# Introduction to Econometrics : Learning Econometrics the Easy Way! – Part 1

Learning Econometrics can be a daunting task. The financial sector and its revolving industries assume that Econometrics is the dividing quality between someone who dabbles in economic theory and is a ”economist”.  Philosophies may differ but in the pursuit of practical solutions the business world sees econometrics as one of the defining aspects in the arsenal of an economist. After all you wouldn’t want a architect who couldn’t translate his art and ideas into practical models.

This series aims to simplify and break down the basic tenets of Econometrics to give you a solid foundation and bank of information to tackle the field with confidence.

We are assuming you understand the gist of what Econometrics is about. If you don’t you can read up on Econometrics and its uses and/or why it is studied for a sharper context. That will aid your understanding.

In the first article we will be covering a few notions:
1) Statistical Inference
2) The Research Format
3) Gathering Data
4) Introduction to the Econometric Model

There is also a video version if you prefer watching:

1 –    Statistical Inference

Statistical Inference is a powerful word which can be important to your understanding of the field. Many students may do econometrics calculations but have a very poor understanding of the ”whys” and the ”hows”. The word ”Statistical” is a reference to the compiled stats for your project. The word ”inference” is using the evidence to see its support (or lack thereof) the hypotheses. All in all statistical inference is the notion of learning something about the real world through analyzing your data.

1. Statistical inference usually involves estimating economic parameters (such as elasticities, a basic economic concept most of you economics students will be aware of) using econometrics methods.
2. Predicting economic outcomes such as the level of income business X may make over a 15 year period.
3. Testing hypothesis rooted in economics such as questioning the impact of using more marketing on a firms profit level.

2 –  The Research Format

Empirical economics research follows a template format and it is important to know this. If you are a higher-education economics student if you haven’t come across this yet, you will.

1. Question
2. Relationships & Variables
3. Constructing Models
4. Choosing Data Methods
5. Sample Data & Analysis
6. Estimates & Tests
7. Validity
8. Consequences/Analyzing/Impact/Evaluation

Confused? Yeah, me too. Those were just little pointers. Now lets explain all this ker-fluffle.
More detail will be dispensed in following articles, but for now lets wrap our head around this simple structure.
1. Every project will start with a question. Any topic that you choose to tackle will involve some sort of question that you are working against. This will be the basis of all economic research.

2. Economic theory essentially opens up a new perspective of looking at the situation in your question. In this way we can start to see : what variables would be involved in the relationship(s)? Around this a economic model can be built on the hypotheses of interest. Questions will naturally occur as research gains traction but it is wise to have a solid foundation to guide you.

3. With our economic model we can start to assume a econometric model. This will make more sense later, but your ideas must now take on a functional form. It would be wise to make assumptions about possible error term candidates and its inherent nature.

4. Sample data is now obtained (see next section for more detail) and a method of analysis of stats can be chosen. This will be based on our assumptions made at the start of the project and our assumptions.

5. Estimates of the unknown parameters need to be calculated. This can be done with a statistics program. Predictions are made and tests are performed against the hypotheses.

6. This helps us to see the validity of our assumptions. We use our econometrics models and diagnostics are performed. E.g. was the correct functional form used? Are the variables relevant?

7. Lastly the implications of the empirical results are now analyzed. Your data and the tests performed against it now undergo statistical inference. This involves understanding the relationship of the variables, how they react with each other, the relationship they possess,  the economic consequences and impact directly in line with your hypotheses.

3  –  Gathering Data

Understanding the difference between non-experimental and experimental data

It is important to grasp how data is gathered. Most economic data is non-experimental. This is because it is usually ”observed” versus being drawn from a controlled experiment. You could imagine a controlled experiment easily in a field like science where variables are clearly defined and the outcome is observed. This is rare in the social sciences; which makes it especially challenging to define economic parameters.

Non-Experimental Data

An example of non-experimental data is surveys. In these cases variables are not repeatable or fixed hence being ”non-experimental”. This includes but is not limited telephone surveys, mail surveys, face-to-face surveys and so on.

These non-experimental data types can be collected in various ways:

1-Time-Series Form Data

This is data collected over discrete interval of time. For example : the price of item X from 1880 – 2007; or the value of Item Y from the time period 2001 – 2014.

2- Cross-Section Form Data

This type of data is collected over sample units. This means units in a [particular] time period.  For example :  Income of a firm in the year 2001, or the level crime went down in 2006.

3- Panel-Data Form

This data follows individual micro-units over time. This is not to be confused with time-series form data. For example : the U.S. Department of Education has ongoing surveys tracking students from the 8th grade to their mid-twenties. Such data sources are useful because they are rich in studies related to economics of labor, household, health, education and more.  Usually collecting this data is to explore a deeper issue or to tackle a issue from beyond the surface level.

There is also two other basic denominations of data:

All economics students should be aware of this : but still I will briefly refresh your memory.
Micro : Data collected against individual units such as, individuals, households, or singular firms.
Macro: Data that is an aggregate (or total sum) : such as, individuals in a county/state/country, firms at a local/state/national level.

Another two basic denominations for you to memorize:

Data can either be flow or stock.
- Flow : Outcome measured over a certain period of time. Example : consumption of fruits in Q3 of 1997.
- Stock: Outcome measured at a certain period of time. Example : the amount of excess stock held by Adidas in warehouses in March 2001.

The final two:

The data can either be qualitative or quantitative.
-Quantitative: When data is expressed numbers or a function of numbers, such as real capita per income.
-Qualitative: Outcomes that are of an ”either-or” quality; or non-numerical. Example : A consumer did or did not purchase a certain good. Or whether an individual is either married or he isn’t.

4 –  Introduction to the Econometric model

Oh no. Time to introduce you to some of the nitty-gritty. Don’t worry we are keeping it as simple as possible. Make sure you get the gist of what is being talked about the rest will align itself eventually.

Economic theory does not claim to be accurate to the point where it can predict exact behavior patterns of individuals and firms. Economic relations are never exact. Rather the average systematic behavior of many individuals/firms is considered. When studying a data-set for sales of Nike shoes we can see the actual amount of shoes sold is the sum of the systematic part. The random unpredictable component (which economists call $\epsilon$) is considered a random error. With this data we can use statistical inference against the hypotheses created against Nike; such as ”How has ”X” marketing campaign affected net profit in Q4 of 2014”. The sales reference the systematic and accurate aspect of the equation. The random error ( $\epsilon$) references the unknown and unpredictable part of the equation. This will be present in any scenario at all times as everything can not be predicted. By taking [[facts]] and the [[unknown]] into account; general trends and relationships between variables and their strength can be discerned through econometrics. We may learn that the Marketing Campaign failed due to a large random error; where a natural disaster caused the warehouses holding the first wave of promotional products for the press got flooded.

Lastly it is important to understand that in every Econometric model there are certain assumptions that need to be made. Whether it is a demand or supply equation, or a production function : there is always a systematic part and a  unobservable random part as we have just discussed. The $\epsilon$ represents a ”noise” or ”grey area” which blurs our understanding of the relationships between variables and this assumption is always present.

The following two tabs change content below.