Econometrics 1 mid term report factors influencing private consumption expenditures in germany

The paper uses the sample data of Germany’s household consumption from 1970 to 2020, since countries with higher or lower level of income usually give a clearer correlation between the d

FOREIGN TRADE UNIVERSITY

FACULTY OF INTERNATIONAL ECONOMICS

-* -

Econometrics 1 Mid-term report

Factors Influencing Private Consumption Expenditures

in Germany

Name of the student: Nguyen Ngoc Ha,

Tobias Schüler, Kim Thao Weil

Student ID: 2013450014,

2190101113, 2190101114

Supervisor: Vu Thuy Quynh

Hanoi, October, 2021

Individual Assessment Table

Evaluator Ha Tobias Kim

Average Score 10 10 10

Abstract

Household consumption has always been a factor that every country wants to precisely measure in order to have a clearer picture of the economy However, the factors affecting a nation’s household consumption are not always unanimous among researchers and economists, and those factors’ magnitude of effects on consumption varies due to different theories

This study aims at identifying the factors that may affect consumption, measuring the effect of those factors towards consumption and giving the researchers’ outlook on the econometric results The paper uses the sample data of Germany’s household consumption from 1970 to 2020, since countries with higher or lower level of income usually give a clearer correlation between the determining factors than countries with middle level of income To identify these factors and specify a regression model, a literature review was carried out and the OLS estimation technique applied to the model

The results imply a positive statistically significant impact of income and interest rate

on the private consumption expenditure in Germany A rise of GNI by one billion euros will be reflected by a rise of consumption of 0.273 billion euros and a rise of one CPI point results in a consumption increase of 9.462 On the other hand, there is

no statistically significant influence of interest rate on consumption within the model, however, taken individually, interest rates exert a negative impact on consumption expenditure

From these findings, the study suggests that policy makers should focus on income and inflation rate when they want to increase the nominal wealth in a nation However, it is important to note that inflation rate has an inverse relationship with real wealth For business owners, the paper shows that raising wages can indirectly affect consumption and lead to long-term revenue increases

I Table of Contents

II List of Abbreviations

CPI Consumer Price Index –

ECM – Error Correction Mechanism

FRED Federal Reserve Economic Data –

GNI – Gross National Income

OECD – Organization for Economic Cooperation and Development OLS Ordinary Least Squares –

PFCE Private final consumption expenditure –

R² - Coefficient of Determination

1 Introduction

Consumption has always been an important factor to determine a country’s economic development However, identifying the components that affect the consumption expenditures of a nation’s households is challenging There are several reasons to this

Firstly, the factors affecting consumption are still up for debate Researchers and scientists have been bringing up theories, models and functions to pinpoint the relation between different variables and consumption However, each person has his

or her own perspective on the main factors affecting a country’s consumption volume For example, John Maynard Keynes asserted that there are numerous factors affecting consumption expenditure, among which income is the one dominant (Keynes, 2018,

p 209 ff.) Afterwards, Franco Modigliani and Richard Brumberg improved Keynes ’theory and developed the Life-Cycle Hypothesis, which states that individuals plan their spending for their entire lifetime, taking into account future income (Hayes, 2021) Additionally, the Nobel-Prize-winning economist Milton Friedman established the Permanent Income Hypotheses, saying that people will spend consistently with their expected long-term average income, implying that changes in consumers’ behavior is unpredictable due to their various individual expectations (Kagan, 2020)

Secondly, even after determining the important factors that have an impact on the consumption within a country, the magnitude of each factor is still difficult to identify

This paper aims at providing insight into the relationship between three major factors and private consumption expenditure of a country These factors are national income (measured by GNI), inflation (measured by CPI) and interest rate (measured by 10-year government bond yields) The factors will be explained in more detail throughout the work To test the hypotheses, the OLS (Ordinary Least Square) regression will be applied, with the assumption that all the requirements are satisfied and there are no externalities The analysis is performed for the case of Germany with

51 observations, each observation represents a year from 1970 to 2020

In this paper, Germany has been selected to investigate the factors that influence the consumption expenditures due to the high-income nature of the nation, as high-income countries show a stronger connection between earnings and consumption compared to middle-income countries (Diacona & Maha, 2015, p 1535)

The paper will start with an outline of previous research and theories Based on this review, the relevant hypotheses will be developed Then, the applied methodology will be explained, including the model specification, data extraction as well as data and variable analysis To test the developed hypotheses, the OLS estimation technique will be used on the specified model Further, multiple hypotheses tests will

be used To sum up the research, a conclusion will be drawn, and an outlook given at the end

2 Literature review

2.1 Underlying economic theories

Income is defined as payments received plus the changes in the value of assets in a specified time period (Raupp & Raupp, 2018, p 156), while consumption is the expenditure on goods and services by households (Carroll, 2016) According to Keynes’ consumption function (Kenton, 2020), the consumption and income are arranged in a linear relationship, which suggests that aggregate consumption is entirely dependent on earnings, more specifically, disposable income, assuming that the autonomous income is constant Keynes assumes that the function is fairly stable over time, partly based on the Psychological Law of Consumption Moreover, this concept states that increased income raises the average consumption spending, however, not to the same extent (Ali & Rahman, 2015, p 15)

Furthermore, the Keynes´ consumption function can be extended by his findings on the relationship between spending, output and inflation, stated in the IS-LM curve One main part of Keynes´ theory is that whenever inflation is high, for example through an increase in money supply, the supply of money exceeds the money demand This would lead to a fall of interest rates which on the other hand would reduce the incentives for savings in an economy and thereby encourage consumption Thus, Keynes states in his theory that inflation has a positive impact in consumption (Krugman & Wellls, 2018, p 640 ff.)

In addition to that, Keynes stated that a negative relationship between interest rates and private consumption expenditures exists This indicates that with a higher interest rate, the household spending decreases due to the substitution effect Moreover, this implies that consumers prefer to save and defer immediate consumption over direct expenditure when more return is expected on savings due to higher interest rates (Keynes, 2018, p 97 ff.)

2.2 Existing empirical studies

The Keynes´ function, despite having been “reformed” many times by economists for the past 100 years, be it Life-cycle Hypotheses, Relative Income Functions or Permanent Income Functions, retains its core value: it implies the positive relationship between income and consumption in a certain period of time This is what the paper focuses on, and it has been supported by previous research numerous times According to a cross-sectional data analysis by Ali-ud-din Khan (Khan, 2014), income has a positive relationship with consumption following the function:

C = 2401.027 + 0.832Y With C being consumption, Y is the income The parameter shows that with every 1 unit increase in income, consumption goes up by 0.832 The F-test gave a result of 96.465, implying the great significance of the model (Khan, 2014, p 5)

Another research done on the Chinese urban resident income levels from 2002 to

2012 using the cluster analysis shows the same pattern (Fangfang Hou, 2015, p 251 ff.) However, at different income levels, the coefficients are different, showing a more in-depth implication of the theory Furthermore, the marginal propensity to consume decreases as income raises

In 2018, research was done in Shangcha, China They applied the Irrational Keynesian Consumption function and Irrational Life-Cycle Consumption Function Model (Yan & Liting, 2018, p 1517 ff.) They have gathered statistics from two different areas: rural and urban In this research, they have also recognized a significant dependence of consumption on income

A research that was done with a similar model to this work was done in Nigeria (Onanuga, et al., 2015) They have built a similar consumption function based on Keynes’ and used OLS method and error correction mechanism (ECM) technique to estimate the model The goodness of fit was 0.85, showing high credibility of the model Moreover, the coefficient of income stood at positive 0.64, demonstrating a positive relationship between income and consumption The result is close to previous research done in the same country, which result was 0.67 The difference in

parameter was due to the difference in time period, according to their speculation The relationship was also proved to be non-proportional, which contradicts Keynes’ belief that consumption is reversible if income decreases This means that if income rises, so does consumption; but when income falls, consumption does not decrease accordingly (Onanuga, et al., 2015, p 12)

Further, this work will analyze the impact of inflation rate on consumer expenditure

in Germany Previous research implies a positive relationship between inflation and customer spending Obinna (2020) has analyzed the impact of inflation on household consumption expenditure in Nigeria from 1981 to 2018 Using the OLS estimation technique and data from the Central Bank of Nigeria, evidence was given that households spend more money on goods and services during high inflationary periods than in periods with lower inflation The estimation of the model led to a small positive influence of inflation on consumption expenditure with a coefficient of 0.1124 (Obinna, 2020, p 109 f.)

The findings of Obinna (2020) are supported by Effah Nyamekye and Adusei Poku (2017) The empirical study observes the impact of inflation on consumer spending based on data from Ghana for the period 1964 to 2013 After performing an OLS estimation, it was concluded that inflation exerts a positive impact on consumer spending in Ghana The results demonstrate that a 1.00% increase in inflation leads

to a 19.20% increase in consumer spending (Effah Nyamekye & Adusei Poku, 2017,

p 5)

De Mello and Carneiro (2000) analyzed the consumption behavior during high inflation in the Latin American countries Venezuela, Peru, Argentina, Brazil, Paraguay and Uruguay Similar to the previously mentioned studies, the OLS estimation method was applied The research supports the findings that inflation positively affects consumption (de Mello & Carneiro, 2020, p 241) The results demonstrate that the presence of persistently high inflation results in volatility in consumption expenditure (de Mello & Carneiro, 2020, p 243) The increase of consumption during high inflation periods is reasoned with the differences between perceived and observed income (de Mello & Carneiro, 2020, p 239)

Additionally, the effect of interest rates on German private consumption spending will be examined in this paper In several researches and studies a negative relationship between interest rates and consumer expenditures has been indicated Hence, an increase in interest rates is predicted to result in lower consumption which Keynes states to be true (Keynes, 2018, p 98)

Frederic Mishkin supports this theory that tight monetary policy, which raises interest rates in a country is contributing to a decrease in consumer purchases and thus states that interest has a negative impact on expenditures (Mishkin, 1976, p 652) This asymmetry between interest rates and consumption has been investigated for a case

in South Africa for the period 2000 to 2018 using data from the Statistics South Africa and the South African Reserve Bank (Fikizolo, 2020, p 19) By using the OLS estimation and Vector Autoregressive approach in this study, it was proven that a negative relationship between interest rate changes and final consumption exists (Fikizolo, 2020, p 20 ff.) The results of the model estimation through economic and econometric tools reveal that an increase of 1.00% in the interest rate has led to a reduction of around 0.60% in private consumption expenditure Simultaneously, a 1.00% decrease in the interest rate caused a rise in consumption by approximately 0.56% (Fikizolo, 2020, p 40)

The results of this study are further validated by Kapoor and Ravi (2009), who examined the causal effect of a higher interest rate on household consumption expenditure by utilizing an Indian banking legislation and investigating detailed data

of the National Sample Survey from 2000-2001 and 2005-2006 (Kapoor & Ravi,

2009, p 2 f.) Applying the OLS estimation method and regression discontinuity approach to evaluate the effects (Kapoor & Ravi, 2009, p 12 ff.), the main result demonstrates that an increase of 50 basis points (0.50%) in the interest rate leads to

an immediate decrease of 12.00% in consumption spending This indicates the existence of immediate substitution of consumer expenditure when interest rates increase (Kapoor & Ravi, 2009, p 20 f.)

Furthermore, an empirical work determines the factors influencing private consumption in Lesotho from 1982 to 2013 using economic data from the Central

Bank of Lesotho, the Ministry of Finance and Bureaus of Statistics (Sekantsi, 2016,

p 66) This study presents evidence that higher interest rates reduce private spending (Sekantsi, 2016, p 58) An error correction model was assessed by using the Autoregressive Distributed Lag approach to cointegration as well as other instrumental variables like the OLS were used (Sekantsi, 2016, p 63 f.) This paper confirms the previous results that higher interest rates negatively affect consumer spending Moreover, the reduction in consumption due to higher interest rates is reasoned by the substitution effect Hence people defer their consumption when opportunity cost of immediate spending is higher (Sekantsi, 2016, p 71)

After analyzing previous research and theories the following hypotheses are stated regarding the relationship between the independent variables and private consumption expenditure in Germany and will be tested in this paper:

(I) Income has a positive impact on private consumption expenditure (II) Inflation exerts a positive influence on private consumption expenditure (III) Interest rate negatively impacts private consumption expenditure

3 Model Specification

3.1 Methodology

The method used to estimate the model in this paper is the Ordinary Least Square approach This is a type of linear least squares method to estimate unknown parameters of a linear regression model Here, the estimates minimize the sum of squared residuals The OLS is built based on 6 assumptions:

1 The regression model is linear in the parameters

2 The population mean of disturbances is 0

3 The disturbance is homoscedastic

4 There are no correlations between disturbances

5 The model is correctly specified

6 All independent variables are uncorrelated with the disturbances

When all of the conditions above are met, the OLS estimators is considered BLUE (Best Linear Unbiased Efficient) according to Gauss-Markov theorem (Gujarati & Porter, 2009, p 92)

With the application of the OLS model it will be able to derive unbiasedness, consistency, and other important statistical properties efficiently The model proved

to be successful with high goodness of fit and relatively similar estimators between consecutive tries within the same region

In this work, the OLS method will be applied on a linear regression model This will

be specified based on the previously outlined theories and research The following chapter will describe the process of model specification, data collection and the analysis of relationships between the independent variables in detail

To extract data, the authors relied on acknowledged sources such as the Federal Statistical Office The model will then be estimated using Stata Focus of this estimation will be the determination of the regression coefficients and coefficient of determination Additionally, with the usage of the t-test and F-test, the statistical significance of the independent variables will be tested

3.2 Theoretical Model Specification

One of the most popular models has been established by John Maynard Keynes who investigated the influence of income on consumption

𝐶 = 𝐴 + 𝑀 ∗ 𝐷 Where:

A: Autogenous income D: Disposable income

From this a more detailed model is derived to represent the impact of different factors

on consumption The linear regression model in this paper contains three independent variables, income, interest rate and inflation, which are the factors influencing the dependent variable of consumption spending The stochastic disturbance term u represents the variables that are not involved in the model but might affect the dependent variable (Gujarati & Porter, 2009, p 62) This error term includes non-observable factors that are not further specified or the regarding data is not available but still have an effect Hence, the expected future income as well as the consumption preferences are not further assessed in this work, despite having an evidenced effect

on private spending (Jappelli & Pistaferri, 2010, p 480 f.)

In this paper, the income in Germany is described by the Gross National Income (GNI), the interest rate is the long-term interest rate, and the inflation is illustrated by the Consumer Price Index (CPI) of all items and the consumption spending by the private final consumption expenditures

Hence, the following population regression model can be established:

𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 = 0+1∗ 𝐺𝑁𝐼 +2∗ 𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 +3∗ 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛 + 𝑢𝑖Where:

0: intercept term 3: regression coefficient of Inflation

1: regression coefficient of GNI 𝑢𝑖: error term

2: regression coefficient of Interest rate

For the sample data collected, the following sample regression model is developed: 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 = 

0+

1∗ 𝐺𝑁𝐼 +

2∗ 𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 +

3∗ 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛 The corresponding stochastic form of the sample regression model is concluded: 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 = 

0+

1∗ 𝐺𝑁𝐼 +

2∗ 𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 +

3∗ 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛 + 𝑢𝑖Where:

of employees, property income and net taxes deducted by subsidies on production (OECD, n.d.) For the purpose of this work, the GNI is seen as superior proxy compared to the GDP GNI measures the income of the residents of a country This income is likely to be spent within this country and thereby affecting consumption GDP measures the value of the production of a country which might generate income abroad Throughout this work, GNI will be displayed in billion euros

To measure the long-term interest rates, yields of 10-year government bonds are used The interest rates imply rates that bonds are currently traded, not the rates at which

the bonds have been issued Thus, the current market interest rates are reflected Long-term interest rates of government bonds encourage either savings or consumption and are therefore well fitted for the model The interest rates will be displayed in percentage (OECD, 2021)

The proxy for the inflation is the Consumer Price Index The CPI measures the price increases of a defined basket of goods and services representing the goods and services purchased by households in the country (Federal Statistical Office, n.d.) The CPI will be shown in points using the year 2015 as basis where the CPI equals 100 points

Consumer expenditure is the dependent variable of the applied model As described

in the outlined studies and the theoretical background of Keynes, income, interest rate and inflation are independent variables affecting consumption expenditure However, the independent variables impact each other as well Higher interest rates lead to weaker economic growth and thus slower inflation and vice versa Low inflation signals a slow economic growth The decision makers of monetary policy would aim

at lowering the interest rates to lower the cost of borrowing and foster economic growth During high inflation phases, interest rates will be raised to slow economic growth and inflation (McArdie, 2014, p 2 f.) Additionally, research of the Federal Reserve Bank of St Louis has provided evidence that inflation positively impacts income The study has shown that if inflation increases, nominal wage growth also rises (Sanchez, 2015) A well-known model explaining the relationship between interest rates and income is the IS-LM model of Keynes Assuming that a country is closed to international trade, a fall in interest rates, for example through an increase

in money supply, will cause the economy to purchase assets leading to an increase in output, income and employment of an economy to increase Interest rates would fall and income increases until money demand and money supply are in an equilibrium (Krugman & Wellls, 2018, p 640 ff.) However, there are no recent empirical studies supporting this relationship between income and interest rates

3.3 Sample Data

To estimate the population parameters a credible sample model must be created which involves obtaining reliable data from valid sources Thus, in this paper the data regarding the variables have been extracted from reputable websites Information about consumption expenditure and CPI in Germany have been attained from the Federal Reserve Bank of St Louis (FRED) This online database consists of numerous economic data time series from several public and private, international as well as national sources On FRED, the user is provided tools to comprehend and interact with the displayed data in multiple ways (FRED, n.d.)

One major original source that FRED uses to illustrate data in meaningful graphs is the Organization for Economic Co-operation and Development (OECD) This international organization works collaboratively with governments, policy authorities and the public and provides a collection of knowledge, data and analysis as well as advice for public policy decisions (OECD, n.d.)

The data of the dependent variable PCFE in Germany has been reported by the OECD and retrieved from FRED Here, the displayed frequency is in quarters starting in

1970 Q1 and the currency units are in Euros (FRED, 2021)

The GNI which represents the income generated by residents in Germany is retrieved from the Federal Statistical Office Destatis, which published the national accounts data from 1925 to 2020 in August 2021 (Destatis, 2021) This organization provides neutral and professional statistics regarding economic, environmental or public issues

in order to display past and present data for decision-making and detailed comprehension of information (Destatis, n.d.)

Next, the CPI of all items in Germany which is also reported quarterly since 1960 Q1

by the OECD was obtained from FRED (FRED, 2021)

As the last independent variable, information regarding the interest rates in Germany were acquired from OECD data which illustrates the long-term interest rates starting

in 1953 (Federal Statistical Office, n.d.)

In this work, a sample model using 51 observations for each variable is developed in order to display a reliable time period of 51 years with data corresponding for each year since 1970 until 2020 (Appendix 1)

To summarize and organize the characteristics of this dataset effectively, the adjusted data has been inserted into Stata-64 to assess the sample statistics for each variable in detail (Appendix 2)

For the dependent variable private consumption expenditure, the sum aggregates to 50,525.643€ billion and a mean of 990.699€ billion has been calculated which represents the average of the consumption spending during this period The arithmetic mean reveals the center of the data where outliers have less effect on the central tendency and hence the expected value of a variable (Wooldridge, 2012, p 704) The minimum of the sample data amounts to 222.369€ billion which is the first observation in 1970 Consequently, the lowest consumption spending was in the initial year, and the following years all record a larger expenditure The greatest value

of consumption has been recorded in 2019 with 1,808.710€ billion which is almost double the average mean The standard deviation of this data amounts to 470.081€ billion which indicates a relatively large spread around the mean because it is affected

by outliers

Secondly, the independent variable GNI has been analyzed and a total of 90,698.028€ billion has been recorded from 1970 to 2020 which results in a mean of 1,778.393€ billion Additionally, this calculated average income in Germany has been affected

by the minimum amount of 261.540€ billion in the first observed year and the maximum of 3,585.963€ billion in 2019 which represent the most significant outliers

in this dataset The permanent increase in GNI suggests that income in Germany has grown during the period, however in 2020 a small decrease of 124.678€ billion has been reported This small decline in the GNI can be reasoned by the COVID-19 pandemic, which affected the economy globally Furthermore, the spread indicated

by a 962.561€ billion standard deviation around the mean implies that the values vary notably throughout these years