Reinhart and Rogoff’s discredited pro-austerity argument that domestic debt has a negative effect on growth is well known. Less well known—although perhaps an even clearer case of fear-mongering—is their implication that domestic debt puts countries at risk of domestic default if austerity is not imposed (“The Forgotten History of Domestic Debt”; e.g., RR state that their “domestic default” data is needed to “know the danger zone for [domestic] debt levels” (p. 4).
According to Reinhart and Rogoff the danger is both outright (de jure default) and de facto default from inflation and financial repression. However, like their discredited “growth” paper, the “default” argument is also based on the improper use of statistics (inference from poorly used descriptive stats in this case).
* The point of noting spatial-autocorrelation is that it suggests that cases may not be statistically-independent cases and cannot be used in standard inferential or econometric methods. What might be called “event correlation” is more specific, i.e., knowing specifically that cases are not independent. Example: If an alien looked at a world map of social indicators from 1945, it might guess from spatial-autocorrelation that a single event happened in Europe that affected most European countries profoundly even if they knew nothing of Earth history. However, knowing the actual history of Earth events, we know that, for example, Japan should be included in that list of non-independent cases, something someone only looking for spatial-autocorrelation might miss. Knowing the true level of interconnectedness of events is more powerful than simple spatial correlation. Serial correlation is the non-independence of events across time. E.g., it is hard to consider the series of economic crises in Argentina as truly independent. There is likely some set of factors (political, historical, institutional, intellectual/educational/academic, structural) that come together to cause the repeated similar crises in Argentina or Brazil.
Problems with their de jure list: Pre-1971 monetary systems are fundamentally different from post-1971 systems and cannot be included in a dataset analyzing the latter. Using only relevant post-1971 cases reduces the de jure dataset from 68 to 43 cases. Six of the cases are for currency users, which are irrelevant for a dataset meant to understand modern currency issuers. This reduces the de jure dataset from 43 to 37 cases. Six more are for foreign-currency-denominated debt, making those countries de facto currency users and again not relevant to a study of domestic default, reducing the list from 37 to 31. For the entire list there is a high degree of unaccounted-for serial and spatial/event* correlation and non-independence of cases. Accounting for the non-independence of cases leaves a small number of cases such as Myanmar, Venezuela, and Zimbabwe. It is doubtful they are lessons of “debt” rather than of governance. The reader can be the judge of what lessons these cases offer to the rest of the world.
What appears to be their de facto list
is also marked by serial correlation and non-independence of cases.
In reviewing their de jure data I noticed (on the website that includes the data for This Time Is Different) that within the excel sheets there is a worksheet that opens to the title “DOMESTIC DEBT: DEFAULT AND RESTRUCTURING, 1800-2012” (it appears on Table 7.2, 7.3, 7.4, as the fourth worksheet of each, link to excel file). It is somewhat different from their de jure list mentioned above; at first I thought it must be a list of the de facto cases they mention. However, there is some overlap with the de jure cases on the list and no explanation of what the list is that I can find (I emailed the authors but have yet to receive a response). But it does seem to list many (non-de jure) “hyperinflationary” episodes and episodes they define as “financial repression” as well as listing a number of cases qualified as “Not counted as a sovereign default,” all of which make me think it is a list of their “de facto” category. I include it here alongside the de jure cases as it seems to capture many of the de facto cases RR reference and is presumably what they base their inflation/financial repression arguments on. It lists 58 cases from 30 countries.
Accounting for this reduces that list from 58 to a handful of cases. Like the de jure list, the remaining cases have unique and severe internal problems, and many indeed are the same as from the de jure list. And again, the reader can be the judge of what lessons these remaining cases offer to the rest of the world. It is also doubtful they are lessons of “debt” rather than of governance.
Note: [bracketed] notes on all lists are mine; all “de facto” examples are in dark blue.
Also: Nersisyan and Wray 2010 and Bill Mitchell’s 2010 post already offer comprehensive critiques of Reinhart & Rogoff (and well before the 2013 paper demonstrating that RR’s “Growth in a Time of Debt” stats were bad). However, the continued influence of Reinhart and Rogoff’s defective research continues to be sufficiently widespread that highlighting further flaws in their work is merited.
Austerity in a Time of Plenty
“The forgotten history of domestic debt has important lessons for the present.”
Reinhart and Rogoff, 2014
“it remains to be seen whether all advanced countries have permanently ‘graduated’ from outright debt default”
Reinhart, Reinhart, and Rogoff, 2015
“it would be folly to assume that current favorable conditions will last forever, or to ignore the real risks faced by countries with high and rising [domestic] debt.”
Kenneth Rogoff, 2019
It is well known that the claim that public “debt” ratios impact growth—a claim told in an attempt to impose austerity in a time of plenty—is false. For currency-issuing countries there is no logical reason why the stock of savings in one’s currency (under the misnomer “debt”) should relate to yearly levels of economic activity in one’s country, and empirical data shows that indeed it does not.
Promulgators of the discredited debt-to-GDP ratio view (notably Carmen Reinhardt and Kenneth Rogoff) also influentially promote a related view, and for the same reason: That the stock of savings in a currency somehow puts a country at risk of “default,” high future taxation, financial repression, and inflation/hyperinflation. So we must fear domestic “debt” not only because of effects on growth, but also because “default” or “hyperinflation”—even in “advanced” countries—is looming if austerity is not imposed. They widely spread this implication in interviews and presentations (constantly mentioning default, “inflation trauma,” “painful restructuring” etc.).
In “The Forgotten History of Domestic Debt” (NBER 2008, revised 2010, published version 2011) they include a dataset of “68 cases” of de jure “domestic default.” (The updated 2010 NBER data is used here; the exact number of cases depends on how date ranges, serial “cases,” the “voluntary” cases of the UK and several other factors are counted).
RR count as “de jure”:
“1. A failure to meet a principal or interest payment on the due date (or within the specified grace period). (These episodes also include instances where rescheduled debt is ultimately extinguished on less favorable terms.) 2. The freezing of bank deposits and or forcible conversions of such deposits from dollars to local currency. 3. The abrogation of indexation clauses, as the United States did in the 1930s and as Argentina is doing at the time of this writing in 2008.” (2010, Appendix III, p. 42)
In the main text “de jure default” is defined as “ranging from forcible conversions, to lower coupon rates, to unilateral reduction of principal (sometimes in conjunction with a currency conversion), to suspensions of payments.” (2010, p. 12)
Reinhart and Rogoff also repeatedly suggest that there are many more cases of de facto default (these cases will be colored dark blue throughout), mostly defined as “choosing” (2010, p.2) inflation/hyperinflation and/or “imposing” “financial repression.” (pp. 13-14).
Irrelevance of pre-1971 cases in the “de jure” dataset
It wasn’t until 1971 that the centuries-long evolution of monetary systems of countries that had either previously borrowed external currencies, used currencies convertible to gold or silver, and/or pegs to foreign currencies finally moved to the standard for modern monetary systems: Self-issuing of currencies not tied to any commodity, pegged to any other currency, and with floating exchange rates.
†† This practice and the interest payment are purely vestigial practices in fully modern monetary systems, which can easily be left at nominal zero permanently; currency issuers have absolutely no need for bond sales for “funds,” nor do bonds have any useful effect on inflation (because 1. purchasers are eager savers by definition 2. bonds are highly liquid 3. interest on bonds relentlessly pumps high powered money into economies and 4. businesses price-in rate increases). When a bond is purchased, the saver has reserves held at the Central Bank account (by a bank) switched to what amounts to a savings account at the same Central Bank. When the bond is paid off, this process is simply reversed. It can always be reversed instantly and for any quantity, and this reserve/bond swap could be done away with permanently with no effects on inflation nor the spending capacity of a currency issuer.
In such a system it is not possible for a currency issuer to be illiquid or insolvent in their own currency. What is called “public debt” (regardless of whether the holders are foreign or domestic) in this system is merely the savings of currency units that are voluntarily converted to “bonds” (bonds/ bills/ notes/ gilts/ treasuries etc.) by savers simply because the country pays (also voluntarily) interest on “bonds.” This is a purely vestigial practice. The sale of bonds from currency-issuers merely transforms one already existing government token (reserves) into another government token (the bond).††
Pre-1971 pegged and convertible currencies can tell us nothing about the liquidity/solvency of countries with modern free-floating, non-convertible currency systems.
Of Reinhart and Rogoff’s ~68 “de jure domestic default” cases, only ~43 (as RR sometimes use wide date ranges as one case, e.g., Dominican Republic 1975–2001, and sometimes count a few years as multiple cases, e.g., Argentina 1982, 1989–90, 2002–2005, it makes exact counts difficult) are after 1971 and can possibly be of any relevance to modern countries.
However, let’s consider all 68 “de jure” cases for a moment.
Irrelevance of Currency Users (As Opposed to Currency Issuers)
Eight of Reinhart and Rogoff’s 68 cases are not for currency issuers at all but rather currency users (a ninth “case” is hard to classify: “United States (9 states) 1841–1842” which is in the “free banking” era, with a complex relationship between US-State banks, US-State bonds, the US dollar unit, and gold/silver). These eight cases have no place in a dataset purporting to show a danger of default for currency issuers.
The eight currency-user cases (plus one “US States” “case”):
- Cameroon 2004
- Gabon 1999–2005
- Antigua and Barbuda 1998–2005
- Grenada 2004–2005
- Dominica 2003–2005
- Panama 1988–1989
- US States (9) 1841–1842, [in the “Free banking” era 1837-1862; relationship of State-bank issue to the US dollar, gold/silver is complex].
- United States (“states and many local governments”) 1873–83 or 1884 [after the free banking era, but before the 1913 creation of the Federal Reserve]
One Canadian Dollar User
- Alberta, Canada April 1935
Taking these eight obviously irrelevant cases out of the dataset reduces the RR cases from 68 to 60 (It reduces the post-1971 list from ~42 to ~36).
Similarly, although a currency issuer with a free floating non-convertible currency can never be illiquid or insolvent in its own currency (no matter who holds the debt), it can with debt denominated in a foreign currency. It is necessary to separate foreign-denominated debts from domestic-currency-“debt” cases. Yet RR include six cases of dollar-denominated debt as “domestic defaults.” The cases are:
- Argentina 1982, 1989–90, 2002–2005
- Bolivia 1982
- Mexico 1982
- Peru 1985
Interest Rates & “Financial Repression” (relevant to the “de facto” list)
† This potential exists from whenever a country begins emitting its own currency unit that is valued only because it is the tax-unit (and legal tender for obligations such as fees, fines, and contracts) of that country, and with no limits whatsoever especially after 1971. Currency-issuers have no need for funding-related bonds/bills/notes/gilts and thus zero need to pay interest on bonds to make sure they sell. This is done for purely vestigial reasons. (Until 1971, especially in wartime, conversion of reserves to non-convertible bonds was potentially useful to prevent a “run” on conversion at high wartime spending levels; reserves were still promised to be convertible to gold internationally, whereas bonds were only promised to be converted to reserves. With no convertibility of reserves at all after 1971, bonds do not even serve this purpose anymore).
Before moving on, it is important to point out that post-1971 currency-issuing countries set their base interest rates wherever they want them, and do this by propping the base rate up above zero by either paying interest on bonds or on reserves (pre-1971 could do as well; sovereigns could annul any convertibility promise at will if causing problems with desired interest rates†). If a government does not sell bonds with an interest rate above zero, and does not pay interest on reserves, then the rate of return on saved currency will always fall to nominal zero.
The only way savers “earn” any interest whatsoever on savings of a government currency-unit—whether Treasury bonds or reserves—is through government intervention, propping rates up above zero. The idea that governments were “keeping rates artificially low” and calling this “financial repression” illustrates a fundamental lack of understanding of how modern monetary systems function.
None of the cases on the RR “de facto“ list that claim “financial repression” based on interest rates should be there. However, we’ll also proceed with that dataset for a moment.
Non-independence of cases
Losing Wars and Crazy Dictators: Not Recommended
Historically, when countries lose a war, they often have trouble re-establishing political stability/credibility, and the regime’s currency struggles to recover, attain, and/or maintain value (either domestic value or exchange rate value). The same difficulty occurs with civil war and revolution (and crazy dictators, for that matter). Indeed, the political stability of a regime seems to be the most important factor—by far—influencing the degree to which its currency has domestic value (i.e., the price level is stable). (Even more than the usually simultaneous factor of war or corruption destroying the productive capacity of a country; productivity loss can be of a scale to explain high inflation, but seldom of a level high enough to account for the almost total loss of domestic value of a currency).
Imagine you see 30 people flee from a restaurant. You would think, of course, that this constitutes 30 independent cases of “deciding to flee.” But if you discovered there was a large kitchen fire you might think of it as a single case: 1 “restaurant fire,” with all 30 “fleeings” being from that single cause. This captures the way RR fail to deal with the non-independence of their examples.
To extend the analogy: If you learned that many of the customers, upon seeing there was a kitchen fire, perished because instead of fleeing, they instead rushed into the kitchen and doused a large grease fire with water (a very foolhardy thing to do) making it much worse, you might be confused. But not if you learned that all those who had perished had graduated together from the infamous “Fall T. Óeires School of Firefighting.” Besides the non-independent cases first mentioned, this is what RR do with their later cases: the countries involved all share advisers that might as well have studied at good ole “Faulty Wires U.” That is to say, they have dangerously flawed misconceptions, acquired from Chicago, Harvard, MIT, etc., about how monetary systems function. Their advice is virtually guaranteed to turn a difficult situation (“less developed countries”) into a multiple fatalities situation (RR’s IMF “serial defaulters”).
As political stability is threatened or declines, the value of a currency declines. A new regime after a civil war, revolution, or losing a war has a hard time getting its currency accepted as valuable even by its own citizens, or sees the previous value collapse. This is, absurdly, called “hyperinflation,” as if price rises and “printing” somehow precede (and cause) the loss of value of the currency system. Price increases and “printing” (if the latter occurs) are always a sign of and response to a collapsing currency, which is in turn the result of a collapsed or collapsing government (often coincident with sharp declines in production).
The RR de facto list, defined in part by cases of “hyperinflation,” includes, as an example, these seven cases:
Greece 1941-1944 [occupied by Germany 1941-1944]
This illustrates the association of losing a war and “hyper” loss of value (i.e., collapse) of currency. Other countries on the RR list that were profoundly affected by WWII and economically unstable after it were Russia 1947 and China 1946-1948 (Russia was technically not a loser but probably suffered economic losses greater than any of the other “winners;” China technically was a “winner” but (re)entered into its own civil war, 1945-1949).
These “nine cases” are deeply connected by the single underlying event of World War II. It is hard to consider them as meaningfully independent cases.
Similarly, on the de jure list we find:
Austria December 1945
Germany June 20, 1948
Japan March 2, 1946–1952
Again, these cases are dubiously “independent.” We really have one global underlying cause leading to the almost total undermining of the political legitimacy of World War II losers (and unimaginable economic destruction, especially of the otherwise “winning” Russia).
Reinhardt and Rogoff give the same treatment to World War I and its aftermath (on the “de facto” list), ignoring the high degree of non-independence of the cases:
And even further back:
Mexico 1850 [Mexican American war (ended 1848)]
They also list as many “independent” cases nations affected by the Great Depression:
New Zealand 1933
United States 1933
The above examples were from the de facto list. The similar list of countries on the de jure list:
New Zealand 1933
United States 1933
United Kingdom 1932
Uruguay November 1, 1932 February, 1937
War, Revolution/civil war/regime change:
United States January 1790
Mexico November 30, 1850
Confederate States of America 1864-1865
Denmark January 1913
Russia December 1917–October 1918
China March 1921
Spain October 1936–April 1939
Angola 1976, 1992–2002
Sri Lanka 1996
Sierra Leone 1997–1998
Mongolia 1997–2000 [previously a Soviet satellite state]
Solomon Islands “1995”*–2004 [*default appears to actually be from 1999, IMF Country Report No. 04/258; civil war: 1999-2003]
(This leaves pre-1971 de jure cases of: The United Kingdom 1672 [Stop of the Exchequer], Argentina 1890, United Kingdom 1749, 1822, 1834, 1888–89 (“these restructurings appear to be mostly voluntary”), Peru 1850, Bolivia 1927, Russia 1957).
This leaves nine post-1971 potentially independent de jure cases:
|Myanmar 1984, 1987||Congo (Kinshasa) 1979
|Dominican Republic 1975–2001
El Salvador 1981–1996
Venezuela 1995–1997, 1998
It leaves the following de facto cases:
Nicaragua 2003, 2008
When we look at all of the Reinhart/Rogoff data taking into consideration the real events driving their data, it becomes clear they have merely listed the losers or most impacted from World War II, the losers/most impacted from World War I, countries that have had revolutions, civil wars, and crazy dictators, the Great Depression, and lastly those that have taken IMF advice and/or were advised by mainstream economists (often native but taught, in US/European universities, to take on foreign-currency-denominated debt, that pegs are useful, etc.).
RR haven’t given us a statistical case for associating domestic debt with default. They have given us empirical data that make emphatically clear that unstable political regimes cannot maintain the value of their currency, moved off pegs, moved off the gold standard, and suffered general monetary, institutional, and governance chaos and bad advice in a multitude of ways. None of these cases is relevant to modern politically stable countries with non-convertible, free floating currencies.
1) Modern currency-issuers cannot suffer from illiquidity or insolvency.
2) Hyperinflation is always a political (or war-related) and/or foreign-currency-denominated-debt phenomenon. Relatedly: There are no cases where government spending has induced high- or hyper- inflation in modern peacetime politically-stable countries; it is theoretically possible but does not happen in practice. In modern peacetime politically-stable countries, a decline in the value of a currency is always either supply-side related, preceded by a decline in political stability, or (less commonly) tax-cuts in a boom (political).
3) “Financial repression” as related to interest rates does not work the way Reinhart and Rogoff think it does, indeed it cannot work the way they think it does. They demonstrate a shocking level of incompetence with their interest-rate financial repression claim.
4) The idea that current resource use incurs future financial constraints so that “taxes will be higher” is a non sequitur. Real-resource decisions never put any financial burden, whether current or future, on a currency issuer. The stock of saved currency (under the misnomer “debt”) is residual to balancing the real resources of an economy. Government money is an organizing tool created by the public for the public, and can only be judged on its effectiveness at achieving public goals; the liability side of the tokens themselves are completely meaningless to a currency issuer.
Herndon, Thomas, Michael Ash, and Robert Pollin. 2013. “Does High Public Debt Consistently Stifle Economic Growth? A Critique of Reinhart and Rogoff.” University of Massachusetts, Amherst, Political Economy Research Unit (PERI). Working paper no. 322.
Mitchell, Bill. 2010. “Hyperbole and outright lies.” Monday, March 1, 2010.
Nersisyan, Yeva and L. Randall Wray. 2010. “Does Excessive Sovereign Debt Really Hurt Growth? A Critique of This Time Is Different, by Reinhart and Rogoff” Levy Institute of Bard College, Working Paper No. 603.
Reinhart, Carmen and Kenneth S. Rogoff . 2009. This Time Is Different: Eight Centuries of Financial Folly. Princeton University Press. (Note Part III of this contains a version of their Domestic Debt paper, thus the review by Mitchell, and Nersisyan and Wray, cover both the “growth” arguments, notably before the revelation of the bad RR stats in 2013 by Herndon et. al., and those reviews cover the domestic debt arguments by RR I revisit here.)
Reinhart, Carmen and Kenneth S. Rogoff . 2010. “Growth in a Time of Debt” American Economic Review: Papers & Proceedings 100 (May 2010): 573–578.
Reinhart, Carmen and Kenneth S. Rogoff 2008/2010/2011. “The Forgotten History of Domestic Debt.” NBER 2008, revised NBER 2010. 2011 in The Economic Journal, Vol. 121, No. 552, Conference Papers (May 2011), pp.319-350.
Reinhart, Carmen, Vincent Reinhart, and Kenneth Rogoff. 2015. “Dealing with Debt” Journal of International Economics 96, Supplement 1 (July): S43-S55.
Rogoff, Kenneth. 2014. “What have we learned from Argentina’s debt default?” World Economic Forum.
Rogoff, Kenneth. 2019. “Modern Monetary Nonsense.” Project Syndicate.
Wall Street Journal article and the “Database of Sovereign Defaults” (Clint Ballinger)
and The Autocorrelation of Hyperinflation (about 7 events, not 58) (Clint Ballinger)
These maps illustrate 1) the robust (multiple measurements) correlation of weak states with the inability to maintain the domestic value of their tax-credit (currency) and 2) the high degree of correlation along lines of similar institutions, geographic/real resource constraints, colonial history, and IMF interference with poverty and post-1971 examples from Reinhart & Rogoff’s lists.
Fragile States Index, Fund for Peace
Political Stability and Absence of Violence/Terrorism, Government Effectiveness: Worldwide Governance Indicators (WGI), World Bank Group
Corruption Perceptions Index, Transparency International
Check out my new book “1000 Castaways: Fundamentals of Economics”
A renegade band of Modern Monetary Theorists has overturned mainstream economics in part by emphasizing that there is not one, but two systems of modern money, the “vertical” and the “horizontal.” They conclusively demonstrate how unifying our understanding of these is crucial for grasping modern economics.
“the key to understanding Modern Monetary Theory is this vertical-horizontal relationship”
1000 Castaways develops Mosler’s statement into a concise, book-length treatment that is accessible to all readers, starting from first principles and, step-by-step, leading the reader up to the complexities of the real world.
Our one thousand castaways develop, before our eyes, a “perfect” economy, and demonstrate how the horizontal and vertical systems of money naturally emerge from even more fundamental organizational needs of a large society.
1000 Castaways then contrasts the Island’s “economics” with real-world “economics,” in an enlightening illustration of the last few steps in our common economic understanding that we must take in order to run our modern economies in a way that maximizes wellbeing.