Loading...

An Empirical Analysis of the Influence of Gross Domestic Product on the Consumption of Animal Products

Bachelor Thesis 2013 32 Pages

Economics - Statistics and Methods

Excerpt

Table of Contents

1. Introduction

2. Conceptual Framework
2.1 Data and Variable Motivation
2.2 Hypotheses on the Relation of the Variables

3. Econometric Strategy - Model and Data

4. Results and Interpretation

5. Conclusion
5.1 Consequences of an Overall Consumption Increase in Animal Products
5.2 Outlook

6. Bibliography

7. Appendix

Diagrams and Tables

Table 3.1: Countries Used in the Estimation.

Table 4.1: STATA output for the fixed effects regression model.

Diagram 4.1: Comparison of Animal and Vegetal Product Consumption.

Diagram 4.2: Deflator for Animal Products for Selected Countries.

Diagram 5.1.1: GDP Deflator for Selected Countries.

1. Introduction

Food is a permanently present and pressing topic in society. From the time where people changed their habits from being gatherers and hunters towards domesticating animals and starting to grow their food instead of collecting it from woods. Through the industrial revolution that changed Europe in a way that it was able to satisfy its growing populations‘ needs for food and nutrition in a way it was never able before. Until today where the number of overweight people on earth equals the one of undernourished ones1. An argument for the importance of food safety and security even today.

But the change has not only happened in the availability of food but also in the consumption patterns, that is in the food basket of households. In recent months Europe became highly aware of the topic when traces of horse-meat were found in else wise labelled meat and other livestock containing products. Even in the products of one of the largest firms in the world, Nestlé, traces of horse-meat were found2. Such a wide public outrage and interest in these findings is easily understood. Given that animal products are on far higher demand in industrialised countries than in developing ones. The same findings for example in India might have produced far less of an interest. Taking into account that most people there are vegetarians3 and almost half (42%) of all vegetarians in India never eat fish, meat or eggs and thus live almost on a strict vegan diet.

But although red meat consumption in India is low, poultry meat consumption has increased enormously over the last few decades. Its meat production is now a major commercial sector, compared to the backyard approach that was dominant just a few decades ago. Poultry is also the fastest growing segment of the agricultural sector4.

The same trend of increased demand for animal products can be observed in many developing countries. And though their intake is far less than in wealthy industrialised countries, a pattern is visible. The food trends are changing and the question must be raised what the influencing factors are. Given the differences between rich and comparably poorer countries, the question is if a causal relation between the wealth of a country and its consumption of animal products can be found. And if predictions of where the consumption is headed can be derived. A very pressing topic considering the billions of people still living in emerging markets. A relation of the changes in economic wealth and the food preferences could give an outlook of which problems and opportunities for the markets and governments lie ahead.

Therefore this paper is concerned with the influence of a nation‘s and person‘s wealth on her consumption preferences. It will relate those two points by means of an empirical analysis and show with the help of an econometric model if an influence of the former on the latter can be derived.

At the end the results will be presented and given the outcome, relevant present and future implications for society, politics and the markets will be discussed.

2. Conceptual Framework

2.1 Data and Variable Motivation

In order to compare countries’ well-beings with each other, the measure of the Gross Domestic Product (GDP) has many advantages. It measures the (dollar) value of all the goods and services that an economy produces during a specified period, such as a year. GDP thus gives quite an objective view of a country‘s wealth. And more specifically, GDP per capita can show the average person‘s wealth compared to a person in another country. The per capita value thus allows comparisons between countries with different population numbers. And since it has been used for many years, data for many countries and over longer time periods is readily available.

To differentiate consumption patterns and preferences, one can use different approaches. Be it calories, saturated sugars against non-saturated sugars or food items in particular.

The overall food intake however, should not change significantly once a person can satisfy their basic needs for calories and nutrients. What should change though, is the combination of products in a person‘s food basket. But even on the way towards achieving nutrition and food safety and security (in developing countries like Asia), the basket is likely to change in its composition due to substitution effects. This means that if a household has more disposable income they might substitute some basic food items like rice with more expensive foods such as meat. In order to find a sensible summary for the problem at hand, some results of studies conducted should be taken into account.

Globally, meat consumption has quadrupled in the last century from around 10 kg/ person/year in the 19th century to 41.2 kg/person/year in 2005. In industrialised countries 82.1 kg/person/year meat are consumed today. Developing countries show a rise in the increase of meat consumed by 5% annually. Though industrialised countries have a per head consumption of meat that is still four to five times higher than in developing countries, there is a clear trend towards an increased demand for meat. In total numbers, Chinas population already consumes 31% of the total meat today, though a large percentage can be attributed directly to its vast number of people. Other animal products such as egg consumption has tripled and milk consumption has increased by half, in developing countries over the last decades5. In China the per-head consumption of animal proteins between 1980 and 2002 was 140%, in Latin America 32%. The increased demand for animal products in developing countries over the last decades has led to an increase in meat production by a factor of 3.5, especially poultry, from 1980 to 2007, milk production almost tripled and egg production increased by a factor of 56.

Emerging economies are expected to continue to grow, along with increased purchasing power and a change in dietary preferences: predictions say that due to the aforementioned substitution effect a large share of primary foods such as cereals, grain, roots or pulses will be replaced by secondary and processed commodities such as animal products, oils, fats, sugar and sweets7.

Especially developing countries seem to have developed an increased demand for food items such as meat, milk, eggs and others, all of which are animal products. Thus either made of/from or produced by animals. And although in industrialised countries the consumption of those is already very high, an increase has still been present over the past decades. In order to include the changes in consumption for as many items as possible, all animal products consumed (animal products) shall be summed up and shown as a single observed variable y it in this paper.

With animal products as a dependent variable it is also interesting to see the change of the opposite commodities - vegetal products - in relation to the observed variable.

Another way to measure wealth in a society, though less accurate, could be health. Industrialised countries are usually associated with a high life expectancy. With increased access to the medical sector and better and cleaner living conditions, stillborn children should also decrease. Thus, the variables life expectancy and infant mortality seem adequate to cover these changes in living conditions.

Food prices are always a relevant topic, especially in recent years when food price speculations made the headlines in newspapers. One could imagine that food prices have an influence on the composition of a households food basket. But also the overall price level in an economy could have a substantial influence on the consumption patterns of the population. If for example rent is less expensive, a share of money is now free to be invested in other items, such as food. In order to measure this effect the overall and food consumer price index (CPI) are to be included.

All these values of course have to be evaluated on a per person basis, just like the GDP above, in order to be comparable across a variety of countries included in the model.

2.2 Hypotheses on the Relation of the Variables

Consumption preferences towards or against animal products should be influenced by the variables chosen in the section prior. But so far no hypothesis on their expected relation was discussed.

As seen in the sections before, the consumption of animal products is highest in industrialised countries. In Germany the GDP in 2008 was almost 10 times higher with 33824.795 compared to a developing country like Indonesia with 3670.2401. The GDP is a good measurement of wealth and therefore more likely to be higher in industrialised countries. It can thus be argued that developing countries as they move towards becoming industrialised ones, change also their eating habits in the direction of industrialised countries, and therefore more animal product demand. This shift from developing to industrialised can be shown by an increase in GDP. Having an increased income to spend the consumption choices can start to increase, to widen up. This leads to an ability to choose from a variety of products, including more expensive, previously not affordable, food items. Schlatzer for example agues that the richer a society in developing countries becomes, the more they try to adopt western lifestyles - which include an animal fat heavy diet8. An increased demand can either lead to more imports or an investment in a local industrial sector, making it more efficient. Like a change from household, self sustaining meat production to industrialised, mechanised processes. This in consequence results in cheaper food items which are then more likely to be consumed even more often. Overall, an increase in GDP should lead to a direct increase in animal products consumed.

Prices for food items should only have a small influence on the consumption of animal products. Food baskets measured by the CPI are less flexible than real food baskets of households. If for example pig meat is more expensive it might be substituted with more poultry. Nonetheless, lower food prices free up money which in return can be used to buy more animal products. If both items become less expensive the consumption of one or both would increase. Thus, there should be a small negative effect of consumer prices for food on animal products - lower CPI for food then leads to a higher demand for animal products.

The same holds true for the overall consumer price index. If other items are available for less, like rent, more money can be spent on other commodities such as food items. A negative influence of the overall CPI should be observed on animal products.

Higher life expectancy and decreasing infant mortality rates are associated with healthier lifestyles and proper access to medical institutions and aid. They are a measure of how healthy a household or country‘s population is able to live. People with access to doctors and education on food safety and security should be more likely to consume more animal products since it is a relatively fast way to satisfy most of the nutrient and calorie needs. On the other hand animal fats are the source for illnesses such as coronary heart disease or obesity9. An educated person that decides to live a long life and has the choice of food items, might tend to decrease their animal product intake considerably. However, a decreased infant mortality rate should have a positive influence on animal product consumption. As should life expectancy.

3. Econometric Strategy - Model and Data

In order to make sense of the data available and derive relevant results, the right model has to be chosen.

Having repeated measurements or observations per individual, that is several countries across several time periods the observations are not independent. The repetition however can be used to attain better estimates for the parameters. Pooling the observations and using OLS gives biased estimates. Thus, a convergence of cross-sectional and time-series analysis is used, called a panel data analysis. Three more or less independent approaches can be characterised. Independently pooled panels, random effect models (RE) and fixed effects (FE) - or first differenced models (FD).

While independently pooled usually refers to a population, the other two are concerned with samples to draw conclusions from. The main difference between RE and FE/FD models is concerned with the error terms. Since RE models do not allow an arbitrary correlation between the unexplained error a i, also called unobserved effect, and x itj (thus random error) when FE/FD models do, the latter are more widely used and thought to be a better way to prove ceteris paribus effects10. This is especially relevant for this analysis. Since countries are being observed over time one can imagine there are constant influences such as religion, culture, geographical differences and other that do not change over time. They are not included in the model since measuring them proves quite complicated for several reasons. Nonetheless, they would be expected to have an influence on the control variables such as eating habits and health. By assuming their constant nature a fixed effects model by time-demeaning, as shown below, takes care of these variables. Just as the means for a i equate zero and are „averaged out“, this happens to those fixed influences.

A general equation for the panel data model in use is

illustration not visible in this excerpt

The estimation procedure for FE removes the unobserved effect a i by time-demeaning the equation as well as the intercept. Thus the general time-demeaned equation for each i and several x i is

illustration not visible in this excerpt

However, two important assumptions have to be considered in order to use the fixed

effects model. The first being that variables that are constant over time i.e gender, ethnicity cannot be included in the equation. Due to the time-demeaning of the data those would cancel each other out in the process - a wanted effect in the case of time constant influences discussed above. Secondly, one assumes time independent effects for every entity that is possibly correlated with the regressors. This „independency of time“ can be viewed as a variable that has no influence on the observed variable. Otherwise one would have to include time as a influential factor in the equation. In this case it is crucial to be sure that „time“ does have no impact on the consumption of animal products. Generally time is not thought of as causing change, but more of a way to take into account changes due to unmeasured causes. This can be applied to the problem at hand just as well. There is no argument supporting that a change in time would influence the consumption patterns of a society, especially looking at a time span of just a few years.

Another basic but important assumption for the fixed effects model analysis to be valid is that the errors u it are homoskedastic and serially uncorrelated across t. Homoskedasticity assumes the standard deviations of the error terms to be constant and that they do not depend on the x values - unlike the unexplained error term a i.

For t an annual measurement was chosen for the years 2000 through 2008, thus 9 observations for each group i. This is on one side sensible in order to use recent trends and data for the model. But is also a necessity for the availability of data for the used variables across countries is the highest. 9 observations is sufficient to increase the data set to a point where meaningful conclusions can be drawn from the model.

For the estimation procedure the statistical program STATA in its version STATA/SE 12.0 for Mac OS is being used11.

The data in this paper is taken from The Organisation for Economic Co-Operation and Development (OECD) and The Food and Agricultural Organization of the United Nations (FAO). Their data is freely accessible and they maintain some of the most thorough databanks across more than a decade for most of the world‘s countries12.

The dependent variable „animal products“ as well as „vegetal products“ are taken from the food balance sheets (FBS) of the FAO.

[...]


1 Löwenstein 2011, p.86.

2 http://online.wsj.com/article/SB10001424127887323495104578313583735359450.html# - 20.02.2013

3 http://www.fao.org/WAIRDOCS/LEAD/X6170E/x6170e09.htm- 20.02.2013

4 FAO 2003b, Chapter 2.3, no page numbers.

5 Schlatzer 2010, p.34f.

6 Schlater 2010, p.34, 36.

7 FAO 2003a, p.175f and Karapinar 2010, p.323f.

8 Schlatzer 2010, p.40.

9 Löwenstein 2011, p.82.

10 A complete list of assumptions for the different models as well as their derivations can be found in Wooldridge (2006). The theory in this section draws all information from Wooldridge (2006), Chapter 14 (p. 485-509).

11 All important outputs and commands used can be found in the appendix.

12 http://www.oecd.org/statistics/ and http://faostat.fao.org - 05.01.2013 10

Details

Pages
32
Year
2013
ISBN (eBook)
9783656952534
ISBN (Book)
9783656952541
File size
1 MB
Language
English
Catalog Number
v298756
Institution / College
Martin Luther University – Lehrstuhl für Ökonometrie und empirische Wirtschaftsforschung
Grade
2,0
Tags
empirical analysis influence gross domestic product consumption animal products

Author

Share

Previous

Title: An Empirical Analysis of the Influence of Gross Domestic Product on the Consumption of Animal Products