The relationship between public debt and economic growth, a replication of the paper “Growth in Time of Debt”

(1)

1 The relationship between public debt and economic growth, a replication of the paper

“Growth in Time of Debt”

Thesis supervisor: Mr. Swapnil Singh

Shirin Ramezani 10445528 Bachelor Thesis

(2)

2

Statement of Originality

This document is written by Student Shirin Ramezani who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text

and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

In 2010 Reinhart and Rogoff published 'Growth in Time of Debt'. Based on an extensive database on debt and economic growth of 20 advanced countries over a 200 year period, and 20 emerging markets over a period of 40 years, they argue that public debt significantly stifles economic growth once it reaches 90% of the Gross Domestic Product. This conclusion was even maintained by the authors in 2013, after some major corrections on their first paper had to be published. Reinhart and Rogoff's paper had quite some influence, notably on

austerity policies, and generated a lot of subsequent research into the debt-growth

relationship, as the recent sharp increase in sovereign debt – normal during wartime and depressions – posed a new challenge to politicians. This paper replicates part of Reinhart and Rogoff's research, using their original data on 20 advanced countries over the post-WWII-period. It shows that there is a negative relationship between debt and growth, but no

significant threshold. However, if the results are corrected for the number of observations per country in each category of public debt/GDP ratio, there seems to be a clear threshold at 30%, a finding that coincides with the results of an earlier replication.

(4)

4

I. Introduction

I.1. Reinhart & Rogoff’s paper: “Growth in time of debt”

The sharp increase in sovereign debt of advanced economies, like Euro members, in the last two decades led to concerns about sustainability of fiscal policies. Now, the issue of interest is whether there exists a systematic relationship between public debt and growth of real per capita gross domestic product (GDP), to what extent public debt influences economic growth and whether it has a negative or a positive impact. To answer these questions, this paper attempts to replicate part of Reinhart and Rogoff’s (hereafter RR; 2010) paper of “Growth in Time of Debt”. RR’s article took the initiative to explore the relationship between public debt and the median and average growth rates. It brought the attention of many

researches to this forgotten issue and led to many articles. Moreover, their research was published in the aftermath of the 2007-2008 financial crisis, it supported austerity policies aimed at reducing high public debt levels. The main result provided by RR shows that there exists a negative non-linear relationship between the real growth rate of GDP and public debt levels. Their main argument shows that particularly there exists a considerable decrease in real growth of GDP if the debt level passes the 90% threshold or high level of debt.

According to RR, government public debt consists of domestic and external debt. Domestic public debt is issued in one’s national currency under domestic legal jurisdiction and is held mainly by local residents. It counts for two thirds of the total public debt (RR, 2008). Debts that carry a government guarantee are not included in public debt. RR stress the robustness of their finding by using a broad range of countries in different time periods. That is 20 advanced economies over a period of 200 years. Part of this dataset is available on their own website. Their research is based on a simple correlation and straightforward method that does not take other growth factors besides GDP ratio into account. They basically claim that the relationship between economic growth and low or medium levels of public debt is not significant.

Moreover, public debt has been highly significant in many countries but its importance was taken into account less. The reason for the lack of interest in studying the effect of public debt on growth could be attributed to data unavailability. Therefore, since the publication of the paper “Growth in Time of Debt”, RR had a major influence on macroeconomic policies worldwide and the paper was the subject of both scientific research and media attention. It has even been discussed in Dutch parliament (Tweede Kamer, 2014, p. 56).

(6)

6 In this respect, Mankiw (2012) mentions that the more a government finances its budget deficit by debt the higher the chances of default and the higher the interest rates, which leads to crowding out private investment and slowing down potential economic growth. This idea was first formulated by Modigliani in 1961. Also according to Laubach (2009) higher real interest rates lead to lower private investment and consequently lower growth. The two terms “deficit” and “debt” can be used interchangeably in economic debates. Government deficit occurs when the expenditure exceeds the revenue in a given year. While debt is the loanable funds that a government issues to finance yearly accumulated deficits.

In the past, large increases in public debt occurred only during wars or times of depressions. As the war debts are less problematic than the debts that are accumulated in peacetime, the debt accumulation in peacetime reflects unstable political economy and presents uncertainty about the future of one’s economy. Therefore, because policymakers could not come up with a reason for the recent peacetime debt explosions the interest in conducting empirical research in this area significantly increased.

In the subsequent section, background information on theory as well as empirical researches done on the debt-growth relationship will be discussed.

I.2. Literature review

First, in the theoretical review the three different aspects of the relationship between debt and growth will be presented. Then mainly empirical researches done in the aftermath of RR’s paper will be discussed. Many articles tried to validate the result of the threshold of 90% debt suggested in RR’s paper. Yet, there are also a series of papers which start questioning the validity of this arbitrarily chosen threshold. More work done to explore the debt-growth relationship will be provided in following sections.

One of the main critiques of RR was conducted by Herndon, Ash and Pollin published in 2013; “Does high public debt consistently stifle economic growth? a critique of Reinhart and Rogoff”. This article replicates RR (2010) and points out shortcomings of RR’s paper: selective exclusion of available data, a coding error and inappropriate weighting of summary statistics. Consequently, RR reacted to this article and published an erratum on May, 2013. These two papers will be used and discussed extensively further on in this replication.

(7)

7

I.2.1. Theoretical studies

In general, the relation between debt and growth is not clearly explored in the theoretical models because they show contradictory predictions. However, theoretical implications drawn from existing models refer to three different viewpoints on debt-growth relationship

Firstly, low or moderate debt levels have a positive impact on growth. In traditional neoclassical models, the existence of capital mobility or the possibility to lend and borrow increases the economic growth of a country (Pattillo et al., 2002). If the growth rate is higher than the interest rate then there is an over accumulation of capital. In this case an increase in public debt would benefit both the generations now and in the future (Saint-Paul, 1992). Also, according to the conventional view, public debt in the short run stimulates investment and has a positive effect on growth.

Second is the neutral relationship between debt and growth that is best explained by Ricardian Equivalence. The Ricardian Equivalence states that government spending policy does not affect the forward-looking consumer’s consumption. The government deficit can either be financed by issuing bonds or by raising taxes while the bonds eventually will be repaid by increasing taxes in the future. The forward-looking consumer knows that this means either taxing today or tomorrow. Therefore, public debt has effect on neither consumption nor private investment and growth. In other words, Ricardian Equivalence attributes no effect at all (Seater, 1993).

The third viewpoint is the adverse or negative relationship of debt and growth. According to the conventional view, public debt has a negative effect in the long run. While in the short run debt stimulates aggregate demand and output and stimulates investors, in the long run capital crowds out, inflation rises and investment decreases which leads to a decrease in output and growth (Elmerdorf & Mankiw, 1998). Increasing debt levels could discourage the investors from investing in the future. The possibility that the government may be unable to pay back its debt to its creditors alters the incentives to invest. The increase in interest rates causes crowding out of investment. Consequently, growth will be negatively affected (Ball and Mankiw, 1995, cited in Irons and Bivens, 2010). In an aggregate growth model of Modigliani, the existence of government debt increases taxes to finance the interest payments and

decreases the utility of individual living in the long-run equilibrium (Diamond, 1965). Ultimately, it seems that the negative effect of high public debt on GDP growth in the theoretical arguments is in line with empirical studies, such as RR, that show a non-linear correlation between public debt and economic growth.

(8)

8

I.2.2. Empirical studies

RR’s paper only considers a simple correlation statistics method to investigate the debt-growth relationship. It does not take other determinants of debt-growth or reverse causality into account. Reverse causality refers to the case where low growth can lead to high public debt. They also use homogenous parameters where homogeneity refers to the idea that the data generating process that describes the cross-country growth is common for all observations. The empirical studies done in the aftermath of RR consider mainly the following issues; heterogeneity, non-linearity, causality and non-arbitrary debt classification.

Kourtellos, Stengos and Tan (2013) investigate other threshold variables besides debt to GDP ratio therefore; they take heterogeneity parameter into account. According to Kourtellos et al. (2013) the effect of debt on economic growth can be influenced by other factors such as a country’s trade openness and quality of institutions - the latter one being the main factor of heterogeneity in the debt-growth relationship. They show that higher public debt results in lower real GDP growth for countries with a Low-Democratic regime which they measure by the quality level of a country’s institutions. So, the debt-growth relationship is not statistically significant in a High-Democratic regime. Finally, their results imply no statistically significant relationship between public debt and growth in advanced economies. The same conclusion is provided by Panizza and Presbitero (2014) that there is no evidence of impact of high public debt on growth in advanced economies in the medium run.

Kumar and Woo’s (2010) empirical research presents a negative relationship between high public debt and growth, controlling for other determinant factors of growth by using the system GMM estimator. They show that an increase of 10 percentage point in initial debt to GDP ratio is associated with an approximately 15 basis points slowdown in annual real per capita GDP growth. They also provide evidence of nonlinearity for the high levels of debt at the threshold of 90% of GDP.

Égert (2012) uses the same 20 advanced economies as in RR apart from the inclusion of Switzerland and the exclusion of Ireland. He finds a negative correlation between public debt and growth but generally with a much lower threshold than 90%. He also presents some evidence of nonlinearity but does not find any threshold.

(9)

9 I.3. Structure of this paper

The discussion proceeds as follows. Section II, provides the descriptive statistics of RR’s paper and describes in detail the effects of their results. Subsequently, in the methodology section the hypothesis and the model used will be discussed, this is followed by the data and sample statistics section. Furthermore, in section IV the results will be provided and in section V they will be discussed. Finally, the conclusion will be posed, including some limitations of this study and recommendations for further research.

I. Reinhart & Rogoff’s approach

II.1. Descriptive statistics and facts

RR present three datasets. First, 20 advanced economies all members of the OECD over the post-WWII period until 2009. The second category is the same 20 advanced economies over the period of 1970-2009 and the third, 20 emerging market economies from 1970-2009. The second classification for advanced countries is to make the comparison of the results between advanced and emerging markets easier. This is because the data for emerging countries is only available over the period of 1970 onwards. Moreover, there exists considerable variation across countries in the immediate years after the WWII. Some countries, such as Germany, had high debt levels while some, such as Australia and New Zealand, experienced no growth decrease at high debt levels.

They present four different public debt/GDP ratios: less than 30% (low debt level), between 30% and 60% (medium-low debt level), between 60% and 90% (medium-high debt level) and 90% and above (high debt level). This classification of debt groups is based on their

interpretation of policy discussion and literature. It is also in line with the World Bank country groupings of income classification. Then, they compare the mean and median of the real GDP growth rates across each public debt/GDP category. Table 1 represents the mean and median growth rate in the post-WWII period for the 20 advanced economies. The main results - as can be seen in the table - are that both mean and median stay relatively similar across low to medium-high debt levels, whereas the median growth rates for countries with a high domestic public debt is about 1% lower than the lowest growth in the other categories and the mean growth rate significantly decreases to negative 0.1%

(10)

10 Table 1: RR (2010) findings of the average and median GDP growth rate in 4 different public debt/GDP categories

RR(2010) from paper--for advanced economies

30% or less 30-60% 60-90% above 90%

Mean 4.1 2.8 2.8 -0.1

Median 4.2 3 2.9 1.6

*Postwar

II.2. Herndon’s correction

According to Herndon et al. (2013) not only data are not available for some countries so consequently not used, but also some available data are missing in RR’s calculation. France’s public debt/GDP ratios in the period 1973-77 and Spain’s level of growth rate in 1959-80 are some examples of this exclusion of available data.1 There also exists a spreadsheet coding error where RR calculated both median and mean of the countries excluding the first five countries in alphabetical order. Finally, RR first calculate the mean or median of the available number of observations for each country and then measure the overall mean and median of the dataset by counting each calculated number as one country observation. Herndon calls this inappropriate weighting because it’s based on summarizing by “country-year”. Table 2,

provides the results of the replication made by Herndon.

Table 2: Herndon et al. (2013) findings of the average and median GDP growth rate in 4 different public debt/GDP categories

Herndon et al. (2013) -- advanced economies

30% or less 30-60% 60-90% above 90%

Mean 4 3 3 1.9

Median 4.1 3.1 2.9 2.3

*Postwar

II.3. Reinhart & Rogoff’s corrections

In 2013 RR published an erratum “Growth in a time of debt” in which they describe the correction of the coding error and the exclusion of the available data in reaction to Herndon et al.’s paper. Table 3 summarizes the changes after the correction.

1

(11)

11 Table 3: RR (2013) findings of the average and median GDP growth rate in 4 different public debt/GDP categories after applying changes to descriptive statistics

Final including New Zealand (Maddison GDP)

30% or less 30-60% 60-90% above 90%

Mean 4.2 3.0 2.5 1.0

Median 4.2 3.0 2.9 1.9

*Postwar

The following graphical presentation of the results of RR paper before and after correction, figure 1 and 2, respectively, show the changes of economic growth in different debt

categories.

Figure 1: graphical presentation of the results of table 1 Figure 2: graphical presentation of the results of table 3

The corrected results- although different- still show a decrease in the mean and median growth rate in the high debt category. The median estimates though are very close to the previous calculations in debt categories below the 90% interval. RR therefore held up their aforementioned conclusions.

II.

Methodology

III.1. Dataset

As said before, RR present three datasets. The first group is 20 advanced economies over the post-WWII period. The second category are the same 20 advanced economies over the period of 1970-2009 and the third, 20 emerging market economies from 1970-2009. The focus of this

(12)

12 paper however is only on the first data set as for the emerging market economies the results of advanced economies are repeated and the decline for the average growth rate is even greater. Below, using RR’s original paper, Herndon’s criticism, RR’s 2013 corrections and the content of the database and before diving into the details, I explain which data from the original database were used for the replication in this paper and why. The main focus of this paper is only the dataset of 20 advanced countries over the period of 1946-2009 or post-WWII. There are three reasons to choose this dataset. Firstly, this is the only dataset spreadsheet available on RR’s website. Secondly, as it is the more recent dataset the results are expected to be more reliable. Lastly, according to RR the relationship between public debt and growth is quite similar between emerging markets and advanced economies.

In the erratum “Growth in a time of debt” (2013), RR already describe the correction of the coding error and the exclusion of the available data.

III.2. Excluded data

Despite RR’s corrections, there are still some available post war data that they excluded from their calculations. Therefore, this paper considers strictly the dataset available between 1946 and 2009 for 20 advanced countries. For the five countries- Finland, Spain, Sweden, UK and the US the data are fully available. Data are available for Portugal since 1950. RR takes Norway into account since 1947 and ignores one observation related to the public debt/GDP ratio of 60-90% with the real GDP growth of 10.2%. However, this paper accounts for all excluded numbers. There are no data available for the Netherlands till 1956 so the

calculations are done for the data available for the period 1956-2009. New Zealand

observations related to 1946-47 of the high debt category with the value of GDP growth rate of 5.3% and 3.6% are excluded from RR’s paper. No data are available till 1956 for Japan, however, the public debt/GDP ratio related to 1954 and 1955 is known. There is no data linked to the growth rate of GDP for these two years. For Italy, Germany and Canada and Australia RR do the calculations since 1951. In this paper, the data related to Germany and Italy are also calculated since 1951 because before that the data is not available. But according to RR five observations each for Canada and Austria are not taken into account. These are included in this paper’s calculations. The five observations related to Canada in the period of 1946-50 are in the high debt category with a growth rate of -1.0%, 4.4%, 1.8%, 2.2% and 7.4% respectively. The five exclusions related to Australia also fall in the high debt category of 90% or more public debt/GDP ratio. The growth rates related to this category are

(13)

-13 3.6%, 2.5%, 6.4%, 6.6% and 6.9% respectively since 1946. For Ireland data are available since 1949 and for Greece since 1970. In total, there is a difference of 14 observations missing in RR’s final version of their corrected calculations.2

They are all taken into account in this paper.

III.3. Weighting

The other problem mentioned in the replication of Herndon et al. (2013) is the inappropriate weighting in calculating the summary statistics. To solve this problem, two measurements are taken into account in this paper. First, counting each year of data from any of the twenty countries as one observation and then calculating the mean and the median of all observations all together. This approach is also taken into account by Herndon. This measurement by country-year approach is in contrast with RR’s way of calculating with only one observation for each country. In RR’s approach, for example for the public debt/GDP ratio of 60%-90%, Norway’s one observation- averaged to 10.20% growth rate in the final calculation- has the same weight as the 32 observations for Ireland which averages to 3.95. Moreover, all the county-year debt and GDP growth observations have been plotted in a scatter diagram to show a general view of the debt-growth relationship. Also, in other to get an idea of the composition of the entire collection of scatters, the individual scatter diagram of each country is plotted in graphs of the same scale.3

Knowing that according to RR’s method one observation for Norway will finally be counted as equal to the average of 32 observations for Ireland, the second measurement approach is introduced. In this way, the weight of each observation is calculated as the ratio of the number of observations of each country to the total sum of observations in a specific public debt/GDP category. By applying this neutralizing method, this paper attempts to correct the impact of RR’s approach that exaggerates the importance of single short-term observations.

III.4. Assumptions

As argued by many researchers, whether causality runs from debt to growth or whether low level growth rates result in higher debt remains a riddle which needs further investigation. In this article, however, it is assumed that different debt levels will lead to different growth rates.

2

Table 1 in the appendix gives an overview of the 14 observations missing

(14)

14 Throughout this paper, it is assumed that the relationship between debt and growth rate is homogenous across countries. In other words, the same debt thresholds are assumed to hold for all the advanced countries included in RR’s dataset.

III. Results

This paper applies three corrections on RR, as described above under 'Methodology'. First, table 4 summarises the results after the first correction for excluding 14 data points.

Table 4: This article’s findings of the average and median GDP growth rate in 4 different public debt/GDP categories after applying changes to descriptive statistics

All the data available between 1946-2009

30% or less 30-60% 60-90% above 90%

Mean 4.13 3.18 2.85 2.22

Median 3.92 2.98 2.93 2.55

Note: Results of this paper’s replication, postwar 1946-2009

By comparing the tables 3 and 4, one can see the exclusion of those 14 observations only has a significant effect on the average of high debt levels. Namely, the average is 1.22 percentage points higher and the median is 0.65 percentage point lower. This difference can be best explained by the exclusion of data points by RR among which the five data points of New Zealand and the five data points of Australia. These countries experienced no growth decrease at high level of debts (more than 90%). It is also noteworthy to mention that high growth high debt observations such as Germany are mainly seen in the years following the World War II. Second, table 5 summarizes the results after the second correction for inappropriate weighting using country-year measurement. That means the average of all observations or each single data point is taken into account. In contrast to the method of RR that first calculated these average of each country and finally calculated the average of averages. The same method is also applied for calculating the median.

Table 5: The average and median GDP growth rate in 4 different public debt/GDP categories based on summarizing all data exclusions by “country-year”

Summarizing all data exclusions by “country-year”

30% or less 30-60% 60-90% above 90%

Mean 4.31 3.14 2.97 2.30

(15)

15 Third, table 6 provides the results using the weighting method by neutralizing the number of observations in calculating the average and median. In other words, I calculated first a weight value for each country by dividing the relevant number of observation of that specific country by the total number of observation in each debt category. Then, I multiplied the weight by the average debt to GDP growth value of each country. Finally, the mean and median of the results has been calculated. The following table shows the results.

Table 6: The average and median GDP growth rate in 4 different public debt/GDP categories after correcting using the weighting method

Correcting the impacts of RR’s approach using the weighting method

30% or less 30-60% 60-90% above 90%

Mean 0.24 0.16 0.16 0.23

Median 0.29 0.15 0.10 0.21

*Postwar

As can be seen in table 6, there would be a decrease in the growth rate mean and median if the level of public debt/GDP ratio passes the threshold of 30%. Otherwise, the growth rate is almost stable until again it reaches the 90% threshold. The same conclusion by means of descriptive statistics - and using the same debt categories as in RR - is given by Égert (2015). According to Égert (2015) real GDP growth is considerably weaker when the domestic public debt exceeds the 30% threshold during the period of 1946-2009.

Finally, a scatter diagram was plotted based on the country-year method. That means by using all the available data for each country in the post war time on debt and growth a scatter

diagram is plotted. As can be seen the x-axis shows the debt levels and y-axis the real GDP growth. In this diagram each dot represents a country in a specific year. As one can see the variance in growth within each debt category is large therefore, the use of mean and median seems not the best option. To understand this even better, the column charts are represented for the mean and median of real GDP growth for each country. Each chart shows one debt category. It can be seen in all the following diagrams that even when the individual

observations are lumped together by country in the form of mean and median, the variance in growth in each debt category is still considerable, especially in the 90% and more debt

(16)

16 -20 -10 0 10 20 30 R e al G D P gr o wt h Debt/GDP

Real GDP growth and Debt/GDP ratio per 'country-year'

0.00 2.00 4.00 6.00 8.00 10.00 Au stra lia Aus tria Be lgi u m Can ad a De n m ar k Finla n d Fran ce G erm an y G re e ce Ire lan d Ita ly Jap an N ew Z eala n d N eth erl an d s N o rw ay Po rtu gal Sp ain Sw ed e n UK US 30% or less public debt

Mean Median 0.00 2.00 4.00 6.00 8.00 Aus tra lia Au stria Be lgi u m Can ad a De n m ar k Finla n d Fran ce G erm an y G re e ce Ire lan d It aly Jap an N ew Z eala n d N eth erl an d s N o rw ay Po rtu gal Sp ain Sw ed e n UK US 30%-60% public debt Mean Median

(17)

17 After making scatter plots for each country it showed that the scatter plot for all the data was made up of several smaller scatters that did not all look like less dense versions of the bigger one, in other words: the data for a single country did not make a representative sample of the whole population. This may mean that individual countries – that have their own monetary and fiscal policies – have their own influence on both debt and growth. Norway is a good example. Disregarding the outliers, most points ‘fall’ within a relatively small area between 10 and 50% of debt and -2 and 8% of growth. This suggests a consistent set of monetary and fiscal policies over the time period. The UK on the other hand has its debt vary between 30 and 240% with a similar variation in growth as Norway. Austria on the other hand, with debt varying between 0 and 60%, realised growth rates between -1 and no less than 20. This means that either the UK’s and Austria’s policies were quite different from those of Norway – and each other - or they changed during the period studied. Consequently, putting 20 different countries just by labelling them advanced and come to general conclusions by simple means of statistic while not taking into account any economic policies seems too simple approach.

-6.00 -4.00 -2.000.00 2.00 4.00 6.00 8.00 10.00 12.00 Au stra lia Aus tria Be lgi u m Can ad a De n m ar k Finla n d Fran ce G erm an y G re e ce Ire lan d Ita ly Jap an N ew Z eala n d N eth erl an d s N o rw ay Po rtu gal Sp ain Sw ed e n UK US 60%-90% public debt Mean Median -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 Au stra lia Aus tria Be lgi u m Can ad a De n m ar k Finla n d Fran ce G erm an y G re e ce Ire lan d It aly Jap an N ew Z eala n d N eth erl an d s N o rw ay Po rtu gal Sp ain Sw ed e n UK US 90% or more public debt

Mean Median

(18)

18

IV. Discussion

One problem mentioned in the replication of Herndon et al. (2013) is the inappropriate weighting in calculating the summary statistics. One solution is to use each year of data from any of the twenty countries as one observation and then calculating the mean and the median of all observations all together. This approach is also taken into account by Herndon, however, the results of this paper - given in table 5 - slightly deviate from Herndon’s calculations that could be the result of data availability.

Throughout this paper, it is assumed that the relationship between debt and growth rate is homogenous across countries. In other words, the same debt thresholds are assumed to hold for all the advance countries included in RR’s dataset. However, this assumption may be too restrictive given that domestic public debt can affect the growth rate differently in different countries. In other words, countries with different monetary and fiscal policies may have different effects on debt and growth.

Moreover, a special case of homogeneity problem is the effect of global shocks on different economies that may differ across countries. RR (2009b) suggest that countries that mainly rely on short-term borrowings are particularly more sensitive to “unexpected” financial crises. Also the macroeconomic conditions embedded in the interest rate and the primary budget balance of a country can determine the relationship between debt and economic growth. Also, the debt to GDP categories may upon further research turn out to have arbitrary

boundaries relative to the actual figures: e.g. if the threshold turns out to be at 24% or 85% of debt/GDP ratios. The presently chosen debt categories cannot show these hypothetical values or other possible outcomes as the classification of debt levels is arbitrary chosen and is not determined by the data.

Finally, in order to get a better understanding of this relationship there are many different points of views. The relationship between the debt/GDP ratio and economic growth can be found either by general trends like RR’s approach or by taking more specific preferences like Koutellos et al.’s approach (2013). These two different approaches of explanation were well explained in psychology for the first time by Charles Darwin in 1857 as ‘lumpers and splitters’. In this context, those who refine a research question by taking more and detailed parameters to answer the question are splitters. Whereas a lumper takes the general trend and assumes that only few main factors are needed to explain the research question and the differences are not as important as the general view. Therefore, RR’s paper is more of a

(19)

19 lumper’s work, which uses a quite simple and straightforward method. So, even though there are many different empirical modeling choices available, RR’s approach explains a lot. RR as the first to provide such broad dataset motivates many researches to question this issue.

V. Conclusion

There is a growing concern in the economies all around the world about the stability of debt which implies risks to long term growth and sustainability. The very recent example of

concern in Europe is Greece with a debt to GDP ratio of 163.6% (Worldbankdata, 2015). This all leads to the need for economists, policy makers and central banks to investigate the

relationship between public debt and economic growth.

This article basically tries to answer two questions, first whether there exists any relationship between the debt to GDP ratio and real GDP growth and second whether there is any

threshold in this relationship or it is linear. The results show that there exists a negative relationship between debt and economic growth. Considering the results driven from the country-year calculations, one can see that there is a decrease in average and median growth rate after the point where the debt to GDP ratio reaches 90% of the GDP. However, in contrast with the conclusion of RR this decline is not significant. According to the

neutralizing weighting method on the other hand, the real GDP growth rate is considerably weaker when the debt to GDP ratio reaches the 30% threshold.

This negative effect of debt levels on economic growth leads to different interpretations by investors about the economic situation. According to Koutellos et al. (2013) investors may consider high level of debt to GDP ratios the result of time inconsistent and inflationary policies that are meant to restore policy makers’ credibility while this may come with related costs such as unemployment and possible depreciation that hinders economic growth. Now the question is to what extent the magnitude of the debt threshold is generally valid across individual countries and across time? And to what extent the debt categories are valid? These can be some suggestions for future and further research.

(20)

20

Bibliography

Balázs Égert. (2015). The 90% public debt threshold: the rise and fall of a stylized fact. Applied Economics 47, 3756-3770.

Carmen M. Reinhart & Kenneth S. Rogoff, “Growth in a time of debt” data, retrieved on April 2015: http://reinhartandrogoff.com/

Checherita-Westphal, C., & Rother, P. (2012). The impact of high government debt on economic growth and its channels: An empirical investigation for the euro area. European Economic Review, 56(7), 1392–1405.

Diamond, P.A. (Dec., 1965). National debt in a neoclassical growth model. The American Economic Review, 55(5), 1126-1150.

Égert, B. (2012). Public debt, economic growth and nonlinear effects: myth or reality?. OECD Economics Department Working Papers 993, OECD Publishing.

Elmendorf, D.W., & Mankiw, N.G. (1998). Government debt. Handbook of Macroeconomics. Herndon, T., Ash, M., & Pollin, R. (2014). Does high public debt consistently stifle economic

growth? A critique of Reinhart and Rogoff. Cambridge Journal of Economics, 38(2), 257–279.

Irons, J., & Bivens, J. (2010). Government debt and economic growth. Economic Policy Institute, briefing paper #271.

Kourtellos, A., Stengos, T., & Tan, C. M. (2013). The effect of public debt on growth in multiple regimes. Journal of Macroeconomics, 38 (PA), PP. 35-43

Kumar, M. S., & Woo, J. (2010). Public debt and growth. IMF Working Paper, JEL classification number: O40, O47, H62, H63

Laubach, Thomas. (2009). New evidence on the interest rate effects of budget deficits and debt. J. Eur. Econ. Assoc. 7(4), 858–885.

Pattillo, C., Poirson, H., & Ricci, L. (2002). External debt and growth. IMF Working Paper Reinhart, B. C. M., & Rogoff, K. S. (2009a). The Aftermath of Financial Crises.pdf, 466–472. Reinhart, Carmen M., and Kenneth S. Rogoff. (2009b). This Time Is Different: Eight

(21)

21 Reinhart, C. M., & Rogoff, K. S. (2010). Growth in a time of debt. American Economic

Review, 100(2), 573–578. h

Saint-Paul, G. (1992). Fiscal policy in an endogenous growth model. The Quarterly Journal of Economics, 107(4), 1243-1259

Seater, J.J. (1993). Ricardian Equivalence. Journal of Economic Literature, 31(1), 142-190 Temple, Jonathan, 1999. The new growth evidence. J. Econ. Lit. 37(1), 112–156.

T. Herndon, M. Ash, R. Pollin. 2014. Does high public debt consistently stifle economic growth? A critique of Reinhart and Rogoff. Cambridge Journal of Economics 38, 257-279.

Tweede Kamer, vergaderjaar 2013–2014, 33 750, nr. 1 p. 56, Retrieved on June 2015 from:

http://www.tweedekamer.nl/kamerstukken/brieven_regering/detail?id=2013Z17169&

did=2013D35249

World bank data, retrieved on 17 July 2015:

http://databank.worldbank.org/data//reports.aspx?source=2&country=&series=GC.DO D.TOTL.GD.ZS&period=#

(22)

22

Appendix

Table 1: Accounting for the years in which country's level of debt and growth rates are excluded in the period 1946-2009

Country Year level of

debt

Total number of

years Extra information

Correct _total RR final count after correction Australia 1946-1951 >90% 5 0

Belgium 1947 >90% a growth rate is missing

Canada 1946-1950 >90% 5 0

France 1950 <30% a growth rate is missing

Italy 1946 >30%,<60% 7 6

Japan 1954-1955 <30% two growth rates are missing

New Zealand 1946-1947 >90% 5 3

Norway 1947 >60%,<90% 1 0

Figure 1: graphical presentation of the results of table 1

-1.0 0.0 1.0 2.0 3.0 4.0 5.0 R e al G D P gr o wt h r ate

Public debt/GDP ratio

RR(2010) from paper--for advanced

economies

Post war Mean Post war Median

(23)

23 Figure 2: graphical presentation of the results of table 3

0 1 2 3 4 5 R e al G D P gr o wt h r ate

Final including New Zealand (Maddison

GDP)

Post war Mean Post war Median

0.00 1.00 2.00 3.00 4.00 5.00 R e al G D P gr o wt h r ate

Calculations using all the data available

between 1946-2009

1946-2009 Mean 1946-2009 Median

(24)

24 Figure 4: graphical presentation of the results of table 5

Individual scatter diagrams of 20 advanced countries -- the x-axis is the public debt and the y-axis is the real GDP growth rate

0.00 1.00 2.00 3.00 4.00 5.00 R e al G D P gr o wt h r ate

summarizing all data exclusions by

“country-year”

1946-2009 Mean 1946-2009 Median 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 R e al G D P gr o wt h r ate

neutralising the long-run effects

Mean Median

(25)

(26)

(27)

The relationship between public debt and economic growth, a replication of the paper “Growth in Time of Debt”

Abstract

Contents

I.

Introduction

I.

Reinhart & Rogoff’s approach

Methodology

III. Results

IV. Discussion

V.

Conclusion

Bibliography

Appendix

RR(2010) from paper--for advanced

economies

Final including New Zealand (Maddison

GDP)

Calculations using all the data available

between 1946-2009

summarizing all data exclusions by

“country-year”

neutralising the long-run effects