1 American Economic Review: Papers & Proceedings 100 May 2010): 573–578 ( http://www.aeaweb.org/articles.php?doi 10.1257/aer.100.2.573 = a in Debt Time of Growth By Carmen M. Reinhart and Kenneth S. Rogoff* - In this paper, we exploit a new multi-country especially against the backdrop of graying pop ( government historical dataset on public ulations and rising social insurance costs? Are ) debt to search for a systemic relationship between high - sharply elevated public debts ultimately a man 1 public debt levels, growth and inflation. ageable policy challenge? Our main result is that whereas the link between Our approach here is decidedly empirical, - growth and debt seems relatively weak at “nor taking advantage of a broad new historical - mal” debt levels, median growth rates for coun dataset on public debt ( in particular, central tries with public debt over roughly 90 percent government debt ) first presented in Carmen M. . Reinhart and Kenneth S. Rogoff of GDP are about one percent lower than other ( 2008, 2009b ) - mean ) growth rates are several wise; average ( Prior to this dataset, it was exceedingly difficult to get more than two or three decades of pub - percent lower. Surprisingly, the relationship lic debt data even for many rich countries, and between public debt and growth is remarkably virtually impossible for most emerging markets. similar across emerging markets and advanced Our results incorporate data on 44 countries economies. This is not the case for inflation. We find no systematic relationship between high spanning about 200 years. Taken together, the debt levels and inflation for advanced econo data incorporate over 3,700 annual observations - - covering a wide range of political systems, insti ( albeit with individual country mies as a group tutions, exchange rate and monetary arrange exceptions including the United States ) . By con - - ments, and historic circumstances. trast, in emerging market countries, high public debt levels coincide with higher inflation. We also employ more recent data on external Our topic would seem to be a timely one. debt, including debt owed both by governments Public debt has been soaring in the wake of the and by private entities. For emerging markets, we find that there exists a significantly more recent global financial maelstrom, especially in the epicenter countries. This should not be sur - severe threshold for total gross external debt public and private ) ( - —which is almost exclu prising, given the experience of earlier severe 2 sively denominated in a foreign currency—than financial crises. Outsized deficits and epic bank the domestically issued ( for total public debt bailouts may be useful in fighting a downturn, but what is the long-run macroeconomic impact, component of which is largely denominated . When gross external debt ) in home currency reaches 60 percent of GDP, annual growth Reinhart: Department of Economics, 4115 Tydings * declines by about two percent; for levels of Hall, University of Maryland, College Park, MD 20742 external debt in excess of 90 percent of GDP, - ; Rogoff: Economics Depart ) ( e-mail: [email protected] ment, 216 Littauer Center, Harvard University, Cambridge growth rates are roughly cut in half. We are not ( MA 02138–3001 e-mail: [email protected] ) . The in a position to calculate separate total exter - authors would like to thank Olivier Jeanne and Vincent R. ( nal debt thresholds as opposed to public debt Reinhart for helpful comments. 1 thresholds ) for advanced countries. The avail - In this paper “public debt” refers to gross central government debt. “Domestic public debt” is government able time-series is too recent, beginning only in debt issued under domestic legal jurisdiction. Public debt 2000. We do note, however, that external debt does not include debts carrying a government guarantee. levels in advanced countries now average nearly Total gross external debt includes the external debts of all 200 percent of GDP, with external debt levels branches of government as well as private debt that is issued being particularly high across Europe. by domestic private entities under a foreign jurisdiction. 2 2009a, b ( Reinhart and Rogoff demonstrate that the ) The focus of this paper is on the longer term aftermath of a deep financial crisis typically involves a macroeconomic implications of much higher protracted period of macroeconomic adjustment, particu - - public and external debt. The final section, how larly in employment and housing prices. On average, public ever, summarizes the historical experience of debt rose by more than 80 percent within three years after the United States in dealing with private sector a crisis. 573
2 574 MAY 2010 AEA PAPERS AND PROCEEDINGS indebtedness stands in This general rise in public G / ebt D D P 100 = 2007 10 0 15 20 0 25 0 0 09 20 stark contrast to the 2003–2006 period of pub - 69 Iceland lic deleveraging in many countries and owes to 44 Ireland direct bailout costs in some countries, the adop - 72 UK 42 Spain tion of stimulus packages to deal with the global 84 US 5. 17 1 recession in many countries, and marked declines 62 Crisis country average 75% ( ) ease incr of in government revenues that have hit advanced Norway 22 Australia and emerging market economies alike. 9 China 21 Thailand 29 Debt, Growth, and II. Inflation Mexico 25 Malaysia 47 Greece The nonlinear effect of debt on growth is 119 Canada 47 ( reminiscent of “debt intolerance” Reinhart, Austria 62 Chile Rogoff, and Miguel A. Savastano 2003 ) and 4 Germany presumably is related to a nonlinear response 44 Japan 182 of market interest rates as countries reach debt Brazil 46 Korea tolerance limits. Sharply rising interest rates, 32 India in turn, force painful fiscal adjustment in the 41 2 o 0% ease cr ) in 1 20 ( f Average for others 49 form of tax hikes and spending cuts, or, in some cases, outright default. As for inflation, an obvious connection stems from the fact that Figure 1. Cumulative Increase in Real Public Debt unanticipated high inflation can reduce the Since 2007, Selected Countries real cost of servicing the debt. Of course, the Unless otherwise noted these figures are for central Note: efficacy of the inflation channel is quite sen - government debt deflated by consumer prices. sitive to the maturity structure of the debt. In Sources: Prices and nominal GDP from International principle, the manner in which debt builds up Monetary Fund, World Economic Outlook. For a complete can be important. For example, war debts are listing of sources for government debt, see Reinhart and arguably less problematic for future growth ) 2009b ( Rogoff . - and inflation than large debts that are accu mulated in peacetime. Postwar growth tends to be high as wartime allocation of manpower and resources funnels to the civilian economy. deleveraging of debts, normal after a financial - Moreover, high wartime government spending, crisis. Not surprisingly, such episodes are asso ciated with very slow growth and deflation. typically the cause of the debt buildup, comes to a natural close as peace returns. In contrast, - a peacetime debt explosion often reflects unsta I. The 2007–2009 Global Buildup in Public Debt ble political economy dynamics that can persist Figure 1 illustrates the increase in ( inflation- for very long periods. adjusted ) public debt that has occurred since - Here we will not attempt to determine the gen esis of debt buildups but instead simply look at - 2007. For the five countries with systemic finan cial crises ( Iceland, Ireland, Spain, the United their connection to average and median growth and inflation outcomes. This may lead us, if any - Kingdom, and the United States , average debt ) - thing, to understate the adverse growth implica levels are up by about 75 percent, well on track to reach or surpass the three year 86 percent bench - tions of debt burdens arising out of the current ) 2009a,b ( crisis, which was clearly a peacetime event. mark that Reinhart and Rogoff , find for earlier deep postwar financial crises. Even in countries that did not experience a major finan - cial crisis, debt rose by an average of about 20 3 debt are not available for many countries. Of course, the percent in real terms between 2007 and 2009. true run-up in debt is significantly larger than stated here, at least on a present value actuarial basis, due to the exten - sive government guarantees that have been conferred on the 3 financial sector in the crisis countries and elsewhere, where Our focus on gross central government debt owes to the fact that time series of broader measures of government for example deposit guarantees were raised in 2008.
3 tIME O IN A wth RO f D EB t G VOL. 100 NO. 2 575 5. 0 6 ( . The bars in Figure very high above 90 percent ) GDP growth bars, left axis ) ( 2 show average and median GDP growth for each of the four debt categories. Note that of the 5 . 5 4. 0 1,186 annual observations, there are a significant number in each category, including 96 above 90 5 n f l a t i o n I percent. ( Recent observations in that top bracket ( line, right axis ) 0 3. ) come from Belgium, Greece, Italy, and Japan. . 4 5 From the figure, it is evident that there is no n obvious link between debt and growth until pub - o i GDP growth t a 4 0 2. lic debt reaches a threshold of 90 percent. The l f n I observations with debt to GDP over 90 percent have median growth roughly 1 percent lower than 3 5 . the lower debt burden groups and mean levels of 1. 0 Using lagged ( growth almost 4 percent lower. 3 ) debt does not dramatically change the picture. The line in Figure 2 plots the median inflation for 0. 0 Average Median Average Median Average Median Average Median 5 2 . the different debt groupings—which makes plain P G / t b e D P D G / t b e D P D G / t b e D / t b e D P D G D % 30 w o l e b o 6 % 0 6 e v o b 9 t 0 3 % % 90 o t 0 a 0 that there is no apparent pattern of simultaneous rising inflation and debt. 0 1. − 2 Table 1 provides detail on the growth experi - Figure 2. Government Debt, Growth, and Inflation: ence for individual countries, but over a much Selected Advanced Economies, 1946–2009 longer period, typically one to two centuries. Interestingly, introducing the longer time-series Notes: Central government debt includes domestic and yields remarkably similar conclusions. Over the external public debts. The 20 advanced economies included past two centuries, debt in excess of 90 percent are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, has typically been associated with mean growth Netherlands, New Zealand, Norway, Portugal, Spain, of 1.7 percent versus 3.7 percent when debt is Sweden, the United Kingdom, and the United States. The , and compared ( under 30 percent of GDP low ) number of observations for the four debt groups are: 443 with growth rates of over 3 percent for the two for debt/GDP below 30 percent; 442 for debt/GDP 30 to 60 percent; 199 observations for debt/GDP 60 to 90 percent; middle categories - debt between 30 and 90 per ( and 96 for debt/GDP above 90 percent. There are 1,180 cent of GDP ) . Of course, there is considerable observations. variation across the countries, with some coun - International Monetary Fund, w orld Economic Sources: tries such as Australia and New Zealand experi - Development Global , OECD, World Bank, Outlook encing no growth deterioration at very high debt inance f , and Reinhart and Rogoff ( 2009b ) and sources cited therein. levels. It is noteworthy, however, that those high- growth high-debt observations are clustered in the years following World War II. Countries Advanced A. Evidence from Emerging Market B. Evidence from Countries Figure 2 presents a summary of inflation and GDP growth across varying levels of debt for 20 We next perform the same exercise for 24 emerging market economies for the periods advanced countries over the period 1946–2009. 1946–2009 and 1900–2009, using comparable This group includes Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, central government debt data to those we used 4 Perhaps surpris- Greece, Ireland, Italy, Japan, Netherlands, New for the advanced economies. ingly, the results illustrated in Figure 2 and Zealand, Norway, Portugal, Spain, Sweden, the Table 1 for advanced economies are repeated United Kingdom, and the United States. The - annual observations are grouped into four cat for emerging market economies. The emerging egories, according to the ratio of debt to GDP during that particular year as follows: years when debt to GDP levels were below 30 percent ( low 4 While we have pre-1900 inflation, real GDP, and public - GDP was 30 to 60 per ; years where debt ) debt / debt data for many emerging markets, nominal GDP data is cent ; and ) high harder to find. ; 60 to 90 percent ) medium debt ( (
4 MAY 2010 576 AEA PAPERS AND PROCEEDINGS Table 1—Real GDP Growth as the Level of Government Debt Varies: Selected Advanced Economies, 1790–2009 ( ) annual percent change ( ) government debt / GDP federal Central Below 30 90 percent 30 to 60 60 to 90 Period percent percent percent Country and above 2.3 Australia 4.6 1902–2009 4.1 3.1 1880–2009 Austria n.a. 2.3 3.0 4.3 3.3 1835–2009 3.0 2.6 2.1 Belgium 3.0 Canada 1925–2009 2.0 4.5 2.2 2.4 n.a. Denmark 1880–2009 3.1 1.7 1913–2009 3.2 3.0 4.3 1.9 Finland 1880–2009 France 2.3 2.8 2.7 4.9 0.9 Germany 1880–2009 3.6 n.a. n.a. Greece 1884–2009 4.0 0.3 4.8 2.5 Ireland 2.4 4.0 4.5 4.4 1949–2009 1.9 0.7 Italy 1880–2009 5.4 4.9 0.7 1885–2009 4.9 3.7 3.9 Japan Netherlands 1880–2009 2.0 2.4 2.8 4.0 New Zealand 2.5 2.9 3.9 3.6 1932–2009 n.a. 1880–2009 2.9 4.4 n.a. Norway 1851–2009 4.8 2.5 1.4 Portugal n.a. Spain 1850–2009 3.3 1.3 2.2 1.6 n.a. 2.7 2.9 Sweden 1880–2009 2.9 1830–2009 2.5 2.2 2.1 1.8 United Kingdom United States 1790–2009 4.0 3.4 3.3 − 1.8 Average 3.4 3.7 1.7 3.0 3.9 3.1 2.8 1.9 Median 352 866 654 445 Observations = 2,317 An n.a. denotes no observations were recorded for that particular debt range. There Notes: are missing observations, most notably during World War I and II years; further details are provided in the data appendices to Reinhart and Rogoff ) and are available from the 2009b ( bolded italics . authors. Minimum and maximum values for each debt range are shown in Sources: There are many sources; among the more prominent are: International Monetary inance orld Economic Outlook f Global Development , OECD, World Bank, . Extensive Fund, w ) 2009 ( other sources are cited in Reinhart and Rogoff . market equivalents of Figure 2 and Table 1 are with significantly higher levels of inflation in ) , to economize on space emerging markets. Median inflation more than ( not reproduced here - ( doubles but the interested reader is referred to the from less than seven percent to 16 per ( cent ) 0 to 30 percent as debt rises from the low ) NBER working paper version of this paper. For 1900–2009, for example, median and aver range to above 90 percent. Fiscal dominance is a - plausible interpretation of this pattern. age GDP growth hovers around 4–4.5 percent for levels of debt below 90 percent of GDP, but Because emerging markets often depend so much on external borrowing, it is interesting to median growth falls markedly to 2.9 percent ) ( for high debt ; the decline is above 90 percent look separately at thresholds for external debt . In Figure 3, we highlight even greater for the average growth rate, which ( public and private ) falls to 1 percent. With much faster population the connection between gross external debt as - growth than the advanced economies’, the impli reported by the World Bank and growth and ( or inflation. As one can see, the growth thresholds cations for per capita GDP growth are in line - with those shown for advanced econo for external debt are considerably lower than the worse ) mies. The similarities with advanced economies - thresholds for total public debt. Growth dete end there, as higher debt levels are associated riorates markedly at external debt levels over
5 O f D A tIME IN EB t G RO wth VOL. 100 NO. 2 577 1 7 have shown that public levels of debt/GDP that GDP growth bars, left axis ) ( 5 5 . push the 90 percent threshold are associated with 5 These obser lower median and average growth. - 6 1 I n f t a l o n i vations, however, present only a partial picture of (line, right axis) . 5 4 the post-financial crisis landscape. Private debt, 5 1 in contrast, tends to shrink sharply in the after - 5 . 3 math of a financial crisis. Just as a rapid expan - sion in private credit fuels the boom phase of the 4 1 cycle, so does serious deleveraging exacerbate the . 2 5 post-crisis downturn. This pattern is illustrated in Inflation GDP growth Figure 4, which shows the ratio of private debt to 1 3 . 1 5 GDP for the United States for 1916–2009. Periods of sharp deleveraging have followed periods of 1 2 lower growth and coincide with higher unem - 5 0 . ployment. In varying degrees, the private sector Average Median Average Median Average Median Average Median in many other countries ) households and firms ( 1 1 − 0 5 . Debt/GDP Debt/GDP Debt/GDP Debt/GDP notably both advanced and emerging Europe ( ) below 30% 30 to 60% above 90% 60 to 90% are also unwinding the debt built up during the 1 − 0 1 5 . boom year. Thus, private deleveraging may be another legacy of the financial crisis that may Figure 3. External Debt, Growth, and Inflation: dampen growth in the medium term. Selected Emerging Markets, 1970-2009 The 20 emerging market countries included are Notes: Remarks IV. Concluding Argentina, Bolivia, Brazil, Chile, China, Colombia, Egypt, India, Indonesia, Korea, Malaysia, Mexico, Nigeria, Peru, The sharp run-up in public sector debt will Philippines, South Africa, Thailand, Turkey, Uruguay, and likely prove one of the most enduring lega- Venezuela. The number of observations for the four debt cies of the 2007–2009 financial crises in the groups are: 252 for debt/GDP below 30 percent; 309 for debt/GDP 30 to 60 percent; 120 observations for debt/GDP United States and elsewhere. We examine the 60 to 90 percent; and 74 for debt/GDP above 90 percent. experience of 44 countries spanning up to two There is a total of 755 annual observations. centuries of data on central government debt, Sources: International Monetary Fund, w orld Economic inflation and growth. Our main finding is that , World Bank, Global Outlook f inance , and Development Reinhart and Rogoff and sources cited therein. ) 2009b ( across both advanced countries and emerging 90 percent and ( / markets, high debt GDP levels above ) are associated with notably lower growth / outcomes. Much lower levels of external debt 60 percent, and further still when external debt ) 60 percent are associated with adverse levels exceed 90 percent, which record outright GDP ( outcomes for emerging market growth. Seldom declines. In light of this, it is more understand- able that over one half of all defaults on external do countries “grow” their way out of debts. The debt in emerging markets since 1970 occurred at nonlinear response of growth to debt as debt levels of debt that would have met the Maastricht - grows towards historical boundaries is remi criteria of 60 percent. Inflation becomes signifi - niscent of the “debt intolerance” phenomenon developed in Reinhart, Rogoff, and Savastano cantly higher only for the group of observations ( 2003 ) . As countries hit debt tolerance ceilings, with external debt over 90 percent. market interest rates can begin to rise quite sud- Debt: III. Private Sector denly, forcing painful adjustment. Illustration An Of course, there are other vulnerabilities Our main focus has been on central govern associated with debt buildups, particularly if - ment debt and, to a lesser degree, external public governments try to mitigate servicing costs by and private debt, since reliable data on private domestic debts are much scarcer across countries and time. We have argued here and elsewhere 5 It is important to note that post-crises increases in pub - that a key legacy of a deep financial crisis is lic debt do not necessarily push economies into the vulner - rapidly expanding public debt. Furthermore, we / able 90 + debt GDP range.
6 MAY 2010 578 AEA PAPERS AND PROCEEDINGS n d e M i a ra n n e m y o l p m e t u e t 32 0 9 6 4 9 1 9 3 2 1 – 6 1 9 1 00 9 – ea Y s e n i l ec P d D G / t b e d h t i w rs 9.8 7.2 6.7 5.5 ea r y e h rs o l l A t 0 27 t 22 0 n e c r e P 17 0 w o F l s und f f o 12 0 s i ti s t c o t ri i H a s a t s l c s e t a t S d e it n U e h f t o 0 7 0 2 1916 1921 1926 1931 1936 1941 1946 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 Figure 4. United States: Private Debt Outstanding, 1916–2009 ( ) end-of-period stock of debt as a percent of GDP Note: Data for 2009 is end-of-June. h Sources: istorical Statistics of the United States , Flow of Funds, Board of Governors of the w Federal Reserve, International Monetary Fund, orld Economic Outlook , OECD, World Bank, and sources cited therein. ) 2009b Global Development f inance , and Reinhart and Rogoff ( shortening the maturing structure of debt. As Debt.” National Bureau of Economic Research ) 2009b ( Working Paper 13946. emphasize and Reinhart and Rogoff Reinhart, S. numerous models suggest, countries that choose Kenneth and M., Rogoff. Carmen 2009a. “The Aftermath of Financial Cri - to rely excessively on short-term borrowing to fund growing debt levels are particularly vul ses.” American Economic Review , 99(2): - nerable to crises in confidence that can provoke 466–72. S. and very sudden and “unexpected” financial crises. Reinhart, Carmen M., Kenneth Rogoff. At the very minimum, this would suggest that 2009b. t his t ime Is Different: Eight Centuries f of traditional debt management issues should be at olly f . Princeton, NJ: Princeton inancial the forefront of public policy concerns. University Press. S. Rogoff, and Reinhart, M., Kenneth Carmen Miguel A. Savastano. REFERENCES 2003. “Debt Intoler - ance.” - Brookings Papers on Economic Activ Kenneth Rogoff. , ed. William C. Brainard and George L. S. and ity (1) M., Carmen Reinhart, Perry, 1–62. 2008. “The Forgotten History of Domestic