Hi everyone, so this is an introductory video to econometrics and econometric analysis, and what we're going to cover is what is econometrics, the basics of econometric data. going to touch on the goals of econometrics, we're going to cover some preliminary basics of economic data, and then finally touch on the simple linear regression. And so first things first, what is econometrics? In simple terms econometrics is the application of statistical methods in economics And so what that has to deal with a lot of the time is regression analysis And so if you know regression analysis, this will be pretty straightforward For those who don't know a regression is the relationship between one or several independent variables and The expected value of a corresponding dependent variable so that might sound a little bit intimidating at first but the concept is actually quite simple.
Basically what it is, is we have some dependent variable y, and then we have a whole bunch of independent variables that affect y through some function. So we have say these independent variables x1, x2, all the way to xn. and in some functional form they affect y. So, why is this interesting? Say a change in x causes a change in y, we want to know how how large that effect is, which direction that effect is in, and a bunch of other neat things.
Typically, we do this because it makes estimating relationships between economic variables very easy. It helps with testing economic theories and hypotheses. It's great for forecasting economic variables and evaluating or implementing government or business policy. It's very important in economics.
And so an example would be say you have wage and some function of say education and experience. So, this would be something a labor economist would do, and they would want to know how a change in education affects labor. lifetime wages or your certain wage, and how experienced, if it changes, how it affects wages well. They might also be interested with certain particular levels of experience in education.
So what we're trying to do is essentially find this functional form in econometrics. And so before we get into that, there's some preliminaries that are very important, such as the types of economic data. And there's three main types. Essentially, the first is called time series. And you might be used to dealing with this already.
It's the most basic of data sets. The second would be called cross-sectional. And you might have heard these before in an econometrics class or a statistics class.
The third is called panel data, which is the hardest to deal with, but we will get into that later. So time series is observations over time. It's the most basic.
An example would be, say, GDP measured for 10 years. one specific country. So Canada's GDP for 10 years.
Which is pretty basic. Cross-sectional would be, it's a little bit different, it would be across subjects instead of across time. An example staying on the topic of GDP would be annual GDP of a certain set of countries in a specific year.
And then panel data, the reason it's slightly more complicated is organizational issues, and also there's a lot more... data generally because it is observations over time and over individuals. So an example would be annual GDP over a certain set of countries between a set of years. So the annual GDP of this set of countries between 2000 and 2005. So now that we have covered the different types of economic data we can sort of get into the simple linear regression. So what economists do is they take one data, or combinations of these data, and they try to find this functional form.
And so the simple linear regression is one with one independent variable. As soon as you add another one, such in this case where we have two independent variables, it becomes what we call a multiple linear regression. which is slightly more complicated but it has basically the same principles. So we're going to start with the simple linear regression.
And so what we're basically trying to do is find a linear function of our dependent and our independent. So we're given all this data, say we have whole bunch of data and what this will look like in the end a lot of the time is what you might be familiar with is a line of best fit so it'll look something like that we're just trying to find a linear line that represents the relationship between Y and X and so we have basically the standard form for a linear line which would be Y is equal to mx plus b, which you should be very familiar with, where m is the slope and x would be your independent variable, y would be your dependent of course, and b would be the intercept. So if we extend this line so it touches the y axis, then this value here would be b, and if we took the slope of the line at any point The slope would be equal to m. So simple enough.
That's basically what we're looking for. The only difference is that with a regression we have a slightly, we have one extra variable enter the function and that variable is the error term. And so our function will generally look something like this. y is equal to some constant, which would be our b. We're going to call it beta 0 or beta 0. Plus our slope, which was our m here.
One independent variable. Then plus our error term, which we often call u. And...
In the population regression, we'll call it u. And so what this error term is, is this function isn't supposed to be a function of the line. It's more of a function of each individual data point.
And so we want to know for every single x, what is our data point equal to? And so we take this line, which is this part. We have, this is the function for the line, the linear line. And then we have this error term, and this error term is not a constant, it changes as x changes, but we just leave it as this u, because what it is, is it's the distance from this line to each individual point. So it's this distance, and it's also at this x, it would be this small distance here, and say we had a point here, so we can show a little bit better, the distance from this line to here.
would be our u. And so that way we can specifically say y is equal to this line. So say we have an x here which is xn.
So at xn we take the the function of the line and so it brings us to this y, but then we have a negative u and the negative u brings us down to here. And so what it is essentially saying is at xn, our point is here. And so this is the very basics of a simple linear regression, what it is. This is the general form of a simple linear regression and what it means. And so what economists are generally trying to find is a causal relationship, which is How does y change if x is changed and all other variables are held constant?
So if this were a multiple regression, we would hold x2 all the way to xn constant, but since it's linear, if we want to assume causality, we have to assume that x is the only variable that affects y. But that is definitely beyond the scope of this video. I hope you understand slightly better what a regression is, what econometrics is, some of the goals of econometrics, different types of data, and the basic form of a simple linear regression, and what it sort of means.
So we're going to go into simple linear regressions in the next model, or sorry, the next video, and we're going to touch on... the interpretation of each of these sort of coefficients and intercepts. And that would be all. Thank you for watching.
You can subscribe and I hope to see you soon.