aimed at treating conditions a licting fewer than 200,000 patients. kets could become Source: Author’s calculations based on data from OptumLabs limited by the value a greater threat to a product creates, future price competi- Figure 4 because an insurer tion across the entire must pass along the market. Consider, for example, the more likely to receive meaningful com- cost to patients via premiums.7 If phar- case of Kalydeco (ivacaftor), which petition in favor of these smaller mar- maceutical manufacturers set a high is a treatment for a subset of cystic kets that might offer larger and more price relative to the value created for a fibrosis patients who also have a par- long-lived profits. specific indication, payers will imple- ticular set of mutations. The likely Beyond simply extending the time ment utilization management pro- patient population is estimated at period during which firms face little grams that limit access to the product. between 2,000 and 3,000, and the drug or no competition, an increasing abil- For products treating multiple condi- costs several hundred thousand dollars ity to develop products aimed at small tions, this ability for payers to restrict per year. Despite the high prices, the patient populations could change a access requires pharmaceutical manu- small patient population means that firm’s optimal pricing strategy. If firms facturers to consider which segments it is unlikely that additional firms will can more accurately predict a drug’s ex of the market it will attempt to serve. attempt to target patients currently ante efficacy — as is true for many pre- The pricing decisions of firms for treated with Kalydeco. cision medicines — they could develop different markets is summarized in Profit-maximizing firms under- more complicated pricing policies that Figure 5.8 Panel A contains a depiction stand the benefits of potentially long- allow them to capture more of the value of optimal pricing decisions for firms lived profits from products targeting they create. This is particularly true if a under the existing uniform pricing sys- these smaller patient populations. As product can be used to treat multiple tem for a product that treats three indi- a result, it may be optimal for firms to conditions with varying efficacy across cations with differing efficacy. Under focus increasingly on products that cre- these conditions. In these settings, firms Scenario 1, patients with Indication ate meaningful value for large patient may be interested in charging prices C receive the least relative value of all populations but where future compe- based on the indication-specific value patients. However, given the large size

22 NBER Reporter • No. 3, September 2018 Market Scenarios for Precision Medicine Under Uniform Pricing and Indication-Based Pricing

Uniform Pricing Indication-Based Pricing

Scenario 1 Scenario 1

Indication A Indication B Indication C Indication A Indication B Indication C

Profit-maximizing price for indication A Consumer surplus Profit-maximizing Profit-maximizing uniform price price for indication B

Value to patient to Value patient to Value Profit-maximizing price for indication C

No. of patients No. of patients

Scenario 2 Scenario 2

Indication A Indication B Indication C Indication A Indication B Indication C

Profit-maximizing price for indication A Consumer Profit-maximizing surplus uniform price Profit-maximizing price for indication B

Patients who do Profit-maximizing not receive drug price for indication C Value to patient to Value patient to Value

No. of patients No. of patients

Source: Illustrative diagram based on author’s theoretical scenarios

Figure 5 of this population, and the broadly to price at the maximum willingness to could increase the amount of innova- high value they receive, a pharmaceu- pay for each indication and consumers tion, as products that previously would tical manufacturer finds it optimal to don’t respond to a higher price through not have generated sufficient profits set a price that causes insurers to allow reduced utilization. In this case, the to justify development are now worth these patients to access the medication. ability to charge multiple prices means more in expectation. However, the This relatively low price allows patients that firms are able to capture far more higher prices for each indication could with Indications A and B to enjoy a surplus than they could under a uni- reduce output — particularly if patients relatively large amount of consumer form pricing system. are exposed to meaningful cost-sharing surplus. In contrast, in Scenario 2 of After considering this simple exam- based on the price of the product. The Panel A, patients who have Indication ple, it becomes hard to imagine how potential scope for these welfare losses C receive such a small amount of value implementing indication-based pricing increases if each indication is quite from the product that the pharmaceu- would result in reduced pharmaceu- small and therefore unlikely to attract tical firm finds it optimal to set a price tical profits or lower average prices. competition after patent expiration. that causes the insurer to limit these However, the welfare implications of In summary, the economics of the patients from accessing the drug. This indication-based pricing remain decid- pharmaceutical market are shaped in decreases output and results in some of edly unclear. If the number of markets part by the scientific research and devel- the consumer surplus from Indication similar to Scenario 2 is large, then insti- opment process. As a result, changes in B patients now being captured by the tuting indication-based pricing could the nature and type of medications that firm. be output-expanding. However, if most can be developed can ripple through the Now consider the situation where markets resemble Scenario 1, the pri- entire system, impacting firms’ innova- a pharmaceutical manufacturer could mary effect of an indication-based pric- tion incentives, competition in phar- charge multiple prices. For ease of dis- ing scheme would be to transfer value to maceutical markets, and public and pri- cussion, Panel B presents a simplified firms. Again, the welfare implications vate drug spending. These changes can version of this for the two scenarios of this transfer are unclear. Greater affect the optimal nature of decisions discussed above, where firms are able expected profits from such a system regarding the protection of intellectual

NBER Reporter • No. 3, September 2018 23 property, as well as pricing strategies Journal of Economics, 119(2), 2004, Economy, Vol. 19, http://www.nber.org/ implemented by firms that themselves pp. 527–64; M. Blume-Kohout and chapters/c14097.pdf will dictate the welfare generated in N. Sood, “The Impact of Medicare Part Return to Text these markets. D on Pharmaceutical R&D,” NBER 5 A. Stern, B. Alexander, and A. Far more work is needed to under- Working Paper No. 13857, March Chandra, “How Economics Can Shape stand the economic factors affect- 2008, and published as “Market Size Precision Medicines,” Science, 335(6330), ing firms in this area. For example, and Innovation: Effects of Medicare 2017, pp. 1131–3. we know that changes in profitability Part D on Pharmaceutical Research Return to Text can affect investments in innovation, and Development,” Journal of Public 6 P. Bach, “Indication-Specific Pricing for but we know little about the quality Economics, 2013, 97, pp. 327–36; P. Cancer Drugs,” Journal of the American of that innovation. In a recent paper, DuBois, O. Mouzon, F. Scott-Morton, Medical Association, 312(16), 2014, pp. David Dranove, Manuel I. Hermosilla, and P. Seabright, “Market Size and 1629–30. and I find little evidence that truly Pharmaceutical Innovation,” The RAND Return to Text novel products are affected by mar- Journal of Economics, 46(4), 2015, pp. 7 Though in some markets, firms are able ginal changes to profitability, but this 844–71. to exploit forced bundling in insurance con- relationship warrants further study.9 In Return to Text tracts to charge prices that exceed the value addition, little is known about the cur- 2 A. Chandra, C. Garthwaite, and created by the products. D. Besanko, D. rent incentives for firms to develop new A. Stern, “Characterizing the Drug Dranove, and C. Garthwaite, “Insurance biomarkers — particularly those that Development Pipeline for Precision and the High Prices of Pharmaceuticals,” could decrease market size by provid- Medicines,” chapter in forthcoming book, NBER Working Paper No. 22353, June ing more information about the opti- E. Berndt, D. Goldman, and J. Rowe, eds., 2016. mal use of existing products.10 Economic Dimensions of Personalized Return to Text and Precision Medicine, University of 8 A. Chandra and C. Garthwaite, Chicago Press. “The Economics of Indication-Based 1 D. Acemoglu and J. Linn, “Market Return to Text Drug Pricing,” New England Journal of Size in Innovation: Theory and Evidence 3 S. Amur, “Biomarker Terminology: Medicine, 377(2), 2017, pp. 103–6. From the Pharmaceutical Industry,” Speaking the Same Language,” available Return to Text NBER Working Paper No. 10038, at https://www.fda.gov/downloads/Drugs/ 9 D. Dranove, C. Garthwaite, and M. October 2003, and The Quarterly DevelopmentApprovalProcess/Drug-Dev Hermosilla, “Pharmaceutical Profits and Journal of Economics, 119(3), 2004, pp. elopmentToolsQualificationProgram/ the Social Value of Innovation,” NBER 1049–90; A. Finkelstein, “Health Policy UCM533161.pdf Working Paper No. 20212, June 2014. and Technological Change: Evidence Return to Text Return to Text from the Vaccine Industry,” NBER 4 N. Bagley, B. Berger, A. Chandra, C. 10 A. Stern, B. Alexander, and A. Working Paper No. 9460, January 2003, Garthwaite, and A. Stern, “The Orphan Chandra, “Innovation Incentives and and published as “Static and Dynamic Drug Act at 35: Observations and an Biomarkers,” Clinical Pharmacology and Effects of Health Policy: Evidence from Outlook for the 21st Century,” forthcom- Therapeutics, 103(1), 2018, pp. 34–6. the Vaccine Industry,” The Quarterly ing, NBER Innovation Policy and the Return to Text

24 NBER Reporter • No. 3, September 2018 Mortgage Lending and Housing Markets

Fernando Ferreira

At the onset of the last housing cri- sis by constructing a panel of housing ket over the course of the housing boom, sis, it was widely believed that the lenders ownership sequences that contains more roughly doubling to just over 20 percent. who extended subprime mortgages and than 33 million ownership spells from However, this came at the expense of the the homeowners who had taken out those 1997 to 2012.1 These data, acquired from government-insured subsector — Federal loans were responsible for the housing CoreLogic, are based on the universe Housing Administration and Veterans boom, bust, and ensuing economic cri- of housing transactions for almost 100 Affairs loans — not the prime mortgage sis. With the benefit of hindsight — and Metropolitan Statistical Areas. We merge sector. Prime mortgages were always the aided by much better data and research them with the Home Mortgage Disclosure dominant loan type across the cycle, with designs — academic researchers now have Act files in order to add more loan fea- their share hovering around 60 percent, a clearer view. The credit expansion dur- tures and demographics. Importantly, this and in fact increasing almost 10 percent- ing the housing boom was not concen- panel includes details on every type of age points from 2000 to 2006. Finally, trated in the subprime those using only cash sector, and the major- to purchase a house ity of foreclosures dur- Funding Sources for Home Purchases, 1997–2012 constituted a rela- ing the crisis were not tively stable 10 to 11 associated with sub- Share of all home purchases (%) percent of the sample prime mortgages. 80 until 2010, after which African-American and 70 this share increased to Hispanic homebuyers 16 percent, due to the 60 paid higher mortgage Prime loans unavailability of credit costs relative to compa- 50 during that period and rable homebuyers dur- 40 the increase in the rel- ing the last cycle, inde- ative number of cash 30 pendent of whether Government loans Subprime loans All cash purchases investors. they used subprime or 20 The aggregate data prime loans. Finally, indicate that subprime 10 those minorities were did not take over the hurt most by the fore- 0 mortgage market dur- closure crisis, especially Q3 1997 Q3 2002 Q3 2007 Q3 2012 ing the housing boom; when they bought the pattern was rather homes at or near the Source: F. Ferreira and J. Gyourko, NBER Working Paper No. 21261 one of broad-based peak of the housing expansion of credit. boom. Figure 1 This has been corrob- Much of my recent orated by other recent research focuses on understanding three option available for financing a home pur- studies. For example, Manuel Adelino, key issues related to subprime mortgages chase — prime and subprime mortgages, Antoinette Schoar, and Felipe Severino and minority borrowers during the last cash, and governmental loans — as well as find that the mortgage expansion was housing cycle: the role of subprime loans for refinancing during an ownership spell. shared across the entire income distribu- during the housing boom, the foreclosure This fixes the missing data problem of tion, as opposed to being concentrated in crisis, and the vulnerability of minority research conducted early in the cycle that low-income groups.2 Christopher Foote, homeowners during the boom and bust. relied solely on subprime mortgage data. Lara Loewenstein, and Paul Willen dem- Figure 1 documents the market onstrate that, since high-income borrow- Subprime Mortgages and shares of the different sources of fund- ers tend to use mortgages with higher loan the Housing Boom ing used by homeowners. The subprime amounts, wealthy borrowers accounted for sector, which included many alternative most of the increase in outstanding mort- Joseph Gyourko and I uncover basic loans issued to higher-risk borrowers, gage debt in dollar terms.3 Neil Bhutta stylized facts about the foreclosure cri- indeed expanded its share of the mar- looks at the dollar value of mortgage

NBER Reporter • No. 3, September 2018 25 inflows to reveal that first-time homebuy- Subprime subsector borrowers’ distress ers, even the ones with low credit scores, spiked first, beginning in 2006, and experienced only modest growth in credit quickly reached double-digit percentage inflows.4 He finds that the largest inflows rates by the time the global financial cri- were in fact due to investors, not house- sis hit in 2008. But we find that this ini- holds. And to put the proverbial last nail tial shock in subprime distress was spa- in the coffin of the “subprime caused tially concentrated in a relatively small the boom” narrative, Stefania Albanesi, number of metropolitan areas in central Giacomo De Giorgi, and Jaromir Nosal California. show that credit growth was concentrated Much less well known is the fact that in the prime segment, and that so-called there was a surge in prime subsector dis- high-risk borrowers had similar growth tress within a few months of the initial in virtually all debt categories during the surge in subprime borrower home losses. early 2000s.5 The rate of home loss for prime borrow- Fernando Ferreira is an associate ers never approached that of subprime professor of real estate and business Foreclosure Crisis borrowers, but it remained high through economics and public policy at The 2012. Because prime borrowers far out- Wharton School of the University Gyourko and I also document how number subprime borrowers, even with of Pennsylvania, where he is also the housing distress evolved over the cycle. a lower foreclosure rate they still account coordinator of the Wharton Applied Distress is defined as a home being lost for twice as many home losses as subprime Economics PhD Program. He is a to foreclosure or short sale. Foreclosed borrowers. Figure 2 shows the total num- research associate in the NBER Public Economics Program, and has co-orga- nized the NBER Summer Institute Housing Foreclosures and Short Sales, 1997–2012 Real Estate meeting since 2014.

Ferreira is also a faculty fel- Total quarterly foreclosures and short sales by funding source (000s) low of the Penn Institute for Urban 100 Research, and was a co-editor of the Prime loans Journal of Public Economics from 2013 80 to 2018. His research interests lie in the 60 intersection of real estate, urban eco- nomics, and public economics. He Subprime loans has written about household loca- 40 tion decisions and valuation of pub- All cash purchases 20 lic goods, with a special focus on Government loans public elementary schools; how the

size and composition of local govern- 0 ments are influenced by housing mar- Q3 1997 Q3 2002 Q3 2007 Q3 2012 kets, political parties, income inequal-

ity, and female leadership; and the Source: F. Ferreira and J. Gyourko, NBER Working Paper No. 21261 causes and consequences of the last housing cycle, including the impact of Figure 2 negative equity on household mobil- ity and the vulnerability of minority homeowners during the boom and homes are explicitly identified in the ber of home losses by quarter in this sam- bust. DataQuick files by a distress code that ple, by type of home financing source. Ferreira obtained his PhD in indicates the date the home was lost by Hence, the foreclosure crisis that economics from the University of the previous owner. A short sale is defined started in 2006 was still not over in California, Berkeley. He also has an as a transaction in which the sales price is 2012. Though it started in the subprime MA in economics from the Federal no more than 90 percent of the outstand- subsector, it did not remain there for University of Rio Grande do Sul and ing balance on all existing debt. very long, and it ultimately became a a BA in economics from the State The initial finding of distress is con- broad housing market phenomenon. University of Maringa, both in Brazil. sistent with much previous research: This pattern of distress is also found by

26 NBER Reporter • No. 3, September 2018 Albanesi, De Giorgi, and Nosal, who Willen use data from the Panel Study sidered in mortgage underwriting. report that the rise in mortgage default of Income Dynamics to circumvent this We find that African-American and during the foreclosure crisis was concen- problem and find that change in abil- Hispanic borrowers are 103 percent and trated in the middle of the credit score ity to pay is the main factor explaining 78 percent more likely to receive high- distribution, not among low-credit score mortgage default.6 cost mortgages for home purchases, borrowers. after controlling for individual credit Gyourko and I also find that loan- Vulnerability of Minority scores, other risk factors, and whether to-value ratios at the time of house pur- Homeowners the borrower used a prime or subprime chase did not vary much by type of loan. A large fraction of this effect is mortgage. But over the course of a home- Improved datasets also helped due to sorting of borrowers across lend- owner’s time in her house, the loan-to- Patrick Bayer, Stephen Ross, and me to ers with a particular characteristic. This value ratio varies, partly as a function understand and to compare mortgage most important lender characteristic is of loan repayment but also as a func- costs by race and ethnicity during the not something observed at the time of tion of house price movements. Rising last housing boom. This was not an easy origination; instead, it is related to the house prices result in behavior of these com- lower loan-to-value panies during the fore- ratios. These ratios Mortgage Foreclosures by Race and Ethnicity, 2004–2009 closure crisis. These on the overall hous- lenders disproportion- ing stock reached their Annual foreclosure rate (%) ally foreclosed proper- lowest levels between 20 ties during the Great 2005 and 2006; this Hispanics Recession. The role of may have influenced lenders in the foreclo- 15 the perception that sure crisis remains an housing markets were Blacks understudied topic. healthy. As home 10 Bayer, Ross, and prices fell after 2006, I in a separate paper current loan-to-value also analyze differ- 5 ratios shot up, turning Whites ential rates of mort- into negative equity, gage foreclosures and and foreclosure rates delinquencies faced by 0 rose. Adelino, Schoar, 2004 2005 2006 2007 2008 2009 minority homeown- and Severino find that ers.7 We first show default rates went Source: P. Bayer, F. Ferreira, and S. Ross, NBER Working Paper No. 19020, and published as “The Vulnerability of Minority that, while all home- up predominantly in Homeowners in the Housing Boom and Bust,” American Economic Journal: Economic Policy, 8(1), 2016, pp. 1-27. owners had negligible areas with large house 90-day delinquency price reductions. Figure 3 and foreclosure rates Within those areas, the in 2004 and 2005, largest default effects were concentrated task, because valid comparison required very large racial and ethnic differences among high-income and high credit- us to identify whites, African Americans, emerged by 2008 and 2009 [see Figure score borrowers. and Hispanics with similar creditworthi- 3]. The numbers are stark: More than Defaults and foreclosures can hap- ness, demographics, loan characteristics, 1 in 10 minority homeowners in the pen because of strategic reasons — keep- etc. We dealt with this empirical chal- sample had a delinquent mortgage in ing up with monthly mortgage payments lenge by assembling a unique panel data 2009, compared with 1 in 25 for white may not be worthwhile once a house set that links individual housing trans- households. has negative equity — and also simply actions, individual mortgage decisions Using the same panel data, we esti- because of changes in homeowners’ abil- and demographics, and individual credit mate that minorities were 3 percent ity to make payments due to the unem- data. We leverage this panel to examine more likely to experience foreclosure ployment shock of the Great Recession. pricing of mortgages that vary by race than white homeowners with similar An important data limitation still faced of the homeowner separately from the credit scores, loan characteristics, demo- by researchers is the difficulty of linking racial composition of the neighborhood. graphics, house type, neighborhood, and microdata on employment and incomes The panel also includes a representative lender.8 This difference is especially pro- with individual-level data on credit and sample of all mortgages, not just sub- nounced for loans originated near the housing choices. Kristopher Gerardi, prime loans. Finally, it contains all stan- peak of prices during the housing boom. Kyle Herkenhoff, Lee Ohanian, and dard risk factors that are typically con- And the differential foreclosure effect by

NBER Reporter • No. 3, September 2018 27 minority status seems to be explained of the Middle Class,” NBER Working 6 K. Gerardi, K. Herkenhoff, L. largely by the lower rates of employment Paper No. 20848, January 2015, and Ohanian, and P. Willen, “Can’t Pay or among African Americans. The Review of Financial Studies, 29(7), Won’t Pay? Unemployment, Negative Taken together, these estimates pro- 2016, pp. 1635–70. Equity, and Strategic Default,” NBER vide evidence that minority households Return to Text Working Paper No. 21630, October drawn into homeownership late in the 3 C. Foote, L. Loewenstein, and P. 2015. housing boom were especially vulner- Willen, “Cross-Sectional Patterns of Return to Text able, both because they acquired assets Mortgage Debt during the Housing 7 P. Bayer, F. Ferreira, and S. Ross, at peak prices and because they suf- Boom: Evidence and Implications,” “What Drives Racial and Ethnic fered unemployment consequences of NBER Working Paper No. 22985, Differences in High-Cost Mortgages? the downturn more acutely. December 2016. The Role of High-Risk Lenders,” NBER Return to Text Working Paper No. 22004, February 4 N. Bhutta, “The Ins and Outs of 2016, and The Review of Financial 1 F. Ferreira and J. Gyourko, “A New Mortgage Debt During the Housing Studies, 31(1), 2018, pp. 175–205. Look at the U.S. Foreclosure Crisis: Panel Boom and Bust,” Journal of Monetary Return to Text Data Evidence of Prime and Subprime Economics, 76, 2015, pp. 284–98. 8 P. Bayer, F. Ferreira, and S. Ross, “The Borrowers from 1997 to 2012,” NBER Return to Text Vulnerability of Minority Homeowners Working Paper No. 21261, June 2015. 5 S. Albanesi, G. De Giorgi, and J. in the Housing Boom and Bust,” NBER Return to Text Nosal, “Credit Growth and the Financial Working Paper No. 19020, May 2013, 2 M. Adelino, A. Schoar, and F. Crisis: A New Narrative,” NBER and American Economic Journal: Severino, “Loan Originations and Working Paper No. 23740, August 2017. Economic Policy, 8(1), 2016, pp. 1–27. Defaults in the Mortgage Crisis: The Role Return to Text Return to Text

28 NBER Reporter • No. 3, September 2018 NBER News

Two New Directors Elected to NBER Board

Diana Farrell has been elected an at- Samuel Kortum, the James Burrows large member of the NBER Board of Moffatt Professor of Economics at Yale Directors. She is president and CEO of the University, is Yale’s new representative JPMorgan Chase Institute, which carries on the board. Kortum’s principal areas out economic and policy research using the of research are international economics, rich administrative data resources that arise industrial organization, and macroeconom- from the firm’s client interactions. ics. In 2004, he and Jonathan Eaton shared Previously, she was a senior partner at the Econometric Society’s Frisch Medal for McKinsey & Company, where she was the their paper “Technology, Geography, and global head of the McKinsey Center for Trade,” which provided a framework that Government and the McKinsey Global has been applied in many subsequent stud- Institute. ies of trade patterns. The researchers were Farrell served in the White House as honored with the 2018 Onassis Prize in deputy director of the National Economic International Trade. Council and Deputy Assistant to the Kortum taught at Boston University, Diana Farrell President on Economic Policy, 2009– the University of Minnesota, and the 10, and was a member of the President’s University of Chicago before joining Auto Recovery Task Force. She currently the Yale faculty. He is a member of the serves on the board of directors of eBay, American Academy of Arts and Sciences, the Urban Institute, and the Washington a Fellow of the Econometric Society, and International School. She is a trustee emer- a past editor of the Journal of Political ita of Wesleyan University, and was a co- Economy. He has been an NBER affili- chair of the World Economic Forum’s ate since 1993 and spent the 1999–2000 Council on Economic Progress. She is academic year at the NBER as a National a member of the Council on Foreign Fellow. Relations and the Aspen Strategy Group. Kortum received his BA from Wesleyan She holds a BA from Wesleyan University and his PhD from Yale. University, which has awarded her a distin- Ray Fair, Yale’s former representative guished alumna award, and an MBA from on the NBER board, was elected to emeri- Harvard Business School. tus status. Samuel Kortum 41st Annual NBER Summer Institute Researchers from 17 countries and the participants were not NBER affil- the role of credit access for house- 427 institutions participated in the iates. Researchers submitted 5,819 holds and firms in both the run-up to NBER’s 41st annual Summer Institute, papers, of which 563 were selected for the 2008 global financial crisis and in which was held in Cambridge dur- presentation. the years following the crisis. Excessive ing a three-week period in July. More Raghuram Rajan, the Katherine liquidity prior to the crisis permitted than 2,800 participants took part in Dusak Miller Distinguished Service mortgage lending with relatively weak 51 distinct meetings arranged by 120 Professor at the University of Chicago’s standards and boosted asset prices. The organizers. Booth School of Business, a former sharp decline in credit after the crisis There were 185 graduate student governor of the Reserve Bank of India, constricted both investment and con- participants, and 545 participants who delivered the 2018 Martin Feldstein sumer spending, placing an important were attending their first Summer Lecture on “The Two Faces of drag on the pace of recovery. An edited Institute. More than 65 percent of Liquidity.” His presentation described text of the lecture appears earlier in this

NBER Reporter • No. 3, September 2018 29 issue of the NBER Reporter. the explanatory variables. Andrews and pre-crisis financial system, post-crisis The 2018 Methods Lectures, on Stock described how to diagnose this lessons on the stabilization role of fiscal “Weak Instruments and What to Do problem, and how to conduct inference policy, and lessons learned about mac- About Them,” were presented by NBER when it arises. roprudential financial policy. Research Associates and Harvard fac- Recognizing that the global finan- The 2018 Feldstein Lecture, ulty members Isaiah Andrews and James cial crisis began a decade ago, the 2018 Methods Lectures, and the presenta- Stock. The so-called “weak instru- Summer Institute also included a day- tions at the Global Financial Crisis ments” problem arises when research- long meeting on “The Global Financial @ 10 conference and several other ers apply instrumental variable meth- Crisis @ 10.” The conference included Summer Institute meetings were video- ods in settings in which the variation presentations on the role of extrapola- taped and can be accessed through the in the exogenous variables accounts tive expectations in inflating pre-crisis NBER Videos tab on the left side of the for only a small part of the variation in asset bubbles, the weaknesses of the NBER homepage.

30 NBER Reporter • No. 3, September 2018 Conferences

Machine Learning in Health Care

An NBER conference on Machine Learning in Health Care took place June 4 in Cambridge. Research Associates David M. Cutler and Sendhil Mullainathan, both of Harvard University, and Ziad Obermeyer of Harvard Medical School organized the meet- ing. The conference was partially funded by a grant from the National Institute on Aging. These researchers’ papers were presented and discussed:

• Sendhil Mullainathan and Ziad Obermeyer, “Are We Over-Testing? Using Machine Learning to Understand Doctors’ Decisions”

• Susan Athey, Stanford University and NBER, “The Impact of Machine Learning on Economics” (Chapter in the forth- coming NBER book The Economics of Artificial Intelligence: An Agenda, Ajay K. Agrawal, Joshua Gans, and Avi Goldfarb, editors, from the University of Chicago Press)

• David C. Chan Jr, Stanford University and NBER, and Jonathan Gruber, MIT and NBER, “Triage Judgments in the Emergency Department”

• Justine S. Hastings, Brown University and NBER, and Mark Howison, Sarah E. Inman, and Miraj G. Shah, Brown University, “Using Big Data and Data Science to Generate Solutions to the Opioid Crisis”

• Jonathan Gruber; Benjamin R. Handel and Jonathan T. Kolstad, University of California, Berkeley and NBER; and Samuel Kina, Picwell Inc., “Managing Intelligence: Skilled Experts and AI in Markets for Complex Products”

• Rahul Ladhania and Amelia Haviland, Carnegie Mellon University; Neeraj Sood, University of Southern California and NBER; and Ateev Mehrotra, Harvard Medical School, “Medication Adherence and Cost Exposure: A Story in Heterogeneity”

Summaries of these papers are at www.nber.org/conferences/2018/MLs18/summary.html

Workshop on Aging and Health

A workshop on Aging and Health cosponsored by the Max Planck Institute for Social Law and Social Policy and the NBER took place June 7–8 in Munich, Germany. Research Associate Axel H. Börsch-Supan of the Max Planck Institute, Fabrizio Mazzonna of Università della Svizzera Italiana, and Research Associate Jonathan S. Skinner of Dartmouth College organized the meeting. These researchers’ papers were presented and discussed:

• Amitabh Chandra, Harvard University and NBER, and Douglas O. Staiger, Dartmouth College and NBER, “Identifying Prejudice in Healthcare by Race, Gender, and Age”

• Axel H. Börsch-Supan; Tabea Bucher-Koenen, Max Planck Institute; and Felizia Hanemann, Technical University of Munich, “Does Disability Insurance Improve Health and Well-being?”

NBER Reporter • No. 3, September 2018 31 • Coen W.A. van de Kraats, Vrije Universiteit Amsterdam and Tinbergen Institute; Titus J. Galama, University of Southern California; and Maarten Lindeboom, Vrije Universiteit Amsterdam, “Light at the End of the Tunnel — Unemployment and Mental Health after Age 50”

• Liran Einav, Stanford University and NBER; Amy Finkelstein, MIT and NBER; Sendhil Mullainathan, Harvard University and NBER; and Ziad Obermeyer, Harvard Medical School, “Does High Healthcare Spending at End of Life Imply Waste? Predictive Modeling Suggests Not Necessarily”

• Mary K. Hamman and John M. Nunley, University of Wisconsin-LaCrosse; Daniela E. Hochfellner, New York University; and Christopher J. Ruhm, University of Virginia and NBER, “Peer Effects and Retirement Decisions: Evidence from Pension Reform in Germany”

• Naoki Aizawa, University of Wisconsin-Madison; Soojin Kim, Purdue University; and Serena Rhee, University of Hawaii, Manoa, “Labor Market Screening and Social Insurance Program Design for the Disabled”

• Andreas Haller and Josef Zweimueller, University of Zurich, and Stefan Staubli, University of Calgary and NBER, “Tightening Disability Screening or Reducing Disability Benefits? Evidence and Welfare Implications”

• Peter Hudomiet and Susann Rohwedder, RAND Corporation, and Michael D. Hurd, RAND Corporation and NBER, “Using Subjective Conditional Probabilities to Find the Causal Effects of Health, Income, Wealth, and Longevity on Retirement”

• Gopi Shah Goda, Stanford University and NBER; Matthew Levy, London School of Economics; Colleen Flaherty Manchester and Aaron Sojourner, University of Minnesota; and Joshua Tasoff, Claremont Graduate University, “Mechanisms behind Retirement Saving Behavior: Evidence from Administrative and Survey Data”

• Nicholas W. Papageorge, Johns Hopkins University and NBER; Kevin Thom, New York University; and Daniel Barth, University of Southern California, “Genetic Endowments and Wealth Inequality” (NBER Working Paper No. 24642)

• Maarten Lindeboom, “Pension Reform: Disentangling Retirement and Savings Responses”

• Ethan Lieber, University of Notre Dame, and Lee Lockwood, University of Virginia and NBER, “Targeting with In-Kind Transfers: Evidence from Medicaid Home Care” (NBER Working Paper No. 24267)

Summaries of these papers are at www.nber.org/conferences/2018/AHs18/summary.html

32 NBER Reporter • No 3, September 2018 East Asian Seminar on Economics

The NBER, the Tokyo Center for Economic Research, the Korea Development Institute, the Hong Kong University of Science and Technology, the Peking University China Center for Economic Research, the National University of Singapore, the Australian National University, and the Chung-Hua Institution for Economic Research (Taipei) jointly sponsored the NBER’s 29th Annual East Asian Seminar on Economics. The conference, which focused on political economy, was organized by Research Associates Takatoshi Ito of Columbia University and Andrew K. Rose of the University of California, Berkeley. It took place in Seoul, South Korea, June 21–22. These researchers’ papers were presented and discussed:

• Abhijit Banerjee, MIT and NBER; Nils Enevoldsen, MIT; Rohini Pande, Harvard University and NBER; and Michael Walton, Harvard University, “Information as an Incentive: Experimental Evidence from Delhi”

• Dongsoo Kang and Changwoo Nam, Korea Development Institute, “Conflict of Interests between Government and Creditors in Corporate Restructuring: Case of Korea”

• Ippei Fujiwara, Keio University, and Shunsuke Hori, University of Tokyo, “Aging and Deflation”

• Kenichi Ueda, University of Tokyo, “Tail Risk Dumping”

• Zhenyu Cui, Stevens Institute of Technology, and Nobuo Akai, Osaka University, “Corruption, Political Stability and Efficiency of Government Expenditure on Health Care — Evidence from Asian Countries”

• Weijia Li, University of California, Berkeley; Gerard Roland, University of California, Berkeley, CEPR and NBER; and Yang Xie, University of California, Riverside, “Crony Capitalism, the Party-State, and Political Boundaries of Corruption”

• Henry S. Farber and Ilyana Kuziemko, Princeton University and NBER; Daniel Herbst, Princeton University; and Suresh Naidu, Columbia University and NBER, “Unions and Inequality Over the Twentieth Century: New Evidence from Survey Data” (NBER Working Paper No. 24587)

• Sunjoo Hwang, Hwa Ryung Lee, and Keeyoung Rhee, Korea Development Institute, “Regulatory Revolving Door in the Financial Industry: Evidence from South Korea”

• Ying Bai, Chinese University of Hong Kong, and Ruixue Jia, University of California, San Diego, “The Oriental City: Political Hierarchy and Regional Development in China, AD 1000–2000”

• Meng-Chun Liu and Chia-Hsuan Wu, Chung-Hua Institution for Economic Research, “Taiwan’s Import Protection after Acceding to the WTO”

• Chen Lin, Chinese University of Hong Kong; Randall Morck, University of Alberta, Edmonton; Bernard Yeung, National University of Singapore; and Xiaofeng Zhao, Lingnan University, “Anti-Corruption Reforms and Shareholder Valuations: Event Study Evidence from China”

• Xiangyu Shi, Yale University; Tianyang Xi, Peking University; Xiaobo Zhang, Peking University and IFPRI; and Yifan Zhang, Chinese University of Hong Kong, “‘Moving Umbrella’: Bureaucratic Transfers, Collusion, and Rent-Seeking in China”

Summaries of these papers are at www.nber.org/conferences/2018/EASE18/summary.html

NBER Reporter • No. 3, September 2018 33 International Seminar on Macroeconomics

The NBER’s 41st International Seminar on Macroeconomics, hosted by the Central Bank of Ireland, took place in Dublin, Ireland, June 29–30. Research Associates Jordi Galí of CREI and Kenneth D. West of the University of Wisconsin, Madison orga- nized the conference. These researchers’ papers were presented and discussed:

• Jing Cynthia Wu, University of Notre Dame and NBER, and Ji Zhang, Tsinghua University, “Global Effective Lower Bound and Unconventional Monetary Policy” (NBER Working Paper No. 24714)

• Anna Cieslak, Duke University, and Andreas Schrimpf, Bank for International Settlements, “Non-Monetary News in Central Bank Communication”

• Alexander Bick, Arizona State University; Bettina Brueggemann, McMaster University; and Nicola Fuchs-Schündeln and Hannah Paule-Paludkiewicz, Goethe University Frankfurt, “Long-Term Changes in Married Couples’ Labor Supply and Taxes: Evidence from the U.S. and Europe Since the 1980s”

• Björn Richter and Moritz Schularick, University of Bonn, and Ilhyock Shim, Bank for International Settlements, “The Costs of Macroprudential Policy”

• Ester Faia, Universitat Pompeu Fabra; Sebastien Laffitte, ENS Paris-Saclay; and Gianmarco Ottaviano, Bocconi University and London School of Economics, “Foreign Expansion, Competition, and Bank Risk”

• Marco Del Negro, Domenico Giannone, and Andrea Tambalotti, Federal Reserve Bank of New York, and Marc Giannoni, Federal Reserve Bank of Dallas, “Global Trends in Interest Rates”

• Luigi Bocola, Northwestern University and NBER; Alessandro Dovis, University of Pennsylvania and NBER; and Gideon Bornstein, Northwestern University, “Quantitative Sovereign Default Models and the European Debt Crisis”

• Atsushi Inoue, Vanderbilt University, and Barbara Rossi, Universitat Pompeu Fabra, “The Effects of Conventional and Unconventional Monetary Policy on Exchange Rates”

Summaries of these papers are at www.nber.org/conferences/2018/ISOM18/summary.html

Advancing the Science of Science Funding Workshop

A workshop on Advancing the Science of Science Funding took place July 19–20 in Cambridge, supported by the Alfred P. Sloan Foundation. Research Associate Paula Stephan of Georgia State University and Reinhilde Veugelers of the Katholieke Universiteit Leuven organized the meeting. These researchers’ papers were presented and discussed:

• Henry Sauermann, European School of Management and Technology and NBER; Chiara Franzoni, Politecnico di Milano; and Kourosh Shafi, University of Florida, “Crowdfunding Scientific Research” (NBER Working Paper No. 24402)

34 NBER Reporter • No 3, September 2018 • Charles Ayoubi and Fabiana Visentin, EPFL, and Michele Pezzoni, Université Nice, “The Important Thing Is Not to Win, It Is to Take Part: What if Scientists Benefit from Participating in Research Grant Competitions?”

• Misha Teplitskiy, Harvard University; Eva C. Guinan, Dana-Farber Cancer Institute; and Karim Lakhani, Harvard University and NBER, “Social Influence in Science Funding Evaluation Panels: Field Experimental Evidence from Biomedicine”

• Alfredo Di Tillio and Marco Ottaviani, Bocconi University, and Peter Norman Sørensen, University of Copenhagen, “Strategic Sample Selection”

• Marc J. Lerchenmueller, Yale University, “Does More Money Lead to More Innovation? Evidence from the Life Sciences”

• Jacques Mairesse, CREST-ENSAE and NBER; Michele Pezzoni; Paula Stephan; and Julia Lane, New York University, “Examining the Returns to Investment in Science: A case Study”

Summaries of these papers are at www.nber.org/conferences/2018/SFIs18/summary.html

The 27th NBER-TCER-CEPR Conference

The 27th NBER-TCER-CEPR Conference, “Globalization and Welfare Impacts of International Trade,” took place in Tokyo July 27. This meeting was sponsored jointly by the Centre for Economic Policy Research in London, the NBER, the Tokyo Center for Economic Research, the Center for Advanced Research in Finance, and the Center for International Research on the Japanese Economy. Shin-ichi Fukuda of Tokyo University, Takeo Hoshi of Stanford University and NBER, and Fukunari Kimura of Keio University organized the meeting. These researchers’ papers were presented and discussed:

• Richard Baldwin, Graduate Institute, Geneva and NBER, and Toshihiro Okubo, Keio University, “GVC Journeys When National and Territorial Comparative Advantage Differ”

• Ayako Obashi, Aoyama Gakuin University, “Trade Agreement with Cross-Border Unbundling”

• Takeo Hoshi, and Kozo Kiyota, Keio University, “Potentials for Inward Foreign Direct Investment in Japan”

• Keith Head, University of British Columbia, and Thierry Mayer, Sciences-Po, “Misfits in the Car Industry: Offshore Assembly Decisions at the Variety Level”

• Gabriel Felbermayr and Marina Steininger, Ifo Center for International Economics; Fukunari Kimura and Toshihiro Okubo, “Quantifying the EU-Japan Economic Partnership Agreement”

• Katheryn Russ and Deborah Swenson, University of California, Davis and NBER, and Kelly Stangl, University of California, Davis, “Trade Diversion and Trade Deficits under the Korea-U.S. Free Trade Agreement”

• Akira Sasahara, University of Idaho, “Explaining the Employment Effect of Exports: Value-Added Content Matters”

• Olena Ivus, Queen’s University, and Walter Park, American University, “Patent Reforms and Exporter Behavior: Firm- Level Evidence from Developing Countries”

NBER Reporter • No. 3, September 2018 35 • Meredith Crowley, University of Cambridge, and Ning Meng and Huasheng Song, Zhejiang University, “Policy Shocks and Stock Market Returns: Evidence from Chinese Solar Panels”

• Heiwai Tang, Johns Hopkins University, and Hiroyuki Kasahara, University of British Columbia, “Excessive Entry and Exit in Export Markets”

Summaries of these papers are at www.nber.org/conferences/2018/TRIO18/summary.html

Japan Project

The NBER held a meeting on the Japanese economy in Tokyo July 30. The seminar was organized by Shiro Armstrong of the Australian National University; Research Associates Charles Horioka of the Asian Growth Research Institute (Kitakyushu), Takeo Hoshi of Stanford University, and David Weinstein of Columbia University; and Tsutomu Watanabe of the University of Tokyo. These researchers’ papers were presented and discussed:

• Shuhei Kitamura, Osaka University, “Land Ownership and Development: Evidence from Postwar Japan”

• Sergi Basco, Universitat Autònoma Barcelona, and John P. Tang, Australian National University, “The Samurai Bond: Credit Supply and Economic Growth in Pre-War Japan”

• Cynthia Balloch, Columbia University, “Inflows and Spillovers: Tracing the Impact of Bond Market Liberalization”

• Martín Uribe, Columbia University and NBER, “The Neo-Fisher Effect in the United States and Japan” (NBER Working Paper No. 23977)

• Yuhei Miyauchi, MIT, “Matching and Agglomeration: Theory and Evidence from Japanese Firm-to-Firm Trade”

• Toshiaki Iizuka, University of Tokyo, and Hitoshi Shigeoka, Simon Fraser University and NBER, “Patient Cost-sharing and Health Care Utilization among Children”

Summaries of these papers are at www.nber.org/conferences/2018/JPMs18/summary.html

36 NBER Reporter • No 3, September 2018 Incentives and Limitations of Employment Policies on Retirement Transitions

An NBER conference on Incentives and Limitations of Employment Policies on Retirement Transitions, supported by the Alfred P. Sloan Foundation, took place August 10–11 in Jackson Hole, Wyoming. Research Associates Robert L. Clark of North Carolina State University and Joseph P. Newhouse of Harvard University organized the meeting. These researchers’ papers were pre- sented and discussed:

• Maria D. Fitzpatrick, Cornell University and NBER, “Pension Reform and Return to Work Policies”

• Leslie E. Papke, Michigan State University, “Retirement Options and Outcomes for Public Employees”

• John Hsu, Partners Healthcare; Samuel H. Zuvekas, Agency for Healthcare Research and Quality; and Joseph P. Newhouse and Lindsay Overhage, Harvard University, “Impact of Fiscal Shocks on Retiree Health Plans and Effect on Work and Retirement Decisions”

• Norma Coe, University of Pennsylvania and NBER, “Impact of Health Plan Reforms in Washington on Retirement Decisions”

• Vanya Horneff and Raimond Maurer, Goethe University Frankfurt, and Olivia S. Mitchell, University of Pennsylvania and NBER, “How Will Persistent Low Expected Returns Shape Household Behavior?”

• Joseph F. Quinn and Kevin Cahill, Boston College, and Michael Giandrea, Bureau of Labor Statistics, “Transitions from Career Employment among Public- and Private-Sector Workers”

• Melinda S. Morrill, North Carolina State University and NBER, and John Westall, North Carolina State University, “The Role of Social Security in Retirement Timing: Evidence from a National Sample of Teachers”

• Robert L. Clark; Robert G. Hammond, North Carolina State University; and David Vanderweide, North Carolina General Assembly, “Navigating Complex Financial Decisions at Retirement: Evidence from Annuity Choices in Public Sector Pensions”

• Gila Bronshtein, Stanford University; Jason Scott, Financial Engines; John B. Shoven, Stanford University and NBER; and Sita Slavov, George Mason University and NBER, “The Power of Working Longer” (NBER Working Paper No. 24226)

Summaries of these papers are at www.nber.org/conferences/2018/EPRTs18/summary.html

NBER Reporter • No. 3, September 2018 37 Program and Working Group Meetings

Economic Fluctuations and Growth

Members of the NBER’s Economic Fluctuations and Growth Program met July 14 in Cambridge. Research Associates Virgiliu Midrigan of New York University and Hélène Rey of London Business School organized the meeting. These researchers’ papers were presented and discussed:

• Alessandra Fogli, Federal Reserve Bank of Minneapolis, and Veronica Guerrieri, University of Chicago and NBER, “The End of the American Dream? Inequality and Segregation in U.S. Cities”

• Gauti B. Eggertsson, Brown University and NBER, and Jacob Robbins and Ella Wold, Brown University, “Kaldor and Piketty’s Facts: The Rise of Monopoly Power in the United States” (NBER Working Paper No. 24287)

• Gita Gopinath and Jeremy C. Stein, Harvard University and NBER, “Banking, Trade, and the Making of a Dominant Currency” (NBER Working Paper No. 24485)

• Anmol P. Bhandari and Ellen McGrattan, University of Minnesota and NBER, “Sweat Equity in U.S. Private Business” (NBER Working Paper No. 24520)

• Davide Debortoli, Universitat Pompeu Fabra, and Jordi Galí, CREI and NBER, “Monetary Policy with Heterogeneous Agents: Insights from TANK Models”

• Fatih Karahan and Ayşegül Şahin, Federal Reserve Bank of New York, and Benjamin Pugsley, University of Notre Dame, “Demographic Origins of the Startup Deficit”

Summaries of these papers are at www.nber.org/conferences/2018/EFGs18/summary.html

Chinese Economy

Members of the NBER’s Chinese Economy Working Group met in Beijing on September 10–11. Research Associates Hanming Fang of the University of Pennsylvania, Zhiguo He of the University of Chicago, Shang-Jin Wei of Columbia University, and Wei Xiong of Princeton University organized the meeting. These researchers’ papers were presented and discussed:

• Thomas J. Chemmanur, Boston College; Bibo Liu, Tsinghua University; and Xuan Tian, Indiana University, “Rent Seeking, Brokerage Commissions, and the Pricing and Allocation of Shares in Initial Public Offerings”

• Wei Xiong, “The Mandarin Model of Growth”

• Lily Fang, INSEAD; Josh Lerner, Harvard University and NBER, and Chaopeng Wu and Qi Zhang, Xiamen University, “Corruption, Government Subsidies, and Innovation: Evidence from China”

38 NBER Reporter • No 3, September 2018 • Quanlin Gu, Peking University; Jia He, Nankai University; and Wenlan Qian, National University of Singapore, “Housing Booms and Shirking”

• Ying Bai, Chinese University of Hong Kong, and Ruixue Jia, University of California, San Diego, “The Oriental City: Political Hierarchy and Regional Development in China, AD1000–2000”

• Haoyuan Ding, Shanghai University of Finance and Economics; Yichuan Hu, Chinese University of Hong Kong; and Ninghua Zhong, Tongji University, “Informal Financing via University Alumni: The Substitution of Political Connections”

• Ning Zhu, Yinglu Deng, and An Hu, Tsinghua University, “Real Life Experience and Financial Risk-Taking: Natural Experiment Evidence from Automobile Traffic Accidents”

• Huasheng Gao, Fudan University; Donghui Shi, Shanghai Stock Exchange; and Bin Zhao, Shanghai Advanced Institute of Finance, “Does Good Luck Make People Overconfident? Evidence from a Natural Experiment in China”

• Jiangze Bian, University of International Business and Economics; Zhiguo He; Kelly Shue, Yale University and NBER; and Hao Zhou, Tsinghua University, “Leverage-Induced Fire Sales and Stock Market Crashes”

• Valerie J. Karplus, MIT; Shuang Zhang, University of Colorado at Boulder; and Douglas Almond, Columbia University and NBER, “Did Scrubbing the Government Clean Up the Air? Polluter Responses to China’s Anticorruption Campaign”

• Kun Jiang, University of Nottingham; Wolfgang Keller, University of Colorado at Boulder and NBER; Larry Qiu, University of Hong Kong; and William C. Ridley, University of Colorado at Boulder, “International Joint Ventures and Internal vs. External Technology Transfer: Evidence from China” (NBER Working Paper No. 24455)

• Christopher Hansman, Imperial College London; Harrison Hong, Columbia University and NBER; Wenxi Jiang, Chinese University of Hong Kong; and Yu-Jane Liu and Juanjuan Meng, Peking University, “Riding the Credit Boom” (NBER Working Paper No. 24586)

Summaries of these papers are at www.nber.org/conferences/2018/CEf18/summary.html

NBER Reporter • No. 3, September 2018 39 Nonprofit Org. NBER U.S. Postage Reporter PAID NATIONAL BUREAU OF ECONOMIC RESEARCH National Bureau of Economic Research 1050 Massachusetts Avenue Cambridge, Massachusetts 02138-5398 (617) 868-3900

Change Service Requested