Summary: Thirteen years ago, the world's top economics journal "American Economic Review" published a paper on the relationship between debt and economic growth, which caused a strong response in European and American political circles, and politicians have cited the paper as the basis for fiscal austerity policy.
Ten years ago, a graduate student accidentally discovered a major error in the paper while doing an econometrics homework, which caused a "bloody storm" in the American economic academic community Two Nobel laureates in economics and the former U.S. Treasury Secretary and former Governor of the Central Bank of India were involved, some of them published more than a dozen op-eds criticizing the paper's data handling errors, some of them did not shy away from protecting their old colleagues, some of them believed that the other party had turned normal academic discussion into personal attacks, and the founder of MMT said that the authors of the paper should "get out of the university office and face the real world...... However, the most surprising thing was that the editor of AER went so far as to say that it was a paper that had been published without peer review......
In recent years, as China's local government debt problem has become increasingly prominent, this seriously erroneous paper has been cited in a large number of articles by Chinese scholars on debt issues, involving many well-known scholars and top journals.
If erroneous theories are deeply rooted in the hearts of the people, the damage to the real world will be incomparably great. At a time when China's local debt problem has attracted widespread attention, the New Economists Think Tank will take you back to the extensive and decades-long academic debate in the U.S. economics community, hoping to provide useful enlightenment for China's current solution to the local government debt problem.
The New Economist Think Tank will continue to track the citation of this paper in China. If you have any comments, please send an email to [email protected].
The author of this article is the New Economist Intelligence Unit
Updated | Academic Ethics Supervision: Reinhardt and Rogoff's Paper on the Negative Impact of Debt on Growth Has Major Errors, Calls on Domestic Academic Circles to Stop Citing (30,000 Words Long Article)
1. Origin: An accidental discovery made by a student while doing an econometrics assignment
On January 4, 2010, the Atlanta Marriott Hotel was held at the annual meeting of the American Economic Association.
Carmen Reinhart, then director of the Center for International Economics at the University of Maryland, and Kenneth Rogoff, former chief economist at the International Monetary Fund (referred to as RR), published a research paper titled "Growth in a Time of Debt." The core points of the paper are:
When a country's public debt-to-GDP ratio exceeds 90%, the country's economic growth rate will decline rapidly.
Pictured: Kenneth Rogoff and Carmen Reinhardt
The paper was published in American Economic Review: Papers & Proceedings (May 2010: 573–578) in May 2010. (Note: AEA: Papers and Proceedings is published annually and contains the AEA Proceedings and selected papers presented at the AEA meeting of the ASSA Annual Meeting.) In 2018, AEA Papers and Proceedings became an independent journal and ceased to be the May issue of ER. Although it will continue to be published in May, it will now be called AEA: Papers and Proceedings)
The RR's paper had a wide and far-reaching impact when it was published, with austerity advocates – advocates of fiscal austerity and immediate and drastic cuts in government spending – citing the paper's purported findings to defend their positions and attack critics.
In fact, RR's impact on public debate is probably more direct than any previous paper in the history of economics. Paul Ryan, chairman of the US House of Representatives Budget Committee and former vice presidential candidate, Olli Rehn, the top economic official of the European Commission, and the editorial board of The Economist magazine, among others, see the 90% threshold as a decisive argument in favor of austerity.
Both academics and policymakers want to learn more about RR's thinking, and Thomas Herndon, a PhD student at the University of Massachusetts Amherst, is one of them.
图：Thomas Herndon
At the time, Thomas Herndon's professor at the University of Massachusetts Amherst gave his graduate class an assignment to choose an economics paper and see if the results could be reproduced. This is undoubtedly a good econometric Xi for aspiring researchers.
The 28-year-old from Austin, Texas, chose RR's Growth in a Time of Debt.
This paper is one of the most politically influential economics papers of the last decade, but no matter how Herndon tries, he has not been able to replicate the results of RR.
"My heart sank," he said, "and I thought I might have made a serious mistake." Because I'm a student, it's just me who makes the mistake, not the famous Harvard professors. ”
His professors were also convinced that he must have done something wrong.
"I remember having a meeting with my professor, Michael Ash, and he basically said, 'Come on, Tom, it's not too hard — you just have to go and fix this. ’”
Herndon checked his work repeatedly, but by the end of the semester, he had not solved the puzzle.
That's when his mentor realized something was wrong.
"We have a puzzle that we can't reproduce the results published in the RR," says Professor Ash. "It really annoys us, it's really a mystery to us". Therefore, Professor Ash and his colleague Professor Robert Pollin encouraged Herndon to continue the project (the three will be referred to herein as HAP).
图：Michael Ash（上）和Robert Pollin
RR did not publish the data on which they relied to reach their conclusions, and in order to find out exactly what the problem was, Herndon emailed RR several times to obtain their data set so that he could compare it to his own work. But an average graduate student with little standing in academia turned to some of the world's top economists for help, and it was naturally difficult to get an answer, until one afternoon, as he sat on his girlfriend's couch, he received a reminder from his email.
"I checked my email and found that I received it
Herndon pulled up an Excel spreadsheet containing Reinhardt's data and quickly discovered a few issues:
In this Excel spreadsheet, the raw data for 20 countries appear in reverse alphabetical order in rows 30 to 49 of the table. "I click on cell L51 and see that they only average rows 30 to 44 and not rows 30 to 49". This was an Excel encoding error, which led them to ignore five countries (Australia, Austria, Belgium, Canada, and Denmark). Of these, three countries (Austria, Belgium, and Canada) have time periods of debt/GDP greater than 90%;
"This error is necessary to obtain their published results, and it will go a long way towards explaining why it is impossible for others to replicate these results. If this mistake turns out to be an actual mistake made by RR, then I hope that future historians will be able to point out that one of the core empirical perspectives that provided the knowledge base for the global shift to austerity in the early 2010s was one
Some other data are also ignored in the calculation of RR. In many cases, this is due to the lack of data in some countries in some years. However, data are available for Australia in 1946~1950, Canada in 1946~1950, and New Zealand in 1946~949. Curiously, this data is not included in the calculations. The data for New Zealand is particularly important because the four years that are ignored are four of the five years in which debt/GDP exceeded 90% (1951 was the fifth year). During these five years, New Zealand's GDP growth rates were 7.7%, 11.9%, -9.9%, 10.8%, and -7.6%, respectively, with an average growth rate of 2.6%. RR excludes data from the previous four years, saying that the average growth rate in New Zealand's high debt years was -7.6%.
The calculation of the RR is based on "1186 annual observations, with a large number of observations in each category, including 96 observations that exceed 90%." "You might think that the average growth rate of -0.1% in the 90% category is a direct average of 96 annual observations. This is not the case. Reinhardt and Rogov calculated the average growth rate for each country and then calculated the average of the growth rates for those countries. For example, the UK had a debt-to-GDP ratio of more than 90% for 19 years, and its average GDP growth rate was 2.4% in those 19 years, while since New Zealand was ignored for four years, it only had a debt-to-GDP ratio of more than 90% for one year, and in that year, its GDP growth rate was -7.6%. Based on 20 annual observations, the average growth rate for these two countries is 1.9%. However, RR calculated an average of 2.4% and -7.6% to get an average growth rate of -2.6%. We don't know if RR made a mistake by accident or if it deliberately chose this unusual calculation, but we do at least know that it supports their view. ”
Herndon was stunned. As a graduate student, he had just discovered a serious problem in a prestigious economic study – the academic equivalent of a National Development League basketball player dunking on LeBron James.
"I can barely believe my eyes when I see basic spreadsheet errors," said Herndon, 28. "I thought, am I mistaken? There must be some other explanation. So I asked my girlfriend (who is a Ph.D. in sociology), 'Am I reading it wrong?'"
"My girlfriend, Kyla Walters, replied: 'I don't think so, Thomas. ’”
In fact, the error Herndon found was so great that even Herndon's professors didn't believe him at first:
"At first, I didn't believe him. I thought, 'He's a student, he must be wrong.'" These are brilliant economists, and he's a graduate student," said Robert Pollin, a professor at the University of Massachusetts Amherst. "So we spurred him on and on, and after about a month, I said, 'Goddamn it, he was right. ’”
Overall, the average GDP growth rate for high debt years calculated by and through and as well as the RR is -0.1%. If the missing data are included, and taking into account the number of high debt years,
HAP concludes: "Contrary to the 'RR' view, there is no significant difference between the average GDP growth rate at a time when public debt/GDP is above 90% and the average GDP growth rate at a low public debt/GDP." ”
On April 16, 2013, HAP published a working paper titled "Does High Public Debt Consistently Stifle Economic Growth?" A critique of RR data analysis was published in the Cambridge Journal of Economics in March 2014. They argued that RR's statistical analysis of the data in the original Excel spreadsheet (which was used to support the conclusions of the paper) was flawed: "When using RR's working spreadsheet, we found coding errors, selectively excluding the available data, and unconventional weighting of the summary statistics. ”
They found that high levels of debt were still associated with lower growth – but the most striking results of the RR paper disappeared. High debt is associated with lower growth, but the relationship is much more modest and there are many exceptions.
2. Response: RR's views on the HAP paper
RR emailed a brief response to the incident to many news outlets, and below is the full text of the response:
We have just received this draft comment and will review it in due course. Roughly speaking, Herndon Ash and Pollen also seem to find that when debt exceeds 90%, the growth rate is lower (they found 0-30 debt/GDP with a growth rate of 4.2%; 30-60 with a growth rate of 3.1%; 60-90 with a growth rate of 3.2%; 90-120, 2.4% and above 120, 1.6%). In fact, these results are of an order of magnitude similar to the detailed results by country that we listed in Table 1 of the AER paper and the median results in Figure 2. And they are similar to the estimates for many large countries. and a growing body of literature, including our own attached August 2012 Journal of Economic Outlook paper (with Vincent Reinhart). However, these strong similarities are not what these authors choose to emphasize.
The 2012 JEP document largely anticipates and addresses any concerns about aggregation (the main point of contention here), and the JEP document not only provides an average for individual countries (as we have already described in Table 1 of the 2010 AER document), but also further provides a period-by-period average. Not surprisingly, the results are broadly similar to our original 2010 AER Table 1 average, as well as the equally prominent median results. It is difficult to see how to interpret these tables and the results of individual countries, as they show that the public debt burden of more than 90 percent is clearly benign.
Vincent Reinhart's JEP paper examines all the periods of public debt backlog in developed countries since 1800 in our database. The overall average results show that the growth rate during periods of public debt backlog (more than 90% of GDP for five years or more) is 1.2% lower than that during periods when debt/GDP is below 90%. (A handful of shorter periods are also included in our table.) Note that since historical public debt overstock events have lasted on average over 20 years, the cumulative impact of small growth differences can be considerable.
By the way, we are very careful to talk about "correlation" rather than "causation" in all of our papers, because our 2009 book, This Time It's Different, certainly shows that debt exploded immediately after the financial crisis. That's why we're limiting our focus on longer debt backlogs in our JEP document, although, as noted, the number of short-term debt backlogs is very limited. In addition, we have generally emphasized the 1% difference median result in all discussions and subsequent writing, which is precisely low-key and cautious, and is also an acknowledgment of the results in our core Table 1 (AER paper).
Finally, our 2012 JEP paper cites papers from the Bank for International Settlements, the International Monetary Fund, and the OECD (among others), which almost all found conclusions that are very similar to the original findings, albeit with slightly different thresholds and many nuances to alternative explanations. By the way, these later papers, of course, require further research because the data we developed and used in these studies are new. nevertheless
Carmen Reinhardt and Kenneth Rogoff
April 16, 2013
In a long, detailed response to HAP ("Full Response From Reinhart and Rogoff") via email, RR began by thanking HAP et al. for acknowledging that their paper was flawed:
"We thank Herndon et al. for their careful attention to our original paper published in the AR, Growth in a Time of Debt, and for noting important corrections to Figure 2 of the paper. But it is a reminder that, despite our best efforts to remain cautious at all times, we will redouble our efforts to avoid such mistakes in the future." however
Next, RR specifically responded to HAP's three questions about their paper:
Regarding the first point:
"We reiterate that HAP accurately points out coding errors for several countries that are missing from the averages in Figure 2. They show that our unintended omission has a fairly modest impact on the average GDP growth rate for the grouping in Figure 2 with debt/GDP values between 0-90%. However, it led to a significant change in the average GDP growth rate of the grouping where the debt/GDP value exceeded 90%. However, our median GDP growth rate is correct. ”
《Growth in a Time of Debt》图2（笔者所配）
"Fortunately, our interpretation of the erroneous data points in Figure 2 is somewhat diluted by the same weight given to the median GDP growth rate of countries with different debt levels in our discussion, which HAP selectively ignores. To quote our opening paragraph:
Countries with public debt as a percentage of about 90% of GDP have a median GDP growth rate about 1% lower than other countries, and Table 1 follows Figure 2, but the same problem is not present. Table 1 gives estimates for all public debt/GDP groupings for all individual countries and for a longer period of time than Figure 2, although Figure 2 is also important in our analysis. Fortunately, we have chosen several different ways to present our results, including mean, median, and individual country averages, which are largely standard robustness checks. however
《Growth in a Time of Debt》表1（笔者所配）
On the second point:
HAP went on to point out some other missing debt data points, which they described as "selective omissions." This accusation, which runs through their thesis, is what we most strongly oppose. The explanation for these "absences" is that, at the time of writing, our public data debt set is still missing, a dataset that no one has been able to build before, and which we have now filled in more completely. Many readers of our work have followed this evolution on our data sites. For example, at the time of writing the 2010 AER paper, there was a gap in the French data from the 1970s, which we filled in later. Other data, including New Zealand's pre- and post-World War II data, have just been included, and we have not reviewed the comparability and quality of previous data compared to more recent data. In fact, HAP only knew we had this data when we sent them the files we were using at the time.
We have no problem with HAP's concerns about any data issues related to our work. Once we were able to process New Zealand's data, we had long since fully integrated it into the early 1800s. All major periods of high debt in developed countries since 1800 and their underlying data are included in our paper "Public Debt Overhangs: AdvancedEconomy Episodes Since 1800" published in the Journal of Economic Outlook, co-authored with Vincent Reinhart (Carmen Reinhart's husband).
But HAP certainly doesn't mean to imply that we're manipulating data to exaggerate our results. For what purpose do we "manipulate" the average GDP growth rate of a country with a debt/GDP value above 90% to -0.1%, yet in the same AER paper, we report a median GDP growth rate of 1.6% from 1946 to 2009, and in a longer sample from 1790 to 2009, we report an average GDP growth rate of 1.7% and a median GDP growth rate of 10%? For this reason, why do we make all the data that we have collected and used in our research over the years (detailed, documented from multiple sources) publicly available?
Regarding the third point:
This brings us to the core conceptual question, which HAP believes greatly influenced our findings.
They argue that we used "unconventionally weighted summary statistics". In particular, for each sample grouping, we take the average growth rate for each country and then take the average of the results. It's natural for us, and it's not unconventional. For example, we don't want to place too much emphasis on Greece, which has a debt/GDP value of more than 90% for 19 consecutive years in a sample from 1946 to 2009. Our approach has been followed in many other contexts where one does not want to give undue importance to a few countries that may have their peculiarities. This approach is very clear from Table 1, which also gives the average GDP growth rate for each country.
Our 2012 paper in the Journal of Economic Outlook is based on a longer time period (1800-2011, while Herndon et al. focus on 1946-2009) and provides period-by-period data for each country, including growth rates and years, so the results are fairly transparent. As Herndon et al. argue, the problem of weighting long periods is more serious than that of weighting short periods, and is more pronounced in longer time series (an earlier version of which was also used in Table 1 of our original AER paper). As we noted in our initial comments after receiving the paper yesterday, our JEP paper anticipates much of the controversy and spreads it through the use of a case study approach. This is where this literature is now expanding.
So what does this mean for debt and growth? Did Herndon et al. get very different results on the relatively short post-war samples they looked at? Not really, they also found that the GDP growth rate was lower when the debt/GDP value exceeded 90% (they found that the debt/GDP growth rate was 4.2% at 0-30; 3.1% at 30-60; 3.2% at 60-90; and above 90). In other words, the growth rate at high debt levels is slightly more than half the growth rate at minimum debt levels. They ignore the fact that these results are close to what we get in Table 1 of the AER paper they criticize, and that, despite the coding errors, these results are not far from the median GDP growth rate results in Figure 2. And these results are not much different from what we reported in our 2012 paper in the Journal of Economic Outlook, in collaboration with Vincent Reinhart, where the average GDP growth rate for highly indebted countries was 2.4 percent, compared to 3.5 percent for countries with debt/GDP values below 90 percent. The following table clearly shows the similarity of all these comparisons:
(This picture is accompanied by the original text)
RR also highlights the enormous economic consequences of small differences in economic growth rates over a long period of time:
There is also the question of whether these growth effects will have a huge impact on the economy. Here, without recognizing that the typical period of high indebtedness lasted more than a decade (an average of 23 years across the sample), it would be a mistake to assume that the difference in economic growth rate of 1% is insignificant. It is completely wrong to say that the difference in growth of 1% for 10 ~ 25 years is very small. If a country grows at a rate of 1 per cent below trend for 23 consecutive years, output will be roughly below trend by 25 per cent at the end of the period, with a large cumulative effect.
Looking at the blogosphere's reaction to this comment, we note that this is not the first time that our academic work has been seen as pandering to political views. It's remarkable that this proposition straddles polar opposites! This time, we are accused of supporting austerity with the wrong analysis. Just a few months ago, our findings on the slow recovery from the financial crisis were blamed for justifying the deep recession and economic weakness that the Obama administration has faced since 2007.
HAP wrote a useful paper and helped reconcile the reasons why one result was inconsistent with all the other results presented in our original paper and in our later research, not to mention those they presented in helpful reviews. Clearly, more research is needed on debt and growth, and we welcome all efforts, and this is a very exciting area. We now have debt data for more countries than the original sample and for longer time periods, which allows this study to move forward.
RR conducted their 2010 paper, "Growth in the Age of Debt."
R&R corrected for errors and other issues, arriving at a median growth rate of 2.5% for GDP in highly indebted countries, compared to the initial findings of an average growth rate of -0.1% and a median growth rate of 1.6%, and an average growth rate of 2.2% in the HAP analysis.
After rechecking the data, R&R made some changes to the 2010 paper. First, they corrected an Excel coding error, which included countries such as Australia, Austria, Belgium, Canada, and Denmark, which were missing from all debt level studies, though the most notable change was an increase in the median economic growth rate of +0.3% for 90% of countries. In addition to this coding change, they found several blank cells containing non-zero entries that had an impact of 0.1%-0.2% on the outcome in either direction.
R&R has also made two other revisions based on data obtained since then. The first is the accession of Spain between 1959 and 1980, a low-debt country with high economic growth during this period, which does not affect the 90% threshold. The other is the revision of the New Zealand data. Two years (1948 and 1949) were added (coding corrections) and the GDP growth rate data was changed to reflect a better source. Official GDP data did not exist in New Zealand before 1955, so the authors used data from the work of Angus Maddison in their original paper. However, the GDP data from the New Zealand Bureau of Historical Statistics seems to be more in line with the country's economic history, so R&R replaces the Madison data with a different series.
After adjustment, R&R's latest (2013) paper, which uses a full dataset rather than just post-World War II data, found that the median growth rate for highly indebted countries was 2.3%, while the median growth rate for countries with debt ratios between 60% and 90% was 2.6%. In addition to R&R, many other studies using more sophisticated econometric techniques have also demonstrated a negative correlation between high debt and economic growth.
III. Criticism of Earlier Periods
Long before the HAP article came out, many people criticized RR's paper, such as Robert Shiller, the 2013 Nobel laureate in economics and a professor of economics at Yale University.
Robert Shiller
罗伯特·席勒 (Robert Shiller)在报业辛迪加（Project Syndicate）上他的专栏文章《Debt and Delusion》中写道：
A paper written last year by Carmen Reinhart and Kenneth Rogoff, titled "Growth in the Age of Debt," was widely cited for its analysis of 44 countries over 200 years, which found that countries suffer when government debt exceeds 90% of GDP and growth slows, with annual growth falling by about one percentage point.
One can be misled into thinking that because 90% sounds very close to 100%, terrible things will start to happen in a country that is in such a mess. But if you read their paper carefully, you will see that Reinhardt and Rogoff chose the figure 90% almost arbitrarily. They chose to divide the debt-to-GDP ratio into the following categories, without explanation: below 30%, 30-60%, 60-90%, and above 90%. It turns out that as the debt-to-GDP ratio increases, the growth rate of all these categories decreases, with only the last category declining.
There is also the question of reverse causation. For countries that are struggling economically, the debt-to-GDP ratio tends to rise. If this is partly the reason why higher debt-to-GDP ratios correspond to lower economic growth, then there is no reason to think that countries should avoid higher ratios, since Keynesian theory implies that fiscal austerity undermines rather than promotes economic growth performance.
The fundamental problem facing many countries in the world today is that investors overreact to debt-to-GDP ratios, fear certain magic thresholds, and prematurely demand fiscal austerity programs. They are asking the government to cut spending while the economy remains fragile. Households are scared, so they are also cutting back on spending, and businesses are discouraged from borrowing money to fund capital expenditures.
The lesson is simple: we should worry less about debt ratios and thresholds and more about the artificial (and often irrelevant) structure of our inability to see these metrics.
Reinhardt dismissed these criticisms as wishful thinking. "We know very well that cause and effect is a two-way street," she said. "But please point me to the events in which our advanced economies, from 1800 to the present, have been heavily indebted and have grown as fast or faster than normal. "But high debt levels are like a weak immune system." ”
As early as July 27, 2010, Krugman criticized the problems with RR's paper in a column in the New York Times, in which they used the example of slow economic growth in the United States at the end of the 40s of the 20th century to illustrate the cost of the debt burden, but ignored the factors of postwar recovery.
Paul Krugman
4. Paul Krugman vs Reinhardt and Rogoff
At 1:30 p.m. on April 16, 2013, Krugman wrote an article in his New York Times column, "Holy Coding Error, Batman," in which he wrote: "The intellectual edifice of austerity economics rests heavily on two academic papers that have been seized by policymakers without proper scrutiny because they say what 'Very Serious People' want to hear." One of them – mainly because in the VSP space it is thought that a definite outcome has been identified – is Reinhardt/Rogoff on the negative impact of debt on growth. Soon, everyone "knew" that terrible things would happen when the debt exceeded 90% of GDP.
"Some of us have never believed this, believing that the observed correlation between debt and growth may reflect an inverse causal relationship. But even it occurred to me that a large part of the so-called results could reflect nothing more than bad arithmetic. ”
"But it seems like that's what just happened. Mike Konczal sums up Herndon, Ash, and Pollin's comments well. According to the review paper, RR mysteriously excludes data on some highly indebted countries that grew well immediately after World War II, which would greatly weaken their results; they use an eccentric weighting scheme in which one highly indebted country's bad growth in one year is as important as another high-indebted country's good growth in many years; and they lose a whole bunch of extra data through a simple coding error. ”
Herndon et al. say that if you solve all these problems, the results will obviously disappear. ”
"If true, this would be embarrassing and worse for RR. but
On April 16, 2013, at 7:30 p.m., Krugman was dissatisfied with RR's response to HAP and added something about the noon content in his column: "...... Their reaction to the new criticism was really, really bad."
First, they argue that another measure – the median growth rate – is not much different from the results of Herndon et al. But that's first and foremost an apple and orange comparison – the truth is, when you compare the results directly, RR looks very wrong. Something is seriously wrong with the paper, pointing out your other results is not a good defense.
Second, they say they like to emphasize median results, which is much more modest than average, but everyone who uses their work likes to cite strong results, and I haven't noticed whether RR has made significant efforts to correct people's notions that debt has a huge negative impact on growth.
Third, they point out that even the cleaned data does show a negative correlation between debt and growth. Yes, but that's where the reverse causality problem arises. More on that later.
Finally, while they acknowledge the problem of reverse causality, they seem to be very much interested in having both – yes yes, we know the problem, but then immediately switch to talking as if debt necessarily leads to slower growth, rather than economic growth.
So about the slow growth/debt relationship: I did a quick and dirty mini-RR for the period from 1950 to 2007 (from 1950, because that's where the aggregate economic database started), focusing only on the G7. If you look at the scatter plot, there does seem to be a correlation between high debt and slow growth:
But I've coded the dots by country – if you look at it, you'll see that most of the obvious relationships come from Italy and Japan, and Britain doesn't seem to have suffered much from high debt in the 50s of the 20th century. Historically, it is clear that both Italy and (especially) Japan have been saddled with high levels of debt due to slower growth, not the other way around.
So it's really disappointing; It's a scary thing when so many things are at stake.
2013年4月17日，克鲁格曼在《纽约时报》的专栏上发表《Further Further Thoughts On Death By Excel》，进一步谈及RR的论文：
There will be some back-and-forth debate about modeling strategies, data selection, and so on, and I'm pretty sure some people will try to say that RR is basically correct. At this point, however, it's fairly clear what the data will say, as the datasets created by others more or less match the data that RR claims to have viewed; For example, here is the growth in developed countries since 1790:
The data show that debt is negatively correlated with growth; However, 90% do not have any red lines. This red line is crucial to RR's impact – the paper would never have had the impact it has now without the "oh my God, we're going to cross 90% unless we do austerity now."
The paper doesn't do anything – it just looks at simple correlations and doesn't make any effort to sort out cause and effect.
On April 20, 2013, Krugman wrote in a New York Times op-ed entitled "The Good Glitch": Now, in fact, coding errors are not the biggest problem. The real takeaway, from an economic point of view, is that RR results are never robust, both because the apparent relationship between debt and growth is rather weak, and because this correlation is clearly at least partly reversed. But economists have been making these arguments for years, but to no avail. A downright, embarrassing mistake shocked the "Very Serious People" and shook their faith in an outcome they were more than willing to believe.
2013年4月26日,克鲁格曼发表了《The Medium Term Is Not The Message》一文,在文中他谈到了知识分子和政策的关系:
This is the real world in which macroeconomic analysis comes into play. The reality is that as an economist, you're either trying to quell the deficit hysteria or helping to exacerbate it.
RR has clearly exacerbated fears. If that's not their intention, it's easy for them to make it clear that they are now opposed to spending cuts, despite the negative correlation between debt and growth. They never did.
I would say that this is part of a broader view: that it is the responsibility of the general public intellectual to talk more than just ideals. I'm not saying you have to draft legislation that can pass Congress, or whatever; I mean, you need to be clear about where you stand on the actual decisions that are being made, not just what we should do but what we won't. Especially when it's clear to you that a lot of people are invoking your work to push for policies that are completely different from your ideals.
For the next month, Krugman criticized RR in a New York Times op-ed every once in a while, and on May 25, 2013, after a month of relentless criticism, RR finally wrote a long letter (8 pages) to Paul Krugman denouncing him for the uncivil and unprofessional way he had treated them in writing.
The letter is quite long, running not only through Krugman's tone, but also through his past work and their past work to defend the work they continue to do.
The following is a section of the letter in addition to the appendix:
Cambridge, Massachusetts, May 25, 2013
Dear Paul,
Back in the late 80s, you helped shape the concept of emerging market debt overhangs. The financial crisis has exposed the fact that the boundaries between emerging market countries and developed countries are not as clear as one might think. In fact, it's a recurring theme in our 2009 book, This Time Is Different: An 800-Year History of Financial Crises. Today, the growth constraints of the developed countries on the periphery of the eurozone have much in common with the growth constraints of the emerging market economies of the 80s of the 20th century.
We admire your past academic work that still influences us today. We are therefore deeply disappointed by your extremely uncivilized behaviour over the past few weeks. You attack us in a very personal way almost non-stop in New York Times columns and blog posts. Now you're doubling down on the New York Review of Books, adding accusations that we didn't share data. Your description of our work and policy impact is selective and superficial. It is grossly misleading about our position on these issues. We would like to say that your logic and evidence on the substance of the policy is not as convincing as you suggest.
You are specifically addressing our 2010 paper on the long-term link between high debt and slow growth. You have the right not to agree with our interpretation of the results. However, it is disturbing that you have completely ignored the follow-up literature, including the work of the IMF and our own in-depth and more complete paper published in 2012 with Vincent Reinhart. Perhaps, acknowledging the latest literature – not to mention decades of theoretical, empirical and historical contributions to the evils of high debt – would inconveniently undermine your attempt to scapegoat us for austerity. You write: "In fact, Reinhardt-Rogoff may have had a more direct impact on public debate than any previous paper in the history of economics." ”
This hyperbolic exaggeration aside, you never seem to mention our other work, which is certainly much more impactful in terms of dealing with the financial crisis. Specifically, our 2009 book (released before our growth and debt work) shows that recovery from a severe systemic financial crisis is long, slow, and painful. As you yourself have admitted on more than one occasion, this is not at all common sense in front of us. During the crisis, and certainly into 2010, policymakers around the world were using our research, as well as their assessments, to help justify sustained macroeconomic easing in monetary and fiscal policy.
You want to blame the positions of some politicians on our paper from late 2010, but you fail to recognize a basic reality: we support radically different policies. Anyone with experience on this issue knows that politicians may cite an academic paper if they serve their purposes. But when authors come up with very different policy conclusions, the policy traction they can gain through this means is limited. You can refer to the appendix to this letter for our views on policy during the financial crisis, which we have publicly expressed in real time. We are not silent.
Very senior former policymakers, after observing the attacks of the past few weeks, have forcefully explained that real-time policies are rarely largely driven by a single academic paper or results.
Notably, in the past, arguing critics have often blamed the IMF's work with countries with temporary or permanent debt sustainability issues. Since its inception after World War II, the IMF's programs have almost always involved some combination of austerity, debt restructuring, and structural reform. When a country that has been running a large deficit suddenly can't borrow any new money, there is always a need for some kind of adjustment, and one of the IMF's usual roles is to act as a lightning rod. Even before the creation of the IMF, a long period of self-sufficiency and hardship was accompanied by a debt crisis.
Now let's turn to the substance. The events of the past few weeks have not changed the basic facts and fundamentals.
First, advanced economies now have higher debt levels than most, if not all, of historical periods. This is public debt and private debt (which is usually made public as the crisis unfolds). In most cases, a significant portion of these debts are held by foreigners, with the notable exception of Japan. In Europe, where exposure to external debt (both public and private) is greatest, financial deglobalization is in full swing. Debt financing has increasingly become a domestic business, which is a difficult one in the context of limited total domestic savings. As for the United States: our only brief period of high debt was World War II-era debt, which was held by domestic residents, not fickle international investors or central banks in China and elsewhere around the globe. This observation does not mean that a "panic" is coming, with bond watchers pushing the rest of the world to sell US Treasuries in unison and US interest rates surging sharply. Carmen's work on financial repression suggests a different picture. But many emerging markets have stepped into bubble-like territory, and we've seen this before. We should not take it for granted that their prosperity makes it possible for them to continue to buy US Treasuries on a large scale. Reversal is possible. Smart risk management means planning for these and other contingencies that could disrupt today's global low interest rate environment.
Second, about debt and growth. HAP's paper uses a different approach that reinforces our core finding that high debt levels are associated with low growth. This fact has been hidden in the discussions of the tabloid media and the blogosphere, but the point is obvious, even with a cursory look at the full range of results reported in the paper they criticize. What's more, this finding featured prominently in our 2012 Journal of Economic Perspectives paper on Debt Over hands, co-authored with Vincent Reinhart, but they did not cite it. The main point of our 2012 paper is that the debt backlog period lasts an average of 23 years, despite the difference in annual GDP growth rates between high and low debt scenarios of about 1%. Therefore, the cumulative impact on income levels over time is significant.
Third, the debates of the past few weeks have not changed the fact that debt levels above 90% (even if you completely reject the total central government debt as a generic transnational "threshold") are very rare, and even rarer in peacetime. From 1955 until the latest crisis, less than 10 percent of advanced economies had a debt-to-GDP ratio above 90 percent, and only about 2 percent had a debt-to-GDP ratio above 120 percent. If governments consider high debt to be a risk-free proposition, why would they avoid it so consistently?
Your most recent April 29, 2013 New York Times blog, The Italian Miracle, was designed to highlight how interest rates have fallen in debt-ridden Italy since the ECB took strong steps to play more of a lender of last resort to the periphery. There is no objection here. However, this positive development is intended to reinforce your view that high debt is not a problem (even for Italy) and that the causal relationship is entirely from low growth to debt. You didn't mention that in this miraculous economy, GDP fell by more than 2% in 2012 and is expected to fall by a similar magnitude this year. Elsewhere, you have expressed your conviction that Italy's long-term secular growth/debt problems, dating back to the 90s of the 20th century, are purely a case of slow growth leading to high debt. This statement is very controversial.
In fact, the argument that you have repeatedly expressed that low growth leads to high debt, but vice versa, is hardly supported by the recent literature on the subject. Of course, as we have already noted, this work has been singularly overlooked in public discourse over the past few weeks. The best and worst that can be said is that the results are mixed. Some studies that look at a more comprehensive growth model have found that debt has a significant impact on growth. We mentioned this in the appendix to the New York Times article. Of course, it is well known that economic cycles affect government finances and, in turn, debt (from growth to debt causality). Cyclically adjusted budgets have been around for decades, and this is your superficial description of the relationship between growth and debt.
As for the ways in which debt can affect growth, there is dramatic debt and there is not.
Debt and drama. Do you really think that if a country is suddenly unable to borrow from international capital markets because its public and/or private debt is considered unsustainable, it will not suffer from slower growth and rising unemployment? This credit crunch would have a serious impact on economic growth and employment, whether or not austerity is implemented. Fiscal austerity has strengthened the procyclical nature of the domestic and foreign credit crunch. This pattern is not unique to this period
Policy response to dramatic debt. With regard to policy responses to this deplorable situation, we stress that restoring access to credit is essential for sustained growth, which is why many countries need to write off senior bank debt. Moreover, there is no reason for the ECB to buy only sovereign debt, as buying senior bank debt in the same way that the US federal government buys mortgage-backed securities would help rekindle corporate credit and working capital. We don't think your appeal to fiscal generosity is a substitute for the fact that the European periphery cannot afford fiscal expansion, whereas for Germany, which can afford it, fiscal expansion will be pro-cyclical. Any overheating of the German economy would put pressure on the ECB to maintain a tighter monetary policy, reversing some of the progress made by Mario Draghi. A better way to capitalize on the strength of Germany's balance sheet would be to agree to a faster and deeper reduction in the debt of peripheral countries and to support the ECB in pursuing a more expansionary monetary policy.
No dramatic debt. There are other examples, such as the United States today, or Japan since the mid-90s of the 20th century, where their debt is not dramatic. We have already mentioned in our New York Times article the obvious fact that we know very little about these periods. We deliberately did not include the puzzling cases of 19th-century England and the Netherlands. An important source of funding for these imperial debts came from the large transfer of resources from the colonies. They had "good" high debt for centuries, because their colonies did not. In our 2012 paper, we made some points explaining why debt overhang could be important even if borrowing capacity doesn't collapse immediately.
Bad shocks do happen. How can you believe that with peacetime debt hitting new records in the coming years, the United States will be able to pursue a strong countercyclical fiscal policy in the face of major unexpected shocks? Moreover, do you really want to find out the answer to this question in a difficult way?
The UK did not issue a reserve currency, was more dependent on the financial sector, experienced greater bank failures, did not experience the same shale gas revolution, was more susceptible to European influences, and was clearly more susceptible to dramatic scenarios than the United States. But you often assert that the situation is the same in the United States and the United Kingdom, and that both countries have the option of unlimited fiscal expansion at no cost. Of course, this does not preclude high-return infrastructure investments that leverage public balance sheets, either directly or indirectly through public-private partnerships.
The policy response to debt has not been dramatic. Let's be clear, in our research and numerous review articles, we have discussed the role of higher inflation and financial repression in debt reduction. As our appendix shows, we did not advocate austerity at a time when the crisis was just over and the recovery was fragile. But it has been six years since the subprime mortgage crisis began in the summer of 2007. If it takes another 10 to 15 years to solve the worsening problem, it will lead to a recession, if not necessarily a full-blown debt crisis. The end may not be a loud bang, but a whimper.
The New York Review of Books' allegations are your hasty negligence in failing to examine the facts before accusing us of serious violations of academic ethics. This has been hinted at in your column in the New York Times, where you posted a link to a program that treats misstatements as fact.
Luckily, the "Wayback Machine" crawls the internet and regularly makes bulk copies of web pages.
The debt/GDP database was first archived on the web page of the University of Maryland Carmen in October 2010. Data were transferred to www.ReinhartandRogoff.com in March 2011. It is put alongside our other data, such as inflation, crisis dates, and exchange rates. This data is often sought out and discovered by researchers who are willing to look. This is evidenced by the increased disclosure of debt data by official institutions. It was only after we provided a roadmap in a detailed reference checklist in a 2009 book (and a previous 2008 working paper) explaining how we discovered the data that the IMF began to construct historical public debt data.
Our interactions with academics and practitioners who study real-world issues in our field are ongoing, and our doors remain open. Therefore, accusing us of not sharing our data is a baseless attack on our academic and personal integrity.
Finally, like many other mainstream economists, we take risk slightly more seriously than you do, as debt by all measures – including aged debt, public debt, private debt, and foreign debt – has reached record levels. This conclusion is not based on one or two papers, as you sometimes suggest, but on long-term economic research and extensive historical experience with the risk of record-high debt levels.
You often quote Keynes. We read through Keynes's writings. He wrote "How to Pay for War" in 1940 precisely because he didn't mind huge deficits – even to support a cause as noble as the battle for survival. Debt is a slow-moving variable that can't – and often shouldn't – come down too quickly. But interest rates change much faster than fiscal policy and debt.
You may be right, and this time it may, after all, be different. If so, we'll admit we're wrong. Whatever the outcome, we intend to put the results in their proper context for the sake of a community of academics, policymakers, and civil society.
Dear to you
Carmen Reinhardt and Kenneth Rogoff
Harvard University
2013年5月26日上午,克鲁格曼在他《纽约时报》的专栏撰文《Reinhart And Rogoff Are Not Happy》回应道:
This will probably go on forever and they and I have other things to do. So let me briefly point out – and I hope to be clear – where their analysis is wrong.
In part, it lies in the downplay of the question of causality – whether high debt leads to low growth, low growth leads to high debt, or whether high debt and low growth are the result of the third factor (as in the case of post-war American demobilization, which they emphasized in their original paper).
As everyone in this debate acknowledges, there is a negative correlation between the data on debt and growth. So draw a line at any point – 80 percent, 90 percent, whatever – countries with debt above that level tend to grow more slowly than countries with debt below that level.
However, there is a huge difference between the statement that "countries with debt above 90% of GDP tend to grow more slowly than countries with debt below 90% of GDP" and the statement that "growth falls sharply when debt exceeds 90% of GDP". The former statement is true; However, R&R has repeatedly blurred this distinction and continues to do so in recent writings.
For countries with debt levels close to 90%, such as the United States and the United Kingdom, this distinction is crucial. The difference between the two is that if a tightening program of a few percentage points of GDP is not implemented, GDP will fall by less than one percentage point at most in the decade from now, as indicated by the actual correlation, and that it will reduce future GDP by 10%, which is exactly what the threshold proposition suggests.
Of course, austerity policymakers have seized on the latter statement, citing RR – and if the authors have ever tried to correct this misunderstanding, or in fact, if they have ever admitted that it is, it has also been done very quietly.
I'm sorry, but the failure to dispel this misconception has done a lot of harm – and that hurt hasn't been significantly mitigated by the variety of rhetoric that austerity might have gone too far.
V. Lawrence Summers' view
Lawrence Henry Summers
On May 6, 2013, Lawrence Summers, a professor at Harvard University and former U.S. Treasury Secretary, published an article in the Financial Times titled "The buck does not stop with Reinhart and Rogoff", expressing his opinion on his colleague's paper:
Over the past few weeks, a large part of the controversy, economic commentary, and political debate surrounding a study by my Harvard colleagues (and friends) Carmen Reinhart and Kenneth Rogoff has been drained up. The article, published in 2010, was widely interpreted to mean that a country's economic growth could stall once the country's government debt-to-GDP ratio crossed the 90 percent threshold.
But academics at the University of Massachusetts have proven — and both professors admit — that the two professors inadvertently left out some relevant data when forming their results due to coding errors. Questions have also been raised about how they weight their observations and what data they use.
Many assert that the debate has undermined the arguments of austerity advocates around the world that deficits should be reduced quickly. Some have even blamed millions of people for losing their jobs on Reinhardt and Rogoff professors, claiming that they are important intellectual ammunition for austerity. Others argue that, even after review, the data support the view that reducing deficits and debt burdens is important for most industrialized countries. Still others say the controversy has raised questions about the usefulness of statistical research for economic policy issues.
Where should these arguments be resolved? First, the whole experience should change the way we conduct economic and statistical research. Professors Rogoff and Reinhardt are rightly seen as attentive, honest scholars. Anyone close to the process of economic research or financial markets will recognize that data errors like theirs are very common.
In fact, JPMorgan Chase & Co. conducted an internal investigation into the $6 billion losses it incurred in the "London Whale" deal last year and found that the mistakes it made were no different from those made by Professors Reinhardt and Rogoff. A simple error in the model meant that the bank was vastly underestimating the risks it was exposed to. In the future, authors, academic journals, and reviewers will need to put more effort into replicating important research results before they can be widely disseminated.
More generally, any important policy conclusions should not be based solely on a single statistical finding. Policy judgments should be based on the accumulation of evidence from multiple studies conducted using different methods.
Even so, without an intuitive understanding of the factors driving the model, one should be reluctant to accept the conclusions of the "model". Academics are rightly and understandably hoping that their findings will inform policy debates. But they are obliged to stop those who oversimplify and exaggerate their conclusions, and sometimes even refute them.
Second, all participants in the policy debate should be healthy with skepticism about retrospective statistical analysis. Trillions of dollars in losses, millions of people unemployed, because the lesson learned from the 60-year experience from 1945 to 2005 is that "U.S. home prices in general always go up." It's not a data problem or an error analysis. It's a kind of data regularity – until it no longer has regularity.
Extrapolating future prospects from past experience is always very problematic and needs to be done very carefully. In retrospect, it would be foolish to believe that a threshold could be estimated based on data from about 30 countries, beyond which debt would become dangerous.
Even if such a threshold exists, why should it be the same in countries with and without national currencies, financial systems, cultures, openness and growth experiences? Any trend of high debt and low growth at the same time is likely to reflect the way in which debt accumulates rapidly as a result of slow growth.
Third, while Reinhardt and Rogoff's study, even before the recent revival efforts, did not support the claims of prominent American and British right-wing figures about the urgency of deficit-reduction efforts, the joy that some leftists derive from embarrassment is inappropriate.
It is ridiculous to blame Reinhardt and Rogoff for austerity. Political leaders who implement austerity measures make policy choices first and then look for intellectual support. There may not be a problem of debt exceeding a certain threshold that would become catastrophic, and while the experience of both the United Kingdom and the United States suggests that fiscal austerity is bad for economic growth in a weak economy with near-zero interest rates, it is a grave mistake to think that debt can or should accumulate unchecked.
With all but the most optimistic forecasts, further action is needed in almost all industrial worlds to ensure that debt levels are sustainable after the economic recovery.
Now is not the time for austerity, but we have forgotten our danger that debt-financed spending is not an alternative to cutting other spending or raising taxes. It's just a way to postpone these painful behaviors.
6. Bradford DeLong, former deputy assistant to the U.S. Treasury Secretary: There is no "90% threshold"
J. Bradford DeLong
Bradford DeLong, a professor of economics at the University of California, Berkeley, a researcher at the National Bureau of Economic Research, and a deputy assistant to the U.S. Treasury Secretary during the Clinton administration, published an article on April 29, 2013, titled "Risks of Debt: The Real Flaw in Reinhart-Rogoff?," in which he wrote:
When Carmen Reinhart and Kenneth Rogoff wrote Growth in a Time of Debt, they asked the question: "Large deficits and epic bank bailouts may help deal with a recession, but what are their long-term macroeconomic implications? Reinhardt and Rogoff argue that the threshold for public debt is: "90% of GDP (annually)", beyond which "growth rates will fall, with both developed and emerging economies alike".
The choice of the word, combined with the charts included in their study, was extremely misleading. The Washington Post editorial board recently criticized the way the U.S. budget deficit and government debt are being handled, arguing that "economists believe there is a 90 percent cut-off line, beyond which it threatens sustainable economic growth."
What emerges in Reinhart and Rogoff's paper is a product of Reinhart and Rogoff's nonparametric approach: data are thrown into four ranges, with 90% being the lower bound of the highest range. In fact, as debt-to-GDP rises, growth rates will gradually and steadily decline. 80% looks no different than 100%.
How worrisome is the decline in economic growth due to rising debt? After all, correlation is sometimes causal.
Much of this decline has come in countries where interest rates tend to be higher, government debt is higher, and stock markets tend to be lower. It has nothing to do with the United States today. Much of this decline has come from countries with higher inflation rates when government debt is high. It has nothing to do with the United States today. This decline is more likely to come from countries where growth is already slow, and as Larry Summers often says, the high level of government debt relative to GDP is not the result of the numerator, but of the denominator. It has nothing to do with the United States today.
How much of this correlation is left for a country with low interest rates, low inflation, rising stock prices, and previously healthy economic growth?
Until interest rates and inflation start to rise above normal levels or the stock market plummets, there is little risk of the U.S. accumulating more government debt. There are huge potential benefits to solving our real low employment and overcapacity problems now.
7. L. Randall Ray, a disciple of Minsky and one of the founders of MMT. Randall Wray).
L. Randall Wray
兰德尔·雷接受《加拿大商报》采访时，直言不讳地批评了RR的《Growth in a Time of Debt》，
On why the RR paper has such an impact, Randall Ray said: First, it's the willingness to believe, and everybody wants to believe that government debt is bad. RR seems to have found evidence to back up what they want to believe, no matter how bad their research is. Second, since they refuse to provide data, no one can check it.
The theories in their papers and books are simply untenable, they do not have a sovereign monetary theory...... Like many economists, they do not understand that a sovereign government that issues debt denominated in its own floating currency cannot be forced to default, and that the government can always afford the interest.
"What I would say is that there is a part of their research – only a part – that is defensible. When you have a severe financial crisis, like the global financial crisis in 2008, it does lead to a slowdown in economic growth. However (and this is the key), it is private debt, not sovereign debt, that has triggered this crisis. ”
Finally, Randall Ray advises RR to "step out of the university office every now and then and face the real world." ”
8. Former RBI Governor Raghuram Rajan: Krugman should not question RR's motives
拉古拉姆·拉詹（Raghuram Rajan）
Raghuram Rajan is the Distinguished Professor of Finance at the University of Chicago's Booth School of Business. From 2003 to 2006, he served as Chief Economist and Director of Research at the International Monetary Fund. From 2013 to 2016, he was Governor of the Reserve Bank of India, the Reserve Bank of India, and from 2015 to 2016, he was Vice Chairman of the Board of Directors of the Bank for International Settlements.
拉古拉姆•拉詹（Raghuram Rajan）发表《The Paranoid Style in Economics》一文，反对克鲁格曼对RR的过分指责：
Perhaps the most famous recent example is Nobel laureate Paul Krugman's attack on economists Carmen Reinhart and Kenneth Rogoff, in which he quickly shifted from criticizing a mistake in one of their papers to accusing them of their commitment to academic transparency. ”
For those who, like I know these two brilliant international macroeconomists, it is clear that these accusations should be dismissed immediately. But there's a bigger question: why the paranoid style has become so prominent.
Part of the answer is that economics is an imprecise science, with exceptions to almost all patterns of behavior that economists take for granted. For example, economists predict that an increase in the price of a certain commodity will reduce the demand for that commodity. But economics majors will undoubtedly remember the "Giffen commodity" they encountered in the early days, which defied the usual patterns. When tortillas become more expensive, the poor Mexican worker may eat more tortillas because she now has to cut back on more expensive foods like meat.
This "irregularity" also occurs elsewhere. When the price of a product rises, customers tend to value it more. One reason for this may be its signal value. An expensive hand-mechanical watch may not be more accurate than a cheap quartz watch, however, since few people can afford it, buying it is an indication that the owner is wealthy. Similarly, investors flock to stocks that have appreciated in value because they have "momentum".
The point is, economic behavior is complex and can vary depending on the individual, time, commodity, and culture. Physicists don't need to know the behavior of each molecule to predict how a gas will behave under pressure. Economists could not be so optimistic. In some cases, abnormal behaviors of individuals cancel each other out, making groups more predictable than individuals. However, in other cases, influence between individuals can cause the group to become a group led by a minority.
The difficulties faced by economic policymakers do not end there. An economic system can have different effects, depending on its quality. On the eve of the 2008 financial crisis, macroeconomists tended to ignore the financial sector in advanced economies models. There haven't been major financial crises since the Great Depression, and it's easy to assume that financial pipelines operate behind the scenes.
Such a simplified model proposes policies that seem to work – until the pipeline is back to normal. And the pipeline failed because herd behavior – shaped by policies that we are only beginning to understand now – drowned it.
So why not let evidence, rather than theory, guide policy? Unfortunately, clear evidence of causation is hard to come by. If high national debt is associated with slow economic growth, is it because excessive debt is holding back growth, or is it because slow growth has led to more debt accumulation in the country?
Many econometricians' careers have been built on finding an ingenious way to determine the direction of causality. Unfortunately, many of these methods cannot be applied to the most important issues facing economic policymakers. Therefore, the evidence does not really tell us whether a heavily indebted country should repay its debts, or increase borrowing and investment.
Moreover, seemingly obvious, common-sense policy solutions often have unintended consequences, because the target of policy is not a passive object in physics, but an active agent that reacts in an unpredictable way. For example, price controls, rather than price reductions, tend to lead to scarcity and the emergence of a black market, in which regulated goods are much more expensive.
All of this means that economic policymakers need to be very humble, open to alternatives (including the possibility that they could be wrong), and willing to experiment. This does not mean that our economic knowledge cannot guide us, only that policies that work in theory – or that have worked in the past or elsewhere – should be prescribed with an appropriate degree of self-doubt.
But for economists who actively engage with the public, it is difficult to influence the hearts and minds of the people by validating their own analysis and hedging their own prescriptions. It's best to explicitly claim one's knowledge, especially if past academic honors prove one's expertise. If this leads to a more intense public debate, then it's not an entirely bad practice.
However, the dark side of this certainty is the way it influences how these economists express opposing views. How do you convince your enthusiastic followers if other equally qualified economists hold the opposite view? The public is often kept in the dark instead of promoting public dialogue and educating the public. It also hinders young, less senior economists from entering the public discourse.
In a major study of public and sovereign debt over the centuries, Reinhardt and Rogoff, who have always been very cautious, made a mistake in one of their working papers. This mistake is not found in their 2009 award-winning book, nor in the subsequent widely read paper on the scholarly debate over their work.
Reinhardt and Rogoff's research broadly suggests that when public debt levels are high, GDP growth slows. While there is a reasonable debate about whether this means that high debt is leading to slow growth, Krugman instead questions their motives. He accused Reinhardt and Rogoff of deliberately keeping their data out of the public domain. Reinhardt and Rogoff were appalled by the allegation, which amounted to an accusation of academic misconduct, and they issued cautious rebuttals, including online evidence that they did not hesitate to share data.
To be fair, given Krugman's strong public stance, he has received tremendous personal criticism from many right-wingers. Perhaps the paranoid style of public debate, focusing on motive rather than substance, is a useful defensive strategy against ardent critics. Unfortunately, it also spills over to confront more legitimate differences of opinion. Perhaps a mutually respectful economic debate is possible only in academia, and there is less public discussion about it.
9. AER Editor: The paper has not been peer-reviewed
Back when RR's paper was first published, many people were skeptical of their findings because it was difficult for many people to replicate their results using data. (There is also a lot of doubt about whether there is a causal relationship between high debt and low growth)
It wasn't until the faculty and students at the University of Massachusetts came to RR for analysis that RR's mistakes in the paper were revealed.
People wonder why the American Economic Review, one of the world's most authoritative economics journals, didn't notice the problems with this paper in the first place.
Business Insider interviewed University of Virginia economics professor William Brown. William R. Johnson, editor of the Review, which was first published in the paper.
William R. Johnson
Johnson said the difference between the American Economic Review: Papers & Proceedings and other editions is that the papers are from presentations at the annual meeting of the American Economic Association. The papers were hand-selected by the AEA (American Economic Association) President-Elect in consultation with the committee.
Because of these unusual circumstances, the editing of the Proceeding paper was not strict enough.
"These papers don't go through the normal peer review like AER's other questions. ”
Here, here's what he said, from the introductory part of the edition:
"The data must still be reproducible", but Johnson said that "the prestige of the authors also plays a role" and that "this is true for all AER papers, not just those that appear in the 'conference minutes'"
It is not necessarily the reviewer's responsibility to reproduce the research results.
"It's largely based on trust and people's reputation. The main mechanism for discovering the bug was after it was released - someone tried to reproduce it. ”
Johnson said that if the paper had gone through the normal review process, it might have been caught, but he was not convinced.
"If it's some internal data error, it's hard to detect without a reproduction. ”
10. Domestic media reports
After HAP published its questioning of the RR paper and caused an uproar in the U.S. economic academic community, the Chinese media quickly took notice of the incident and reported on it.
The article "Doctoral Students Challenge Harvard's Authority, Government Debt Problem Heated Discussion" was published, which briefly reviewed and analyzed the incident.
Published an article titled "Harvard Economist Asserts High Debt Leads to Slowdown in Economic Growth, Alleged Excel Coding Error" (a translation of Krugman's column in the New York Times), which reviews some of the criticisms of the RR paper.
The China Financial Information Network released "Harvard Professor: It is better to rely on debts than to repay money", reviewing the latest developments of the incident.
The publication of "The Founder of the 'Debt Threshold' Theory Insists 'Right'" summarises RR's latest response to the incident in the Financial Times.
Published an opinion piece by Andreas Hoefert, Chief Economist and Chief Investment Officer for Europe at UBS Wealth Management, "Reinhardt and Rogoff Are Wrong".
发布《哈佛学者批克鲁格曼行为“无礼”》（翻译自Benjamin Pimentel的“Economists accuse Krugman of ‘uncivil behavior’”）。
Publication of "Raghulam Raja: The Paranoid Style of the Public Economist" (translated from "The Paranoid Style in Economics" by Raghuram Rajan, former Governor of the Reserve Bank of India).
11. The views of domestic scholars
Zhang Ming, deputy director of the Institute of Finance and Economics of the Chinese Academy of Social Sciences and deputy director of the National Finance and Development Laboratory, published an article entitled "Deepening the Understanding of Leverage Ratio: A Review of the Reinhart &Rogoff Controversy" on his official account "Zhang Ming Macro Finance Research", the full text of which is as follows:
By far the hottest event in the international economics community of 2013 was the controversy surrounding a paper published in the American Economic Review a few years ago by renowned economists Reinhart and Rogoff, both of whom are professors at Harvard University. The central idea of the paper is that international experience shows that once a country's government debt-to-GDP ratio exceeds 90%, the country's median economic growth rate falls by 1 percentage point. It is said that this article has laid a theoretical basis for the IMF, Germany, France, and other countries to demand that southern European countries implement fiscal austerity policies after the outbreak of the European debt crisis. By the beginning of this year, several American economists found that Reinhart and Rogoff's paper had operational errors in the data calculation process, subjective bias in the sample selection process, and questionable weighting methods. In addition, it is difficult to determine whether high debt leads to low economic growth, or whether low economic growth leads to high debt, and it is difficult to determine the correlation and causal relationship.
For a time, criticism of Reinhart and Rogoff was rampant. More notably, behind these criticisms is a reinforcing view that since the level of a country's government debt stock does not have a significant impact on the country's economic growth rate, a country can implement fiscal deficit policies without restraint in order to boost its own economic growth. Even if there is a huge amount of government debt in the future, the country can still roll over or even absorb the debt through quantitative easing. In a word, government debt is no longer a problem.
The author does not agree with the latter view. While hindsight suggests that a 90% threshold for government debt is problematic for countries at different stages of development, with different levels of financial markets and economies of varying size, it does not mean that Reinhart and Rogoff's warnings of high government debt can be ignored. In fact, after reading the challenger's paper and several rounds of responses from Reinhart and Rogoff, I find that both sides acknowledge that high government debt affects economic growth, with the difference being only to the extent of it. Although the world is changing rapidly, a lot of common sense has maintained its tenacious vitality. For example, the phrase "paying off debts" does not change depending on the level of development of financial markets or the complexity of financial instruments. It is true that a government can reduce its government debt by creating inflation, but this can cause serious damage to the reputation of that government. In short, if you come out to mix, you will have to pay it back sooner or later.
The outbreak of the current global financial crisis provides us with a valuable opportunity to understand the concept of leverage more comprehensively. The leverage ratio includes both the total leverage ratio of the economy as a whole (national leverage ratio) and the leverage ratio of sub-sectors (households, firms, governments), and the "ratio" refers to the ratio of the total debt or sub-sector debt to the country's GDP. The author believes that after the current round of financial crisis, our understanding of leverage ratio has deepened in at least the following three points:
First, in the past, we focused on government leverage while neglecting the private sector. After the current round of financial crisis, we look back and realize that both are actually important. In the past, we have ignored private sector leverage on the implicit assumption that the private sector is more rational than the government and that its decisions about increasing or decreasing debt are reasonable. In the aftermath of the crisis, however, we found that the accumulation of private sector leverage was often accompanied by the expansion of asset price bubbles. Once the asset price bubble bursts, the deleveraging of the private sector will lead to the outbreak of a financial crisis, and the government's bailout will mean the transfer of private sector debt to government debt, that is, the government will have to respond to the deleveraging of the private sector by increasing leverage.
The United States is a prime example. Before the subprime mortgage crisis, the U.S. government had very low leverage, but its national leverage ratio exceeded 300%. After the subprime mortgage crisis, private sector leverage fell rapidly, but at the same time government leverage continued to rise. Another example is China. Although the leverage ratio of the Chinese government is still within a manageable range (even with contingent debt, the debt-to-GDP ratio is currently around 60-70%), the leverage ratio of Chinese companies is at a high level globally (over 100% of GDP). This means that once China's corporate sector begins to deleverage in the future, China's government leverage ratio may quickly climb into dangerous territory, and the possibility of a local debt crisis cannot be ignored.
Second, in the past, we have focused on the level of external debt at the expense of the level of domestic debt. In the wake of the current financial crisis, we have turned to the fact that both are important. It is true that foreign investors investing in domestic bonds are more volatile and more susceptible to exchange rate risk, global liquidity and global risk appetite than domestic investors, so excessive accumulation of external debt usually leads to debt crises and currency crises. However, the excessive accumulation of domestic debt is also very dangerous. On the one hand, the accumulation of domestic debt is usually accompanied by the inflation of asset price bubbles, which in turn lead to the transfer of domestic debt forms (from the private sector to the government); In many cases, even if the size of a foreign debt is limited, an excessively high level of domestic debt alone can lead to the outbreak of a debt crisis.
Summing up the above two points, it is not difficult for us to find that the first point is that it is important which department owes money, but more important is the level of the overall debt level; the second point is that it is important who owes money, but what is more important is the level of the overall debt level. In other words, in the past, we may have overlooked the importance of total leverage, but in the future, the significance of total leverage monitoring will become more and more prominent.
Third, the process of deleveraging is very painful, but whether the process of deleveraging can be successfully achieved is crucial to the recovery of the macro economy and financial markets. In the deleveraging phase, financial institutions are reluctant to lend, and households, enterprises and governments are less willing to borrow, which will undoubtedly lead to a contraction in investment and financing, a decline in economic growth, a rise in unemployment, and an increase in social unrest, and the whole process will be very painful. However, if debt is not reduced to sustainable levels through deleveraging, the macroeconomic and financial market recovery will be difficult. In other words, when it comes to deleveraging, there are two options: "Nirvana rebirth" and "it's better to die than to live". The fact that the US aggregate leverage ratio has been significantly lowered in recent years, and the deleveraging of the eurozone and Japan has not yet begun, may explain why the current economic growth prospects in the US are much better than those in the eurozone and Japan.
In this sense, it is not terrible if too much debt leads to a debt crisis debt. What is scary is the wrong way to deal with a crisis. Reducing real debt levels, whether through debt restructuring or inflation, is the way to address the root cause (of course, the latter can seriously damage the government's reputation). Responding to a crisis through debt rollover or debt transfer is likely to delay the problem rather than solve it if it does not reduce real debt levels (the absorption of debt through growth is in many cases wishful thinking). The above enlightenment may have important reference significance for both the euro district government and the local government in China.
Comment that even if RR's data processing and statistical methods are correct, its "underlying logic" is wrong.
"The indirect effect of debt on economic growth is much larger and more long-term than the direct effect. The direct effect, or the multiplier effect, is rooted in the Keynesian tradition of macro-fiscal analysis. Indirect effects are the "balance sheet effect" that surpasses the multiplier effect, one is the asset effect and the other is the capital effect. Together, they are referred to as "extra" investment effects, which are different from consumption effects, which end at the multiplier effect. ”
Wang's policy conclusion is that "a 90% debt ratio threshold or warning value is meaningless. All policymakers really need to care about is a concise balance sheet logic. The logic of the balance sheet is the logic of "debt-bound assets" and "asset effect amplification into capital effect". This means, firstly, following the golden rule of debt policy, that is, "debt can only be used for investment", and secondly, it also means that the "high-quality assets" rule should be high-quality assets: not only have the potential to maintain and increase their value, or at least not lower than their replacement value, but also be sufficient to generate sufficient "flow income", either in the form of financial cash flow and profit flows, or service flows as indicators and conditions of economic growth. These service flows contribute to economic growth and are linked to a "service-oriented government". ”
Professor Wang not only criticized the logic of the RR paper, but also proposed a methodological reflection, arguing that the destructive impact and consequences of "variable thinking" on social science research should be fundamentally reflected. It fundamentally removes the precious "speculative thinking" and "associative thinking", and also cleanses the mind and soul from the roots. It wraps the highly self-manipulating "pseudo-scientific" research into the cloak of science (mathematics), discards the logic of cause and effect, and pushes the correlation of worship to the point of no return, resulting in many (but not all) discoveries and conclusions that are more stupid and superficial than truths and insights. ”
12. Citations of the paper in Chinese journals
The New Economist Think Tank is based on the data published by CNKI
Since Carmen Reinhart and Kenneth Rogoff's 2010 book Growth in a Time of Since 2020, with the increasing seriousness of China's local government debt problem, Chinese scholars' attention to debt-related issues has also increased significantly, with the number of citations of the paper increasing by 42% year-on-year to 139. From 2020 to the present, the number of citations of the paper has remained high, with more than 100 times.
The New Economist Think Tank is based on the data published by CNKI
There are many scholars who cite "Growth in a Time of Debt" in their papers, and their institutions are also diverse, among which the 10 institutions that cite the paper the most times are shown in the figure, and the Central University of Finance and Economics has 35 papers, ranking first. Among the top 10 institutions, there are also 20 papers with the least number of citations.
The New Economist Think Tank is based on the data published by CNKI
There are also a large number of papers cited in "Growth in a Time of Debt" and published in well-known journals, and almost all the famous journals in the field of economics in China, among which "Journal of Financial Research" and "Journal of Fiscal Research" have published the most papers, with 12 and 11 papers respectively, mainly because the paper focuses on the relationship between government debt and economic growth, which is directly related to finance and finance.
Bibliography:
［1］Economists' Spreadsheet Error Upends the Debt Debate，bloomberg
［2］Reinhart, Rogoff... and Herndon: The student who caught out the profs，BBC News
［3］Reinhart and Rogoff correct austerity research error, BBC News
［4］Doubt cast on research supporting austerity, BBC News
［5］REINHART AND ROGOFF: 'Full Stop,' We Made A Microsoft Excel Blunder In Our Debt Study, And It Makes A Difference, BusinessInsider
［6］How a student took on eminent economists on debt issue - and won, Reuters
［7］Shocking Paper Claims That Microsoft Excel Coding Error Is Behind The Reinhart-Rogoff Study On Debt, BusinessInsider
［8］REINHART AND ROGOFF FIRE BACK: Critique Of Our Work Still Confirms What We've Found On High US Debt, BusinessInsider
［9］Does high public debt consistently stifle economic growth? A critique of Reinhart and Rogoff, Cambridge Journal of Economics
［10］Reinhart and Rogoff Release Errata,Committee for a Responsible Federal Budget
［11］Reinhart And Rogoff Are Not Happy,Paul R. Krugman,The New York Times
［12］Carmen Reinhart And Ken Rogoff Write Huge, Brutal Letter To Paul Krugman Blasting Him For Being Uncivil，BusinessInsider
［13］Holy Coding Error, Batman,Paul R. Krugman,The New York Times
［14］Further Further Thoughts On Death By Excel,Paul R. Krugman,The New York Times
［15］The Good Glitch,Paul R. Krugman,The New York Times
［16］The Medium Term Is Not The Message,Paul R. Krugman,The New York Times
［17］Reinhart and Rogoff's Full Response To The New Paper Criticizing Their Austerity Research，The Huffington Post
［18］Reinhart and Rogoff's Second Response To Critique Of Their Research，，The Huffington Post
［19］Meet the 28-Year-Old Grad Student Who Just Shook the Global Austerity Movement，New York Magazine
［20］JOURNAL EDITOR: The Famous Reinhart-Rogoff Debt Paper Did Not Go Through The Normal Refereeing Process，BusinessInsider
［21］Why Reinhart and Rogoff Results are Crap, Randall Wray
［22］Debt And Growth, Yet Again, Krugman,The New York Times
［23］Debt and delusion, Robert Shiller
［24］The buck does not stop with Reinhart and Rogoff, Lawrence Summers, Financial Times
［25］Why do people hate deficits? Dylan Matthews，The Washington Post
［26］Debt and Delusion，Robert Shiller，Project Syndicate
［27］The Paranoid Style in Economics，Raghuram Rajan，Project Syndicate
［28］Risks of Debt: The Real Flaw in Reinhart-Rogoff?？Bradford DeLong，Project Syndicate
[29]《深化对杠杆率的认识 ——兼评Reinhart&2Rogoff争议》,张明,张明宏观金融研究
[30] "Professor Wang Yongjun's Commentary on Reinhardt and Rogoff's Debt-Growth Papers", New Economist
[31] Simple Statistics, Gary Smith, Jiangsu People's Publishing House