How does obesity affect economic growth in the Netherlands and Finland?

1

How does obesity affect economic

growth in the Netherlands and

Finland?

Bachelor Thesis

Faculty Economics and Business

ABSTRACT:

This paper separately investigates the impact of obesity on economic growth

in the Netherlands and Finland, between 1981 and 2014. It covers the

importance of the neoclassical approach of Solow (1956) and endogenous

growth model of Arora (2001). By assuming an endogenous growth approach

and time series data, we construct Ordinary Least Squares (OLS) regressions

to explain economic growth by obesity rates, lagged GDP, lagged growth, net

Foreign Direct Investments and education. We find a significant positive

effect of obesity on economic growth in the Netherlands, whereas there is no

effect in Finland. This implies that obesity promotes economic growth in the

Netherlands, which contradicts our first expectations. Further research on the

relationship between lagged GDP and obesity rates is needed.

Author:

Tufan Kiziltekin

Student number:

10650024

Year:

2016-2017

Supervisor:

Rutger Teulings

Track:

Economics

2

Statement of Originality

This document is written by Tufan Kiziltekin, 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

C

ONTENTS

1. I

NTRODUCTION

... 4

2. E

CONOMIC

G

ROWTH

... 5

2.1

I

NTRODUCTION

... 5

2.2

S

OLOW MODEL

(

EXOGENOUS

) ... 6

2.2.1

S

TEADY

S

TATE

... 8

2.2.2

R

EMARKS

... 10

2.3

E

NDOGENOUS MODEL

... 11

2.3.1

D

EFINITION

... 12

2.3.2

E

FFECTS OF

TFP

GROWTH

... 13

2.3.3

D

ETERMINANTS

TFP ... 14

3. O

BESITY

... 15

3.1

D

EFINITION

... 15

3.2

E

CONOMIC COSTS OF OBESITY

... 16

3.2.1

M

ARGINAL

R

ATE OF

T

IME

P

REFERENCE

(MRT) ... 17

4. M

ETHODOLOGY

... 19

5. D

ATA

... 20

5.1

D

ESCRIPTION

... 20

5.2

S

UMMARY

S

TATISTICS

... 22

6. R

ESULTS

... 23

7. C

ONCLUSION

... 25

8. R

EFERENCES

... 25

9. A

PPENDIX

... 29

9.1

S

OLOW MODEL

... 29

9.2

OLS

ASSUMPTIONS

... 29

4

1. Introduction

In the 1970s, obesity is reported as one of the most important nutritional diseases. Currently, obesity rates show a powerful increase and differ per country (OECD Health Statistics, 2016). Fogel (1994) argues that gross nutrition (diet) explains a third of income growth in Britain between 1790 and 1980. This reflects the importance of nutrition on economic growth. Moreover, Philipson & Posner (2008) perceive obesity as a major issue of health economics and public finance.

In general, healthier employees are more productive and earn higher wages (Bloom et al., 2004). Then, it might be that obesity hurts the productivity levels and decreases general income and consequently reduces life quality. In addition, obesity boosts the

marginal rate of time preference (MRT) (Komlos et al., 2004). This rate expresses the value of future consumption in terms of current consumption (Smith et al., 2005). In the case of health issues, future consumption usually reflects future health risks. Thus, obese individuals with a higher MRT in comparison to healthy people prefer current food consumption over the future health risks of obesity. This refers to an increase in aggregated consumption, which might stimulate economic growth. Finally, obesity raises medical expenditures and stresses the health care system (Baum & Ruhm, 2009, p.635).

Arora (2001) argues that health increases the pace of economic growth in

industrialized countries, which permanently boosts the slope of their growth path. Gyimah-Brempong & Wilson (2004) confirm the positive relationship between health and economic growth in the member countries of the Organisation for Economic Co-operation and Development (OECD). Then, if people and governments consider health a significant indicator of their economic wellbeing, they will approach health differently. For example, George Osborne, chancellor of the exchequer between 2011-2016, announces that the UK plans on introducing sugar taxes on soft drinks in 2018 to fight obesity (BBC, 2016).

Despite various papers on the impact of health, few studies show the magnitude of obesity on economic growth. This paper examines this relationship and anticipates a negative impact of obesity. The World Health Organization (2016) determines a doubling in worldwide obesity rates in the period after 1980. Therefore, the research period of this paper is 1981-2014. The sample includes the Netherlands and Finland, which are the only two OECD countries with consistent data regarding obesity. Moreover, we assume an

endogenous growth model for these countries and use time series data to construct Ordinary Least Squares (OLS) regressions. Finally, we find a significant positive effect of obesity on economic growth in the Netherlands, whereas it has no effect on the economic growth of Finland. This partly satisfies the hypothesis of literature but contradicts our first expectations. The structure of the paper is as follows. First, chapter 2 and 3 cover literature

5 introduces an empirical model to explain economic growth. Moving on, chapter 5 discusses the data. The results of the research follow in chapter 6 and the conclusion follows in chapter 7. Finally, the remaining two chapters consist of the references and appendix.

2. Economic Growth

This chapter analyses economic growth. It discusses the neoclassical growth model of Solow (1956) and the endogenous growth models of Romer (1981) and Arora (2001).

2.1

Introduction

An essential indicator of economic activity in a certain period is the Gross Domestic Product (GDP) (England, 1998). In addition, a more accurate estimation to underpin

economic progress is to correct the GDP for inflation, which refers to real GDP (Rodriguez & Sanchez, 2005). Often, to determine the standards of living, the GDP is expressed per capita. The higher this value, the better the standards of living.

It is possible to determine GDP (Y) in three different ways: output, expenditure and income. In principle, these measurements should result in the same value of GDP. So, production is equal to expenditures and income (Office for National Statistics, 2016). We cover the two most relevant approaches in this paper, namely the output approach and expenditure approach. Equation (1) defines the output approach, which puts emphasis on the value added due to production (Landefeld et al., 2008). In contrast, the expenditure approach interprets Y as the sum of consumption (C), government spending (G), investments (I) and net export (NX), see equation (2) and (3).

𝑌 = 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐺𝑟𝑜𝑠𝑠 𝑂𝑢𝑡𝑝𝑢𝑡 − 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐼𝑛𝑝𝑢𝑡 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑠

(1)

𝑌 = 𝐶 + 𝐺 + 𝐼 + 𝑁𝑋

(2)

𝑁𝑋 = 𝐸𝑥𝑝𝑜𝑟𝑡 − 𝐼𝑚𝑝𝑜𝑟𝑡

(

3)

The degree in which (real) GDP changes, reflects economic progress. Then, we need the GDP in the current period and the lagged GDP to track economic progress. Economic growth occurs when there is an increase in GDP. Usually, we have special interest in equation (4) and (5), which define the economic growth rate.

𝑌(0) = 𝑙𝑎𝑔𝑔𝑒𝑑 𝐺𝐷𝑃 ; 𝑌(1) = 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝐺𝐷𝑃

(4)

𝑌(1)−𝑌(0) 𝑌(0)

=

𝑑𝑌

𝑌

= 𝑒𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑔𝑟𝑜𝑤𝑡ℎ

(5)

In fact, there are various theories which try to determine and explain economic growth, each with different parameters and assumptions. In general, an economic model

6 exists of exogenous and endogenous variables (Jones, 1998). A commonly used

macroeconomic exogenous model to explain economic growth is the one of Solow (1956), which describes a neoclassical growth approach. An alternative approach is the endogenous growth model (Romer, 1989). Both approaches define four major sources of economic growth: labor (L), capital (K), human capital (H) and productivity levels (A). The following sections will broaden the exogenous and endogenous growth models.

2.2

Solow model (exogenous)

The work of Solow (1956) is a neoclassical approach for long term economic growth. For example, he assumes full employment under the population and a fixed savings rate. This model defines GDP as total output production and predicts this with equation (6). From this equation, it follows that output depends on labor and capital.

Next, we introduce labor-augmenting technology in the production function (6), which results in equation (7). This function expresses labor as effective labor (AL). Solow (1956) assumes that the production function has constant returns to scale (CRS). This states that the input products have a one-to-one relationship with the output, For example, if z=2 in this equation, a twice as large capital stock results in a doubling in output, which indicates that the production function is homogenous of degree 1. With this in mind, equation (8) defines capital per effective labor (k) and output per effective labor (y). Finally, we combine functions (7) and (8) and define z as 1/AL, which results in equation (9). This shows the intensive form of Solow (1956), which makes output per effective labor a function of capital per effective labor.

𝑌 = 𝐹(𝐾, 𝐿)

(6)

𝑌 = 𝐹 (𝐾, 𝐴𝐿) → 𝑧𝑌 = 𝐹(𝑧𝐾, 𝑧𝐴𝐿)

(7) 𝑌 𝐴𝐿

= 𝑦 ;

𝐾 𝐴𝐿

= 𝑘 ;

𝐴𝐿 𝐴𝐿

= 1

(8) 𝑌 𝐴𝐿

= 𝐹 (

𝐾 𝐴𝐿

,

𝐴𝐿 𝐴𝐿

) = 𝐹 (

𝐾 𝐴𝐿

, 1) = 𝐹 (

𝐾 𝐴𝐿

) ; 𝑦 = 𝑓(𝑘)

(9)

The impact of k on y is as follows. When k is low, there is relatively more effective labor than capital. Then, effective labor has relatively low capital to perform with, such that the output per effective labor is low as well, see equation (9). Because the labor force is considerably efficient and the capital stock is low, a slight increase in the capital stock is a sizeable stimulus for production. Vice versa, when k is high, there is relatively more capital than effective labor and equation (9) shows that output per effective labor is high as well. Then, the return of additional capital decreases, because there is already a considerable capital stock and there is no sufficient effective labor to handle it.

7 So, the neoclassical model assumes diminishing benefits of capital: the first order condition of y with respect to k is positive, while the second order condition is negative. Hereby, Solow’s model (1956) meets every INADA-condition in part 9.1 of the appendix (Correia, 1996). As a consequence, the production function is concave, as could be seen in figure 1. This means that if we move along the y-curve, the growth rates decelerate. Thus, the initial growth and initial levels of y are major indicators of future growth. Table 1

summarizes these findings. Initial

Levels

Low High

k New units of capital are valuable: high benefits / productivity

lim

𝑘→0

𝑓′(𝑘) = ∞

New units of capital are less valuable: low benefits / productivity.

lim

𝑘→∞

𝑓′(𝑘) = 0

y -Production increases quickly -More economic growth

-Production increases slowly. -Less economic growth Table 1: summary indicators in Solow (1956)

Figure 1: production function Solow, source Economic Discussions (2016)

The next step is to understand the underlying mechanisms affecting k. First, equation (8) shows that if the capital stock grows more rapidly than effective labor, k increases.

Capital accumulation only appears if economic agents invest (I) a part of their income. Subsequently, equation (10) describes investments, which reflects that economic agents invest an exogenously determined savings rate (s) of their income.

𝑌 𝐴𝐿

= 𝑦 ;

𝐾 𝐴𝐿

= 𝑘 ;

𝐴𝐿 𝐴𝐿

= 1

𝐼 = 𝑠 ∙ 𝑓(𝑘)

(10)

8 Second, the exogenous depreciation rate (δ) hurts k. Moment gleaming capital

becomes less valuable due to usage: the more production it undergoes, the more it loses its value (Parker, 2012). Equation (11) defines the growth in productivity (g) and growth in labor (n). These growth rates increase the denominator of k. Intuitively, if employees become more effective or grow, they desire more capital to keep k constant. Thus, the exogenous growth rates g & n are negative forces for k.

𝑑𝐴

𝐴

= 𝑔 ;

𝑑𝐿

𝐿

= 𝑛

(11)

In sum, investment conflicts the depreciation rate, growth in technology and growth in labor to keep k constant. With this in mind, we define the total change in capital per effective labor (ǩ) as follows:

ǩ = 𝑠 ∙ 𝑓(𝑘) − (𝛿 + 𝑛 + 𝑔)𝑘

(12)

2.2.1 Steady State

The key in Solow (1956) is the long run (stable) equilibrium: steady state. This indicates that the net change in k is zero, see equation (13). Then, equation (14) describes that there is an optimum (k*), in which investments are equal to the break-even point of k (𝛿 + 𝑛 + 𝑔). Moreover, figure 2 plots the situation under steady state and shows that if k does not change, y does not alter either. This indicates that capital (K) and output (Y) grow at the same pace as effective labor (AL), which follow from the definitions in equation (8), presented in the previous section. In short, the economy is at its balanced growth path, see table 2.

ǩ = 0 ; 𝑠 ∙ 𝑓(𝑘 ∗) = (𝛿 + 𝑛 + 𝑔)𝑘*

(13)

𝑘 ∗=

_{(𝛿+𝑛+𝑔)}𝑠 ∙ 𝑓(𝑘∗)

=

𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑠 𝐵𝑟𝑒𝑎𝑘𝑒𝑣𝑒𝑛−𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡

(14)

𝑌

𝐴𝐿

= 𝑦 ;

𝐾

𝐴𝐿

= 𝑘 ;

𝐴𝐿

𝐴𝐿

= 1

9 Figure 2: steady state level in Solow (1956), source EUR Macro (2016)

Required investment =

𝒌 (𝜹 + 𝒏 + 𝒈)

Potential output/labor =

𝒇(𝒌)

Savings =

𝒔

∙

𝒇(𝒌)

Variable Symbol Steady-State Growth

Capital per effective labor 𝐾 𝐴𝐿= 𝑘 0 Output per effective labor 𝑌 𝐴𝐿 = 𝑦 0 Capital per labor 𝐾 𝐿 = 𝑘𝐴 g Output per labor 𝑌 𝐿 = 𝑦𝐴 g Total Output 𝑌 = 𝑦(𝐴𝐿) n+g Total Capital 𝐾 = 𝑘(𝐴𝐿) n+g

Table 2: summary conditions Steady State (Balanced Growth Path)

Looking at table 2, the only source to raise living standards (Y/L) is g, which is an exogenous variable in Solow (1956). Thus, we solely depend on an external shock to improve our standards of living. Then, it is interesting to know how the economy reacts after a positive external shock on g, while being in steady state.

Equation (14) describes that an increase in g increases the break-even investments. Consequently, the economy evolves towards a lower k (k**). Figure 3 presents this

mechanism from a graphical perspective. The break-even investment line becomes steeper, see black line of required investments. Then, it intersects with actual investments at k**,

10 which is lower than the initial steady state level k*.

If we analyze the total effect, table 2 and figure 4 show that an increase in g causes a jump of Y and K to a higher level, which refers to the level effects. When the economy is back at its balanced growth path, everything remains constant. Thus, Solow (1956) concludes that if technology grows faster, it results in a permanent level effect and a temporary growth rate effect.

𝑘 ∗=

_{(𝛿+𝑛+𝑔)}𝑠 𝑓(𝑘∗)

=

_{𝐵𝑟𝑒𝑎𝑘𝑒𝑣𝑒𝑛−𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡}𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑠

Figure 3: level effect due to an increase in productivity, source EUR Macro (2016)

Figure 4: productivity shock in Solow (1956)

2.2.2 Remarks

A power of Solow’s model (1956) is that it contributes to the economic growth theory. It accepts that the economy transitions towards a steady state level, which is a consequence of savings behavior, population growth, technological growth and depreciation rates. Furthermore, equation (15) presents the augmented Solow model (Mankiw et al.,1992), in which human capital (H) enters the production function. Mankiw et al. (1992) argue that human capital accelerates the way to steady state, which indicates faster growth in the initial periods.

11

𝑌 = 𝐹(𝐾, 𝐻, 𝐴𝐿)

(15)

Moreover, if countries express the same characteristics, they have the same steady state level (k*, y*). A poor country, with lower levels of capital and output, grows faster than the richer country, until it reaches the same steady state. This describes convergence and implies that the poor country can catch-up the richer country (Mankiw et al., 1992).

However, the model also has limitations. Namely, economic growth in the

neoclassical approach has an external aptitude and implies that on the long run economic growth slows down. The determinants of factors affecting steady state are unknown, it is something outside the model, which makes the theory nonconcrete. For example, Solow (1956) perceives the upward trend of the standards of living in figure 5 as a consequence of a boost in an exogenous variable g. In short, the model fails in explaining international income differences (Prescott, 1998).

1980 1990 2000 2010 2015 Figure 5: GDP/capita in $ in OECD countries between 1980 and 2015, source OECD GDP (2017)

2.3

Endogenous model

Romer (1989) is one of the first footprints of endogenous growth models. The main objective is to internalize the factors affecting economic growth instead of accepting it as exogenous. The proponents of endogenous growth models argue that economic growth does not depend on the initial levels of output per capita and reject a certain steady state level (Mankiw et al., 1992, p.422). In fact, capital has no diminishing benefits anymore which can lead to different growth paths and cross country differences (Trier & Gullock, 1989).

12

2.3.1 Definition

First, equation (16) reflects the elements of a basic endogenous production function (Arora, 2001). It follows that this framework reflects output levels by productivity levels (A) and reproducible factors (K), including intangible – and tangible capital. Moreover, the model does not restrict intangible capital to human capital but includes knowledge as well (Mankiw, 2010, p. 240). Next, it fights the assumption of diminishing returns of capital in Solow (1956) by assuming a constant marginal productivity of capital (MPK) and constant marginal

productivity of labor (MPL) (Arora, 2001).

𝑌 = 𝐴(𝑑)𝐾

(16)

For this paper, the most relevant deviation from a neoclassical approach is that productivity (A) reflects an endogenous variable, which depends on potential determinants (d) (Mankiw, 2010, p.240). Because there are various proxies for productivity, there is no exact definition of it. Hence, we prefer the term total factor productivity (TFP) to indicate productivity levels, which is hard to observe (Nelson, 1973).

Accordingly, we use an estimation technique to measure the impact of the growth in TFP on economic growth. This technique follows from equation (17), which reflects growth accounting (Nelson, 1973, p.464). The purpose of growth accounting is to account for different sources of economic growth and determine the magnitude of each source. So, it breaks down the impact of growth in TFP, labor and capital on economic growth. The advantage of using growth accounting is that the growth in output, capital and labor are straightly perceivable. Then, we reform equation (17) to equation (18) and measure the growth of TFP (dA/A) by subtracting the effect of labor and capital from economic growth. Thus, we determine the growth in TFP by a residual, which we call the Solow-residual. Growth accounting is less relevant in the neoclassical approach, since the Solow-residual is exogenous (g). Δ𝑌 Y

=

(𝑀𝑃𝐾 ∙ 𝐾) 𝑌

∙

Δ𝐾 𝐾

+

(𝑀𝑃𝐿 ∙ 𝐿) 𝑌

∙

Δ𝐿 L

+

Δ𝐴 A (17) Δ𝐴 A

=

Δ𝑌 Y

−

(𝑀𝑃𝐾 ∙ 𝐾) 𝑌

∙

Δ𝐾 𝐾

−

(𝑀𝑃𝐿 𝑥 𝐿) 𝑌

∙

Δ𝐿 L (18)

13

2.3.2 Effects of TFP growth

Figure 6: changes in total factor productivity 1961-2009, source ACIAR (2017) Looking at figure 6, we observe a boost in the growth of TFP in developed -, developing - and transitional countries. We use the endogenous model of Arora (2001, p.708) to reflect the consequences of this development for economic growth, see equation (19).

𝑔 =

𝐴 − 𝜌 −𝛿

𝜎

(19)

It follows that the economic growth rate depends on TFP (A), marginal rate of time preference (

𝜌

), depreciation rate

(𝛿)

and risk aversion

(𝜎).

Then, other things equal, growth of TFP permanently boosts the economic growth rate. This differs from the interpretation of Solow (1956), which is presented in section 2.2.1: growth in productivity has temporary effects on economic growth. Figure 7 summarizes these findings.

Exogenous model Endogenous model

14

2.3.3 Determinants TFP

The previous section shows that the growth in TFP has an upward trend. The next step is to analyze some determinants of TFP.

I.

Foreign direct investment (FDI)

Foreign Direct Investment (FDI) refers to taking a controlling position and having decision power in a foreign firm (De Mello, 1999, p.134). In other words, FDI allows countries to interact with each other, which might affect TFP and economic growth.

First, FDI enhances the investment flows and therefore might promote economic growth through capital accumulation. Second, it results in more product variety within the recipient country and makes the production process more efficient as well (Borensztein et al., 1998, p.121). Consequently, the costs of implementing technology are lower and in return the demand for new technology is higher. Finally, FDI results in cross-border externalities generating knowledge. The outflow and inflow of foreign investments transfer knowledge from one country to another. This might take the form of labor training, skill acquisitions or promotion of new technology (De Mello, 1999).

Next, Borensztein et al. (1998) regress economic growth on FDI for a sample of 32 OECD and non-OECD countries. They conclude that the independent effect of FDI on economic growth is a point of concern. It seems that FDI negatively affects economic growth in countries with a low level of human capital. However, when FDI interacts with higher sufficient level of human capital, it contributes to economic growth. In short, a sufficient human capital stock is a necessity to generate a positive effect of FDI.

II.

Human Capital

Benhabib & Spiegel (1994) conclude that human capital significantly stimulates TFP: the higher the stock of human capital, the higher the productivity levels. There are two major mechanisms through which this effect takes place. First, human capital contributes to the research & development (R&D) and diffusion of new knowledge (Engelbrecht, 1997, p.1486). Second, human capital is responsible for the absorption of new knowledge. In general, human capital has various proxies. The most important investment of human capital is education: the higher the level of education, the higher the stock of human capital and its contribution to TFP (Becker, 1994, p.17). In fact, education has two

measurements: quality and quantity. For example, a quantitative measurement of education might be the total amount of years of education. The level of education (primary/

secondary/tertiary) indicates a qualitative measurement. According to Barro (2001), both measurements of education significantly contribute to economic growth. In addition, he

15 concludes that the effect of education is more significant at the secondary and higher levels of education.

3. Obesity

Arora (2001) finds that health enhancement increases the ability of human capital and

boosts economic growth rates of industrialized countries. In addition, Fogel (1994) concludes that gross nutrition (diet) clarifies a third of income growth in Britain between 1790-1980. Cawley et al. (2015) determine a significant increase in obesity rates, looking at the past 30 years. This chapter explains the relationship between obesity and economic growth.

3.1

Definition

The definition of obesity is excessive weight accumulation (Baum & Ruhm, 2009, p.636). An individual’s weight depends on the energy balance: calorie intake (food) versus calorie expenditure (activity) (Fogel, 1994, p.5). Baum & Ruhm (2009, p.637) argue as follows:

‘Individuals are more likely to gain weight if they have a high-energy intake and low metabolism or physical activity.’

So, weight gain occurs if an individual takes more calories than needed (expended), which results in a surplus of calories. The higher the caloric surplus, the more weight gain. Then, if weight gain accumulates over time, it might lead to obesity, which is in essence the most extreme case of being overweight (Chou et al., 2004). In fact, overweight and obesity are not exactly the same. We distinguish them by checking the Body Mass Index (BMI) (Huxley, 2014). This index divides the weight in kilograms by the square of height in meters (kg/m2), which indicates the relative weight.

The BMI has different categories and interpretations (Built Lean, 2016). To begin with, BMI degrees between 18.5 and 24.9 refer to normal weight, whereas the range 25-29.9 indicates overweight. Obesity is only present if the BMI is equal to or greater than 30.0. In specific, BMI measures between 30 and 34.9 refer to moderate obesity, while the ranges greater than 35 indicate severe obesity (Finkelstein et al., 2010). These measurements hold for male and female. The following figure summarizes the classes of BMI:

16 Figure 8: classes BMI, source University of Utah Health Care (2016)

Obesity has lifetime health consequences, which hurt the life quality. It increases the risk of developing a variety of other health problems such as coronary heart diseases, high blood pressure, type 2 diabetes and many forms of cancer (National Obesity Observatory, 2010). Furthermore, Mokdad et al. (2003) argue that the higher the degree of obesity, the higher the future health risks.

3.2

Economic costs of obesity

Obesity bears economic costs (Baum & Ruhm, 2009, p.635). For example, the total

economic cost of obesity in England, between 1998 and 2002, increased from 2628.9 million to 3340 million pounds (£). In addition, in the United States, the total costs of obesity for full-time employees equals $73.1 billion per year (Finkelstein et al., 2010).

The economic costs of obesity consist of two parts: direct costs and indirect costs. The direct costs refer to the direct consequences of treating obesity, whereas the indirect costs indicate the adverse social consequences like decreased productivity levels (National Obesity Observatory, 2010). Obese individuals suffer from absenteeism due to premature mortality or sickness, which refers to not working at all and therefore leads to lost earnings (Gates et al., 2008).

In addition, presenteeism describes the health-related limitations at work. Gates et al. (2008) empirically validate the relationship between obesity and presenteeism, by running surveys on 341 manufacturing employees. They conclude that moderately or highly obese employees show a significant increase in the total time needed to complete a specific task or job, which results in a 4.2% loss of productivity. This is 1.18% more than their non-obese colleagues and leads to an annual decrease of $506 in general income.

Another channel affecting wages is medical expenditures. Finkelstein et al. (2010) conclude that obesity boosts annual medical expenditures. Then, if obese employees opt for the opportunity of employer-provided health insurance, the employer should pay a higher premium for obese employees in comparison to healthy workers. As a consequence, the

17 employer might pass these excess costs on to obese workers. Supplementing Finkelstein et al. (2010), Bhattacharya & Bundorf (2009) determine a considerable decrease in the wages of obese employees in comparison to their similarly insured non-obese colleagues. In specific, amidst workers with employer-provided health insurance, obese employees earn $1.42 less per hour than their non-obese colleagues.

However, if we approach GDP by the expenditure approach, higher health

expenditures contribute to economic growth, as could be seen from equation (2) presented in chapter 2. Figure 9 shows that there is indeed an upward trend in health expenditures in the Netherlands and Finland. In addition, the World Health Organization (2016) indicates an increase from 5.3% to 9.4% in health expenditures expressed as a share of GDP, in the period 1994 – 2014.

𝑌 = 𝐶 + 𝐺 + 𝐼 + 𝑁𝑋

Figure 9: health expenditures in current PPP per capita, source World Health Organization (2016)

3.2.1

Marginal Rate of Time Preference (MRT)

The next consequence of obesity is the rise in the MRT (

𝜌)

, which might slow down or eventually boost economic growth.

First, the MRT explains intertemporal choices by reflecting the trade-off between current consumption and future consumption (Smith et al., 2005). MRT is a mechanism to discount the future value to obtain the present value. Equation (20) describes this

mechanism by using 1 as the future value. It follows that the interpretation is essentially the same as the Discounted Cash Flow Method (Sloan, 1996). Regarding health issues, the MRT is a common measurement to explain health affecting behaviour. In the case of obesity, individuals make a trade-off between current food consumption and future health

18 risks.

1 (1 + 𝑀𝑅𝑇)

⁄

(20) Smith et al. (2005) indicate a positive relationship between BMI and MRT: the higher the BMI, the higher the MRT. Then, from equation (20) follows that obese individuals have a heavier discount factor than people with a healthy weight. So, they have less concerns about the future health risks of obesity. Thus, they put more emphasis on current food

consumption. Moreover, according to Smith et al. (2005), an increase in MRT elevates the risk of obesity. Supplementing this finding, Komlos et al. (2004) argue that people with a high MRT tend to exercise less. Combined with more current food consumption, the probability of the intake of excessive calories increases (Chou et al., 2004). Then, the whole process becomes a vicious circle: BMI stimulates MRT and MRT stimulates BMI, see figure 10.

Figure 10: vicious circle BMI and MRT

The next step is to understand the impact of MRT on economic growth. If we approach economic growth endogenously, an elevation of MRT deteriorates economic growth, see equation (19), elaborated in chapter 2. Another effect of the rise in MRT on economic growth goes through the channel of consumption. Looking at figure 10, a higher MRT results in more orientation towards current food consumption. So, individuals gain more utility by taking in more calories in the present. Consequently, if we approach GDP by the expenditure approach, obesity stimulates economic growth, see equation (2), which is discussed in chapter 2. So, the total effect of an increase in MRT on economic activity is ambiguous, it depends on the approach of economic interpretation.

𝑔 =

𝐴 − 𝜌 −𝛿_𝜎

𝑌 = 𝐶 + 𝐺 + 𝐼 + 𝑁𝑋

MRT Increased food consumption Decreased Exercising Caloric surplus BMI

19

4. Methodology

This chapter introduces and explains an Ordinary Least Squares (OLS) model to statistically test the effect of obesity on economic growth.

This paper prefers the endogenous growth model as the main starting point. This approach accepts that productivity shocks can permanently change the slope of economic growth rate, presented in chapter 2. It is only interesting to examine the effects of obesity if it has a long-term effect on economic growth, otherwise the impact of obesity would taper off on the long run and is therefore not a point of concern.

In this research, economic growth is the dependent variable. Moreover, 5

independent variables explain economic growth. Thus, the empirical model is as follows:

EG(t) = α + β1 OBS(t) + β2 GDP(t − 1) + β3 EG(t − 1) + β4 FDI (t) +

β5 AVG_ENROL(t) + ε

(21)

Variables:

α : constant

EG(t) : economic growth rate OBS(t) : obesity rates

GDP(t-1) : lagged GDP levels

EG(t-1) : lagged economic growth rate FDI(t) : foreign direct investments AVG_ENROL(t) : average education levels

ε : error term

This paper shows two separate regressions, which makes it possible to distinguish the effect of obesity in the Netherlands from the effect in Finland. Because of the restricted size of the dataset, we prefer a p-value of 0.10 (α=10%) to indicate a significant result. So, we have a 10% probability that the regression results do not hold. Both results follow from equation (21), which implies a linear relationship between the dependent and explanatory variables. Thus, this paper analyses economic growth by using the Ordinary Least Squares (OLS) method.

To check whether this method holds for our dataset, we need to formally check the following four OLS assumptions (Stock & Watson, 2012):

1. Linearity between the dependent and independent variables 2. No large outliers

3. Standard normally distributed error terms (µ=0 and σ=σ2₎ 4. No perfect multicollinearity

20 Figure 2.1, see appendix, presents the first two assumptions by plotting the linearity between the variables. It seems that there is indeed a linear relationship. In addition, these plots determine some outliers, which might be a point of concern. Linear prediction creates gaps between the observed and predicted data points, which is the definition of residuals. If the average of these residuals is zero and the residuals have an expected constant deviation from this average (constant variance), there are no big outliers (Stock & Watson, 2012). These conditions refer to normally distributed error terms (µ=0 and σ=σ2_{). Therefore, if} assumption 3 holds, assumption 2 of OLS holds as well.

Then, we formally check the normality of the residuals with a Jarque-Bera normality test (Jarque & Bera, 1987). The p-values of these checks are 0.059 for the Netherlands and 0.329 for Finland. Consequently, assuming a critical p-value of 0.10, the results of the Netherlands do not reject normality, whereas the test for Finland rejects normality. So, we assume homoscedasticity for the Netherlands and heteroscedasticity for Finland (robust standard errors). Figure 2.2, in the appendix, plots the distribution of the residuals in comparison to a normal distribution.

Finally, to fully supplement the approach of OLS, the last assumption of no perfect multicollinearity requires a formal analysis. Correlations between dependent variables might determine this problem (Stock & Watson, 2012). It is only important to know the absolute value of the correlation, so the direction (positive or negative) of the correlation does not matter. The more strength the correlation has, the more the two variables affect each other. In specific, a correlation of 0.6 or higher reflects a strong interaction.

We present a correlation overview, see table 2.3 in the appendix. The problem of multicollinearity seems to occur between lagged GDP and obesity rates, namely a correlation of 0.98. However, we resolve the problem if we take the first difference of the obesity rates. With this in mind, the primary reason for the multicollinearity might be the upward trend of both variables. In other words, the dataset does not violate the assumption of OLS.

5. Data

This section discusses the data and exists of two parts: data description and summary statistics.

5.1

Description

First, the Netherlands and Finland outline the sample of this research, which are two member countries of the OECD (OECD, 2016). This organization strives for policies to improve economic and social well-being of people around the world. Figure 10 and table 6, in section 5.2, show that there is a considerable increase in obesity rates in the period

1981-21 2014 (n=34). In specific, the growth rate of obesity in this period is 1.8 in the Netherlands, which is almost a doubling, and 1.5 in Finland. Then, we would like to know how this development affects the macro-economic direction of the two countries.

To begin with, the only consistent data points regarding obesity in OECD countries are accessible for the Netherlands and Finland (OECD Health Statistics, 2016). The

downside of this setting is that the size of the dataset is limited. If an individual’s BMI is equal to or greater than 30.0, it is considered obese, which is discussed in chapter 3. The obesity rates, measured over the total population and expressed in percentages, are self-reported. Hence, there is uncertainty about the validity of these values. It is possible to have an under- or overestimation, which refer to measurement errors.

Moving on, all approaches to measure GDP should result in the same value, which is discussed in chapter 2. We prefer the output approach to measure lagged GDP levels (t-1) in euros (€), by assuming current prices under Purchasing Power Parity (PPP) (OECD, 2016). Equation (1) describes the output approach, which is presented in chapter 2. Moreover, this approach determines current economic growth rates in percentages,

measured in percentages (OECD, 2016). In this paper, these growth rates represent lagged growth (t-1) as well. This is not the most ideal approach to measure lagged growth but is not a problem in this case, because of the limited size of the dataset.

𝑌 = 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐺𝑟𝑜𝑠𝑠 𝑂𝑢𝑡𝑝𝑢𝑡 − 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐼𝑛𝑝𝑢𝑡 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑠

Next, the paper uses the average gross enrolment ratios of primary education and secondary education (Worldbank Databank, 2016) as a proxy for education. The gross enrolment ratio divides the total enrolment within a country, for a specific level of education, by the total amount of the group within an official age range. The expression is in

percentages. The data points for secondary education in 1998 are missing for both

countries. To make the dataset free from missing observations, the gross enrolment ratio in this year is calculated by taking the average of 1997 and 1999. This is not an accurate representation of the observations in 1988, but it results in consistent data, which benefits the validity of the empirical framework. Finally, the dataset includes net FDI, see equation (22), expressed as a percentage of GDP (Worldbank Databank, 2016). This is a power to explain the openness of economy.

22

5.2

Summary Statistics

Variable Mean (NL) Mean (FIN) Std.Dev. (NL) Std.Dev. (FIN)

EG 2.1 2.1 1.9 3.2

OBS 8.3 11.2 2.7 3.4

GDP 392935.6 117347.4 168085.9 53559.1

FDI 12.9 2.2 18.4 3.3

AVG_ENROL 1334163 413339.6 51744.4 19413.8

Table 3: summary statistics Mean & Std. Dev.

NL: the Netherlands FIN: Finland

Variable Min. (NL) Min. (FIN) Max. (NL Max. (FIN)

EG -3.8 -8.3 5.1 6.3

OBS 4.7 6.7 13.3 18.3

GDP 173819 33682 652748 203338

FDI 0.8 -3.5 87.4 10.8

AVG_ENROL 1207850 381375 1415282 450402

Table 4: summary statistics Min. & Max.

Figure 10: Trend OBS for the Netherlands (NL) and Finland (FIN), source OECD Health Statistics (2016) 0 5 10 15 20

Sel

f

-

rep

o

rted

o

b

esi

ty

r

ates

Time span: 1981-2014

FIN

NL

23

OBS Start (1981) End (2014) Growth rate

Netherlands 4.7 13.3

𝟏𝟑. 𝟑 − 𝟒. 𝟕

𝟒. 𝟕

= 𝟏. 𝟖

Finland 7.2 18.3

𝟏𝟖. 𝟑 − 𝟕. 𝟐

𝟕. 𝟐

= 𝟏. 𝟓

Table 5: obesity growth rates

6. Results

The Netherlands(n=34) Finland(n=34)

R

2

0.44

0.45

Regress EG(t) on: Coef. Std. Error P-value Coef. Std. Error (Robust) P-value

OBS(t)

0.09

0.06

0.09*

0.09

0.08

0.23

EG(t-1)

0.35

0.16

0.04*

0.31

0.18

0.10*

GDP(t-1)

-0.20

0.09

0.04*

-0.80

0.48

0.11

FDI(t)

0.04

0.02

0.09*

0.45

0.17

0.01*

AVG_ENROL(t) -0.03

0.08

0.67

0.04

0.27

0.88

Constant

4.97

9.58

0.61

-2.51

13.25

0.85

Table 6: results OLS regressions significant at 10% *

Looking at table 6, the R2 reflects in which degree the regressors explain the total variation of the dependent variable. It seems that for both countries, our model predicts 45% of economic growth.

Then, table 6 shows that obesity rates are significantly different from zero in the Netherlands. In specific, a 1% increase in obesity in the Netherlands promotes economic growth by 0.09%. Section 5.2 indicates an average economic growth of 2.1% in both countries. Then, the coefficient of obesity reflects almost 1

24of the average growth, which

might be realistic if we perceive this size as the maximum. The positive sign of obesity is partly consistent with our expectations. Namely, chapter 3 expects that obesity might hurt or stimulate economic growth: a boost through increased food consumption and medical spending or a downfall because of decreased productivity levels, which follows from equation (2) and (19) in chapter 2. It seems that the aggregated consumption and health

24 expenditure dominate the negative consequences of obesity, in the 1981-2014 period. The data description concludes that obesity rates are self-reported. Then, another perspective might be that the negative economic effects of obesity are only at a micro level, which do not significantly affect the aggregated level due to an underestimation of the real obesity rates. Finally, the strong correlation between lagged GDP and obesity rates, indicated in the methodology, might influence the sign of the obesity rates.

𝑌 = 𝐶 + 𝐺 + 𝐼 + 𝑁𝑋

𝑔 =

𝐴 − 𝜌 −𝛿_𝜎

Although Finland has the same coefficient for obesity rates as the Netherlands, it does not significantly differ from zero. It might be that obesity tend to boost economic growth through increased food consumption and health expenditures but the total change in these observations seems marginal. It is probably be the case that food consumption is a small part of total household consumption in Finland and therefore has no significant effect on economic growth. A different option is that health has less impact on economic activity in comparison to the Netherlands.

Next, the initial economic growth significantly contributes to current growth in both countries. In specific, a 1% increase in lagged growth boosts current growth with 0.35% in the Netherlands and 0.30% in Finland. So, the initial growth does not deteriorate current growth. Therefore, the countries show no convergence, which is an assumption of the neoclassical approach. This supports our belief of endogenous growth, which states that there might be an indefinite growth in output levels. However, the concave production function of Solow (1956) seems to hold in the Netherlands. Table 5 indicates that a 1€ increase in lagged GDP significantly slows down current economic growth with 0.20%. In contrast, lagged GDP is not significantly different from zero in Finland. This means that there is no specific growth model to generalize economic progress.

Moving on, our findings show that there is a significant positive effect of FDI in both countries and therefore do not support the critics of Borensztein et al. (1998) on the

independent impact of FDI on economic growth. In Finland, a 1% increase in net FDI as a share of GDP boosts economic growth with 0.45%, which almost equals 1

5of average

economic growth.

We are curious about the degree of realism, because this is quite a heavy impact. Finally, education has no significant effect in both countries, whereas literature anticipates a positive effect. This dataset only includes primary education and secondary education, which are the most basic forms of education. The contribution of education on economic growth might be higher if we include tertiary education, which goes beyond the necessary types of education. So, the level of education might not be high enough to

25 indicate a significant positive effect on economic growth in Finland. In addition, the sign of the coefficient of education is negative for the Netherlands. I belief that this observation is a measurement error, this is not realistic.

7. Conclusion

This research examined the impact of obesity on economic growth for the Netherlands and Finland by using time series data and OLS regressions. It introduced the neoclassical model of Solow (1956) and the endogenous growth approach of Arora (2001). We preferred the endogenous growth model to measure permanent growth effects of obesity. Our empirical framework pointed out a significant positive effect of obesity in the Netherlands, whereas it has no effect in Finland. In specific, a 1% increase in obesity rates implies a 0.09% boost in economic growth rates in the Netherlands on the condition that we have 34 observations. It seems that the aggregated expenditures is the dominating channel for this result. These results are contradictory to our anticipations but are rational according to the literature. The scarcity of data regarding obesity over the whole population restricts us to focus on the Netherlands and Finland, which are two relatively small industrialized countries. This is a limitation of this paper, because it results in a low degree of generalizability. In addition, the economic impacts of obesity on economic growth in developing and transitional countries are not investigated. In fact, the economic consequences of obesity can differ per country. Hence, our first suggestion for further research is to include more countries with different characteristics to improve the generalizability and reliability of the empirical framework. Second, we recommend to break down the effect of obesity on the economy to clarify the isolated micro-economic consequences. It is an option to regress wages, total factor productivity or labor productivity on obesity rates. Economic growth is an aggregated approach with various mechanisms, there is too much going on at the same time to fully demonstrate it. The

R

2confirmed this statement and indicated a 45% explanation of the variation in economic growth, while we have included 5 explanatory variables. Finally, the relationship between GDP levels and obesity needs more clarification. This paper focusses solely on the impact of obesity on growth rates. We perceived a correlation of 0.98 between lagged GDP levels and obesity rates. Although it does not indicate multicollinearity, it might refer to a reverse causality.

8. References

Arora, S. (2001). Health, Human Productivity, and Long-Term Economic Growth. The

Journal of Economic History, 61, 699-749.

12-26 17.

Baum, C.L., Ruhm, C.J. (2009). Age, socioeconomic status and obesity growth. Journal of

Health Economics, 28, 635-648.

Becker, G. S. (1994). Human capital revisited. In Human Capital: A Theoretical and Empirical Analysis with Special Reference to Education (3rd Edition) (pp. 15-28). The University of Chicago Press.

Benhabib, J., & Spiegel, M. M. (1994). The role of human capital in economic development evidence from aggregate cross-country data. Journal of Monetary economics, 34(2), 143-173.

Bhattacharya, J., Bundorf, M.K. (2009). The incidence of healthcare costs of obesity. Journal

of Health Economics, 28, 649-658.

Bloom, D. E., Canning, D., & Sevilla, J. (2004). The effect of health on economic growth: a production function approach. World development, 32(1), 1-13.

Borensztein, E., De Gregorio, J., & Lee, J. W. (1998). How does foreign direct investment affect economic growth? Journal of international Economics, 45(1), 115-135.

Cawley, J., Meyerhoefer, C., Biener, A., Hammer, M., & Wintfeld, N. (2015). Savings in medical expenditures associated with reductions in body mass index among US adults with obesity, by diabetes status. Pharmacoeconomics, 33(7), 707-722.

Chou, S., Grossman, M., & Saffer. H. (2004). An economic analysis of adult obesity: results from the Behavioral Risk Factor Surveillance System. Journal of Health Economics, 23, 565-587.

Cole, M. A., & Neumayer, E. (2006). The impact of poor health on total factor productivity. The Journal of Development Studies, 42(6), 918-938.

Correia, I. H. (1996). Should capital income be taxed in the steady state? Journal of Public

Economics, 60(1), 147-151.

De Mello, L. R. (1999). Foreign direct investment-led growth: evidence from time series and panel data. Oxford economic papers, 51(1), 133-151.

Engelbrecht, H. J. (1997). International R&D spillovers, human capital and productivity in OECD economies: An empirical investigation. European Economic Review, 41(8), 1479-1488.

England, R. W. (1998). Measurement of social well-being: alternative to gross domestic product. Ecological Economics, 25(1), 89-103.

27 Finkelstein, E. A., daCosta DiBonaventura, M., Burgess, S. M., & Hale, B. C. (2010). The costs of obesity in the workplace. Journal of Occupational and Environmental

Medicine, 52(10), 971-976.

Fogel, W.R. (1994). Economic Growth, Population Theory, and Psychology: The Bearing of Long-Term Processes on the Making of Economic Policy. The American Economic Review,

84, 369-395.

Gates, D. M., Succop, P., Brehm, B. J., Gillespie, G. L., & Sommers, B. D. (2008). Obesity and presenteeism: the impact of body mass index on workplace productivity. Journal of

Occupational and Environmental Medicine, 50(1), 39-45.

Grier, K. B., & Tullock, G. (1989). An empirical analysis of cross-national economic growth, 1951–1980. Journal of monetary economics, 24(2), 259-276.

Guerrini, L. (2006). The Solow–Swan model with a bounded population growth rate. Journal

of Mathematical Economics, 42(1), 14-21.

Gyimah-Brempong, K., & Wilson, M. (2004). Health human capital and economic growth in Sub-Saharan African and OECD countries. The Quarterly Review of Economics and

Finance, 44(2), 296-320.

Huxley, R. (2014). Metabolic mediators of the effects of body-mass index, overweight, and obesity on coronary heart disease and stroke: a pooled analysis of 97 prospective cohorts with 1· 8 million participants. The Lancet, 383(9921), 970-983

Jarque, C. M., & Bera, A. K. (1987). A test for normality of observations and

regressionresiduals. International Statistical Review/Revue Internationale de Statistique, 163-172.

Jiménez-Rodríguez, R., & Sánchez, M. (2005). Oil price shocks and real GDP growth: empirical evidence for some OECD countries. Applied economics, 37(2), 201-228 Jones, C. (1998). introduction To Economic Growth 2nd Edition.

Komlos, J., Smith, P. K., & Bogin, B. (2004). Obesity and the rate of time preference: is there a connection? Journal of biosocial science, 36(02), 209-219.

Landefeld, S. J., Seskin, E. P., & Fraumeni, B. M. (2008). Taking the pulse of the economy: Measuring GDP. The Journal of Economic Perspectives, 22(2), 193-193.

Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A contribution to the empirics of economic growth. The quarterly journal of economics, 107(2), 407-437.

Mankiw. (2010). Macroeconomics. United States of America: Palgrave Macmillan.

Mokdad, A. H., Ford, E. S., Bowman, B. A., Dietz, W. H., Vinicor, F., Bales, V. S., & Marks, J. S. (2003). Prevalence of obesity, diabetes, and obesity-related health risk factors,

28 2001. Jama, 289(1), 76-79.

National Obesity Observatory. (2010). The economic burden of obesity. [NHS]

Nelson, R. R. (1973). Recent exercises in growth accounting: new understanding or dead end? The American Economic Review, 63(3), 462-468.

Philipson, T.J., Posner, R.A. (2008). Is obesity Epidemic a Public Health Problem? Journal

of Economic Literature, 46:4, 974-982.

Parker, J. (2012). Macroeconomic Theory. Oregon.

Prescott, E. C. (1998). Lawrence R. Klein lecture 1997: Needed: A theory of total factor productivity. International economic review, 525-551.

Romer, P. (1989). Endogenous technological change (No. w3210). National Bureau of Economic Research.

Sloan, R. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings? (Digest summary). Accounting review, 71(3), 289-315.

Smith, P. K., Bogin, B., & Bishai, D. (2005). Are time preference and body mass index associated? Evidence from the National Longitudinal Survey of Youth. Economics & Human

Biology, 3(2), 259-270.

Solow, R. M. (1956). A contribution to the theory of economic growth. The quarterly journal

of economics, 65-94.

Stock, J. H. & Watson, M. W. (2012) Introduction to Econometrics. Third edition.

International edition. Boston, Pearson.

Websites:

ACIAR (2017). Retrieved from:

http://aciar.gov.au/files/mn-158/s12_3-trends-total-factor-prod.html BBC (2016, March 16). Sugar tax: How will it work? Retrieved from:

http://www.bbc.com/news/health-35824071 Built Lean (November, 2016). Retrieved from:

http://www.builtlean.com/2013/07/17/bmi-chart/#fn-17539-4

EUR Macro (2016). Retrieved from:

http://www.eurmacro.eu/tutor/c11.html

OECD (2016). Retrieved from: http://stats.oecd.org/

OECD GDP (2017). Retrieved from:

29 OECD Health Statistics (2016). Retrieved from:

http://stats.oecd.org/index.aspx?DataSetCode=HEALTH_STAT

Office for National Statistics (2016). Retrieved from: http://webarchive.nationalarchives.gov.uk

University of Utah Health Care (2016). Retrieved from:

http://healthcare.utah.edu/bariatricsurgery/images/bmi-key.jpg

World Health Organization (2016). Retrieved from:

http://www.who.int/mediacentre/factsheets/fs311/en/

9. Appendix

9.1

Solow model

 INADA conditions (Guerrini, 2006, p.15)



𝑓(0) = 0

 Continuously differentiable

 Strictly increasing:

𝑓’(𝑘) > 0



Second order condition is negative:

𝑓’’(𝑘) < 0



lim

_𝑘→0

𝑓′(𝑘) = ∞



lim

𝑘→∞

𝑓′(𝑘) = 0

9.2

OLS assumptions

30

The Netherlands Finland

Figure 2.2: distribution residuals (normality)

Variable OBS(t) EG(t) EG(t-1) GDP(t-1) FDI(t) ENROL_AVG(t)

OBS(t)

1

(1)

EG(t)

-0.25

(-0.22)

1

(1)

EG(t-1)

-0.20

(-0.22)

0.54

(0.43)

1

(1)

GDP(t-1)

0.98

(0.98)

-0.32

(-0.27)

-0.25

(-0.19)

1

(1)

FDI(t)

0.63

(0.30)

0.02

(0.39)

-0.01

(0.16)

0.65

(0.33)

1

(1)

AVG_ENROL(t) 0.63

(-0.01)

-0.37

(0.10)

-0.38

(0.03)

0.66

(0.05)

0.44

(0.32)

1

(1)

Table 2.3: correlation table for the Netherlands (Finland)