Bachelor Thesis in Economics, 15 credits
Economics C100:2
Spring term 2020
Crowding out or crowding in?
A study on the effects of the public debt ratio on
private investments of countries in the euro area.
Alexander Nilsson
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Guidelines and tips
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Community
Ask the community
Ask the community for help and clear up your study doubts
University Rankings
Discover the best universities in your country according to Docsity users
Free resources
Our save-the-student-ebooks!
Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors
The Impact of Public Debt Ratio on Private Investments: An Examination of the Euro Area, Study notes of Accounting
The relationship between public debt ratio (PDR) and private investments (PI) in euro area countries. Using various regression models, the study examines how changes in the PDR affect the PI. The findings suggest that higher PDR levels reduce investments for industries that require more external financial resources and increase the sensitivity of investment to internally generated funds for credit-constrained firms.
What you will learn
- What is the role of internally generated funds in private investments when public debt ratio increases?
- How does a one percent change in public debt ratio affect private investments?
- Which industries are most affected by changes in public debt ratio?
- What is the relationship between public debt ratio and private investments in euro area countries?
Typology: Study notes
2021/2022
1 / 54
This page cannot be seen from the preview
Don't miss anything!
Related documents
Partial preview of the text
Download The Impact of Public Debt Ratio on Private Investments: An Examination of the Euro Area and more Study notes Accounting in PDF only on Docsity!
Bachelor Thesis in Economics, 15 credits
Economics C100:
Spring term 2020
Crowding out or crowding in?
A study on the effects of the public debt ratio on
private investments of countries in the euro area.
Alexander Nilsson
This page is left intentionally blank
This page is left intentionally blank
Abstract
The evolution of public debt in the euro area into unsustainable levels is discussed now more than ever
after the recent financial crisis of 2008 and the European sovereign debt crisis that followed. With
countries like Greece and Italy hovering around the level of 181.2- and 134.8 percent of GDP in public
debt in 2018 (AMECO database), it is hard not to be worried about the future. Macroeconomic theory
predicts that increased bond-financed government expenditure will crowd out private investment. This
paper tests the credibility of this theory by empirically examining the effect of public debt ratios on
private investment in 26 countries in the euro area from 1999 to 2018. By using panel data regression,
accounting for fixed effects, it emerged that public debt ratios are negatively correlated with private
investment as is predicted by the crowding out theory. However, no statistically significant negative
correlation was found when introducing an instrumental variable, military expenditure.
Keywords: Private investments, Public debt ratio, Crowding out, euro area
Contents
- Introduction
- Literature review
- Theoretical framework
- 3.1 Reduced national savings
- 3.2 Aggregate production function
- 3.3 Government budget identity
- 3.4 The fiscal policy transmission mechanism
- Data and Method
- 4.1 Data
- 4.2 Model specifications................................................................................................................................
- 4.2.1 Basic OLS model
- 4.2.2 OLS model including control variables
- 4.2.3 Panel data model...............................................................................................................................
- 4.2.4 Log-log model
- 4.2.5 Quadratic model
- 3.2.6 Instrument variable model
- 4.2.7 Variable descriptions
- 4.3 Endogeneity issues
- 4.3.1 Omitted variable bias
- 4.3.2 Misspecification of functional form
- 4.3.3 Measurement error and error-in-variables bias.................................................................................
- 4.3.4 Missing data and Sample selection
- 4.3.5 Simultaneous causality
- 4.3.6 Heteroskedasticity
- 4.3.8 Instrumental variables.......................................................................................................................
- Empirical results
- 5.1 Omitted variable bias
- 5.2 Panel data regression
- 5.3 Log-log transformation
- 5.4 Quadratic transformation
- 5.5 IV regression
- Discussion
- Further research
- Conclusion
- References
- 8.1 Written sources
- 8.2 Data sources.............................................................................................................................................
- Appendix
1. Introduction
The purpose of this paper is to examine whether the increased public debt ratio will crowd out private
investments (Shortening as PDR and PI hereafter) in euro area countries. In other words, that increased
public investments as the government issue more debt will reduce private investments. The aim is to
provide concrete results of the effect PDR will have on the private sector and to serve as a basis for
future policies, especially for countries that currently hold high public debt. The paper will not examine
if crowding in- or out private investments harm the economy. This question is left open for future
research.
High PDR is nothing new in the euro area, however, after the financial crisis of 2008, many euro
countries experienced dramatic increases in their public debt in what is often referred to as “The
European debt crisis”. The average debt of all European Union countries increased by 20% between
the year 2007 and 2010. Greece was already highly indebted before the crisis with 107.4% of GDP in
2007 which increased to 144.9% three years later in 2010. Even countries with very low PDR before
the crisis like Ireland (24.8% in 2007) were affected. In Ireland's case, their PDR nearly quadrupled in
size to 92.5% in 2010 (Beker 2014 , 3 - 4 ).
The European Union have mechanisms in place to limit member states accumulation of excessive debt.
For example, a debt rule from 1992 that limit the general government debt of member states to a
maximum of 60 percent as a share of their GDP. However, the average PDR of the 27 member states
at the end of the year 2018 was 77.8%, which is substantially higher than the debt rule limit (Eurostat).
A total of 11 member states reported a higher PDR than 60% at the end of 2019. These countries can
be punished with fines, sanction or closer surveillance in the future if they do not make progress in
decreasing their debt ratio down to below the 60% limit (Lledó, Victor, Sungwook Yoon, Xiangming
Fang, Smaba Mbaye, and Young Kim. 2017, 81).
One very serious consequence of a high PDR is that the country is not able to pay back the debt and
have to default which in turn can cause financial panic both domestically and internationally. An
increased PDR can thereby be seen as an increased risk. Investors thereby demand a risk premium of
higher interest to compensate for the risk of default (Juessen, F., Linnemann, L. and Schabert, A.
2010). This higher interest rate can in turn crowd out domestic investments as the cost of borrowing
becomes higher.
The financial crowding out effect is ambiguous as it does not necessarily have to crowd out private
investments, but can instead crowd in some. Friedman argues that this is because of the portfolio
crowding out effect and the fact that can be either positive or negative.
A more modern view than Friedmans’ is that of Elmendorf and Mankiw ( 1999 ) that argues that the
conventional view of the effect of increasing government debt is that the effect is different between
the short- and long run. In the short run, when governments increase their debt by running a deficit,
this raises the households disposable income and perhaps lifetime wealth. This will, in turn, increase
the aggregate demand as household spending on goods and services increases. This is possible in the
short run because of sticky wages and prices, but the aggregate demand effect will have less importance
in the long run as wages- and prices adjust. And that in the long run, government debt will lead to
lower investments which in turn will reduce the domestic capital stock and lower output and income
as a consequence.
Hubbard (2012) instead highlights that the effect can depend on who is holding government debt.
Central banks in growing economies regularly purchase government debt to expand the money supply
and in an attempt to keep prices constant (McCallum 1984). Hubbards argument is that government
debt held by the central bank does not crowd out private investment since this a naturally occurring
process for central banks in growing economies to hold prices constant. And that many empirical
studies of government debt fail to include that the central bank may be the one holding the government
debt which may be the source of ambiguous results.
One important alternative theory that is often discussed within the field of government debt is the
Ricardian equivalence theorem. Which states that consumers are rational and forward-looking,
meaning that they will expect higher future taxes if the government accumulate more debt. If for
example, the government finance a tax cut by increasing government debt, then rational forward-
looking consumers will save the increased income they get in the current period to be able to pay
expected higher taxes in the future needed to pay back the debt. Thereby, the difference between
financing government expenditure through increasing debt or increasing taxes have the same effect
and are equal according to the Ricardian equivalence theorem. (Birch Sörensen & Whitta-Jacobsen
2010, 439 - 444). Some economists argue that the capital stock remains constant and interest rates will
not increase when the government debt is increased given that the Ricardian equivalence holds.
Thereby no change in the private investment will occur due to a change in government debt.
This because an increase in government debt is offset by increased private savings since the consumers
are forward-looking and knows that they will have to pay off government debt accumulated today by
higher taxes in the future. This means that the capital stock remains constant since households have
saved funds to smooth out their consumption over time. And that interest rates remain unaffected
(Hubbard, 2012).
Barro (198 9 ) argues instead that the world in non-Ricardian and that, therefore, this assumption does
not hold. Among his criticisms is that people do not live forever and thereby can transfer the cost of
higher future taxes onto future generations. Furthermore, the discount rate is higher for a typical person
compared to the government because of imperfect private capital markets and that future taxes and
income are not known. Lastly, the Ricardian equivalence assumes full employment, which is a bold
assumption that does not hold in reality.
Just as Friedman (1978) argues that government debt can both crowd- in and out private investments,
many economists have different views of the magnitude and the direction that the effect of government
debt has on different macroeconomic factors. For example the effect on interest rates, which in turn
can reduce investments as the cost of borrowing increases. While some economists including Friedman
argues that the effect is large and positive, that increasing government debt will lead to high-interest
rates, others have found empirical evidence that there is no effect on interest rates (Hubbard 2012 ,
This paper is related to the previous work of Huang, Panizza & Varghese (2018) who studied the
effects of government debt on total investments and corporate investments. They examined 172
countries between the period of 1980 - 2016 and found a strongly statistically significant negative
relationship between the government debt over GDP and the investment-to-GDP of a country. They
then further researched how private firms and different industries were affected by government debt
over GDP. The findings show that higher PDR levels reduce investments for industries that need more
external financial resources and that it increases the sensitivity of investment to internally generated
funds for firms that are already credit constrained. The latter part will not be examined in this paper.
Instead, the focus will be on solely the effect on the PI and in a smaller sample of 26 euro area
countries.
By assuming that the net interest receipt part of the current account is small enough to be neglected,
then this gives the following identity:
Substituting the equation (3. 4 ) into (3. 3 ) yields the following identity:
Elmendorf and Mankiw (1999) argue that if the Ricardian equivalence theorem does not hold and that
the private savings will not increase by the same amount as public savings (T - G) decreases. Then, if
the government increases its public debt by either decreasing taxes or increasing government spending
while holding the other constant, either domestic investment or net foreign investment have to fall to
compensate.
3.2 Aggregate production function
Hubbard (2012) argues describes how public debt can crowd out productive capital and raise interests
by using a Cobb-Douglas production function:
𝛼
( 1 −𝛼
)
Where (Y) is output, (A) is the coefficient for multifactor productivity, (K) is the capital stock, (𝛼) is
is the coefficient on capital in the production function and (L) is labour units. Hubbard (2012) argues
that the interest rate (r) is determined by the marginal product of capital (MPK). And that the total
return of capital (MPK*K) in the economy as a share of output (Y) is equal to 𝛼:
Therefore, the interest is determined by:
( 1 −𝛼
)
If government debt is assumed to fully crowd out capital then:
Given that these assumptions hold, then an exogenous change in government debt causes the interest
rate to increase given that all other factors are held constant:
2
Because 0 < 𝛼 < 1 and 𝑌, 𝐾 > 0. Output and capital cannot be negative and the coefficient on capital
is assumed to be bigger than zero but less than one.
This theory predicts that the interest will increase as government debt increases. An increase in the
interest rates will increase the cost of borrowing and thereby have a negative effect on investments.
3.3 Government budget identity
How can governments finance an increase in government spending? The answer can be drawn by
looking at the simple Government budget identity by Carlin and Soskice (2006, 176):
Note:
G= government expenditure on goods and services in nominal terms
iB= interest on the outstanding stock of bonds
T= tax revenue measured net of transfers
∆𝐵 =new bonds issued in the current period
∆𝐻 =new high powered money printed by the government
From this simple identity, it is easy to distinguish that the government have three options to choose
from in order to finance their spending both on current projects (G) and on previously issued debt (iB).
The first option would be to increase tax revenues (T), however, it’s not always easy for the
government to raise tax revenues on short notice to finance unexpected changes in government
spending. Instead, the government can issue bonds (∆𝐵) to acquire funds by increasing its outstanding
government debt.
4. Data and Method
This section starts with a presentation and discussion of the data used, followed by the model
specifications. And lastly, a presentation of the method used in the study.
4 .1 Data
The data used in this paper for gross fixed capital formations, general government gross debt, short-
and long interest rates and the output gap was collected from the AMECO database, which is the
annual macro-economic database of the European Commission's Directorate-General for Economic
and Financial Affairs. The data for military expenditure as a share of GDP was collected from Eurostat.
The dataset contains 26 countries with observations for all variables across the years 1999-2018. The
availability of cross-sectional and time-series data allows for the use of panel data regressions.
Key variables in the estimated model are PI and PDR. Another variable of interest is military
expenditure as it is used to estimate the effect of the PDR that is not correlated with the error term
through an IV-regression. PI is represented by gross fixed capital formation at current prices as a share
of GDP and PDR is represented by general government gross debt as a share of GDP.
Gross fixed capital formation “consists of resident producers’ investments, deducting disposals, in
fixed assets during a given period.” (Eurostat. 2020 ) and is, therefore, a good proxy for domestic PI.
Fixed assets have to be produced and thereby it does not include for example land or natural resources.
Furthermore, fixed assets must be intended for the production of goods and services for more than a
year (OECD. 2020 ). Fixed assets can be tangible, such as machinery or factories, or intangible for
example as patents or intellectual property rights.
The general government gross debt can be described as the nominal value of outstanding government
gross debt at the end of the year consolidated between and within sectors of general government
(Eurostat. 2020). The OECD (2020) describes that the general government debt consists of the central-
, state, local governments and social security funds controlled by these units together. The liabilities
that sum up the government debt consists of currency and deposits, debt securities and loans. The stock
of general government debt is then divided by each respective country’s GDP to express public debt
as a share of GDP.
Military expenditure is proxied by the general government total expenses on defence as a share of total
GDP. The output gap is defined as the gap between the current- and potential GDP at constant prices.
For long- and short interest rates, central government 10-year bonds for long-, and 3-month intrabank
for short interest rates are used.
Table 4 .1: Descriptive statistics
Variable Observations Mean Standard
deviation
Min Max
Private investments
Public debt ratio (%)
Long interest rates (%) 498 4.144 2.424 0.09 22.
Short interest rates
Output gap (%)
Military expenditure
Table 4 .2: Outliers in PI, PDR and military expenditure
Variable Greece Estonia Ireland Luxemburg
Private
investments
Public debt
ratio (%)
Long interest
rates (%)
Short interest
rates (%)
Output gap
Military
expenditure
Note: The top row in each cell is the maximum value for the given variable of that country. The bottom row is the minimum
value for the given variable of that country.
𝑐
is the error term, and that is the variation in the data that cannot be explained by the model. In other
words, it is the deviation of the true PI of observed countries in each period from the predicted PI from
the model.
4 .2.2 OLS model including control variables
The next step is to include control variables in the regular OLS model to mitigate omitted variable
bias:
𝑐
0
1
𝑐
2
𝑐
𝑠
3
𝑐
𝑐
In this multiple regression model, the interpretation of the population coefficients, 𝛽 1
2
3
, is still
the change in the PI of a one-unit change in the observed variable (PDR, i
s
or Gap) given that all other
included variables are held constant. That means that the coefficient of the variable of interest, PDR
1
), is the change in the PI of a one-unit change in the PDR given that both are and Gap is included
and held constant. Thereby, these included variables are “controlled for” which mitigates omitted
variable bias. This is explained in Section 4.3.1. Omitted variable bias down below.
4 .2.3 Panel data model
The dataset used contains longitudinal data and therefore it is possible to control for time- and fixed
effects of unobserved variables. Therefore, a panel data model is constructed as follows:
𝑐𝑡
0
1
𝑐𝑡
2
𝑐𝑡
𝑠
3
𝑐𝑡
𝑐
𝑡
𝑐𝑡
Where: 𝜆 𝑡
4
2000
21
2018
and 𝐵 4
2000
= 1 if the year observed is the year 2000 and
2000
= 0 otherwise. The year 1999 is used as a reference year, and therefore, is omitted to prevent
perfect multicollinearity. And: 𝛼 𝑐
4
𝐵𝑒𝑙𝑔𝑖𝑢𝑚
25
𝑈𝑛𝑖𝑡𝑒𝑑 𝑘𝑖𝑛𝑔𝑑𝑜𝑚
4
𝐵𝑒𝑙𝑔𝑖𝑢𝑚
= 1 If
the country observed is that of Belgium and 𝛾 4
𝐵𝑒𝑙𝑔𝑖𝑢𝑚
= 0 if any other country is observed. Austria
is used as a reference country, and therefore, is omitted to prevent multicollinearity.
The variable 𝛼 estimates the entity fixed effects that vary across entities but not overtime. In this case,
it is the country-specific fixed effects such as norms that may differ across countries but stay constant
within countries over time.
This variable is included to control for omitted variables that vary across countries but not overtime.
By including an entity fixed effect variable, the model allows for country-specific intercepts during
regressions given by 𝛼 that represent the effect of being in country 𝑐 (Stock & Watson 2015, 403-405).
The variable 𝜆 estimates the time fixed effects that vary across time but remains constant for all entities.
Like the previously mentioned entity fixed effects, specific intercepts are thereby allowed in the model
but instead for each time-period where 𝜆 represent the effect of being in time-period 𝑡. Furthermore,
this will control for omitted variables that cannot be observed that vary across time but not across
entities. (Stock & Watson 2015, 407-409) This could in, for example, be bilateral financial regulations
in the EU. New accounting- or credit rating rules can have a big influence on private investments.
They may not change over shorter periods but remains the same for all countries as the policies apply
to all EU countries. The panel used in the models is balanced as there are observations available for
all variables across all countries and time-periods (Stock & Watson 2015, p.397).
Panel data is a mix of cross-sectional and time-series data. Persistent in time series data is
autocorrelation, that there is a correlation between variables over time. Autocorrelation emerges from
the fact that the value of a variable one time period can influence the variable in the next time-period
(Stock & Watson 2015, 411-414). One example of a variable in the model that is plausible to suffer
from autocorrelation is the variable “output gap”. Large output gaps indicate that the economy is
performing well above or below its potential and the economy soon will turn and move towards an
equilibrium output. It is plausible to assume that the output gap will be smaller in the time-period
thereafter. Therefore, the variable suffers from autocorrelation as the previous time-periods influence
the time-periods that follow.
4 .2.4 Log-log model
The relationship between the PDR and PI may not be linear and therefore a log-log model is
constructed as follows to test if the relationship is logarithmic:
𝑐𝑡
0
1
𝑐𝑡
2
𝑐𝑡
𝑠
3
𝑐𝑡
𝑐
𝑡
𝑐𝑡
The interpretation of the population coefficient of PDR, 𝛽
1
, in this model is the PDR elasticity of the
PI. That is the percentage change in the PI resulting from a one percent change in the PDR. This is not
changes in the absolute terms of the variables, but instead in their relative size.