1 An Introduction

Statements about the world can be divided in two broad categories: Normative and Positive statements. Normative statements are statements concerning values. Normative statements are non-verifiable, we are unable to appeal to data to establish their accuracy. These statements will often, but not always, be phrased with should, ought to etc. On the other hand, Positive statements are statements of facts. Positive statements are verifiable, we can appeal to data to verify their truthfulness.

There are 2 common pitfalls that students should avoid when trying to identify the nature of a statement. First, positive statements are often mistaken to mean true. However, the requirement that a statement be verifiable does not imply that it has to be true. Secondly, we can easily craft verifiable assertions about normative statements. For example: “most economists believe X ought to be” is verifiable. We could survey economists and see if it is true that most economists agree. As such, it is positive.

Economists can have personal values, ethical codes etc. However, these should be set aside when we analyze how the world is. The public often sees economists disagree in public, on TV or during public policy debates. Often, these disagreements arise because they do not share a common set of normative values. Despite these apparent disagreements, economists actually tend to agree on many topics. The topics on which economists have a broad agreement are usually topics for which economists would agree that the issue has largely been “resolved”. For example, on the benefit of trade or on the difference between cap-and-trade programs versus pollution (or Pigouvian) taxes (Both subjects you will learn about in this textbook).

Correlation and Causation

Now that we know that Economists tend to focus on positive statements, we need to pay attention to the distinction between causation and correlation:

• Causation: There is a causation between two variables when changes in one variable causes a change in another variable. For example, when the atmospheric pressure decreases, the boiling point for water decreases (Causation).
• Correlation: There is a correlation between two variables when the two variables move together but we can’t say that one variable caused the other variable to change. Correlation can be caused by chance or a confounding variable. For example, when altitude increases, the boiling point of water decreases. Generally, when altitude changes, the atmospheric pressure decreases and that will cause the boiling point of water to change. Nevertheless, we can’t say that altitude causes the boiling point of water to change. Suppose you fly in a pressurized airplane cabin at 35,000 foot. The atmospheric pressure in the cabin is equal to the atmospheric pressure at sea level. The boiling point at sea level and in the cabin would be the same: an increase in altitude does not cause the boiling point of water to increase.

Economics is often used to enlighten public policy design (how to make effective public policies). As such, causations are much more important to economists than correlations. If, for example, we want to design policies that increase income for the poorest households, we need to identify variables that causes earnings to increase (Education, for example). We can then design public policies that will affect the variables that will cause earnings to increase and attain our public policy goals. If however, we only know that 2 variables are correlated (let’s say a specific employability social program and earnings), it is possible that implementing public policies based on this correlation will not attain the desired social outcomes.

There are different types of correlation and causation:

• Reverse Causation: Reverse causation arise when we believe that variable x causes variable y to change but variable y actually causes variable x to change. The causal link is the reverse of what we believed.
• Positive Correlation: An increase in variable x is accompanied by an increase in variable y.
• Negative Correlation: An increase in variable x is accompanied by a decrease in variable y.

Finally, two variable that have a causal relationship will be correlated, but two variables that are correlated don’t necessarily have a causal relationship. This leads to the famous statement: “Correlation doesn’t imply causation.”

Experimental Design

Correlation is easy to find, but proving that we identified a causal effect is much harder. We want to identify the treatment effect: the effect of a treatment on those treated by it. In a perfect world, we would be able to observe the treated in 2 different universes. Both universes would be exactly the same, except that in one universe the treatment is applied and in the other universe the treatment is not applied. Obviously, we are unable to observe specific individuals in multiple universes. Instead. we measure the average treatment effect on the treated. Since we are unable to observe individuals in multiple different universes, we instead take 2 groups of individuals: the treatment and the control group. We apply the treatment to the treatment group and we use the control group as a way to measure what would happen to the treated if they had not been treated. This allows us to measure the average impact of the treatment on the treated. The most common way that scientists conduct such research is by using randomized controlled trials.

In order to measure the causal impact of the treatment properly, we need to make sure that our control group and experimental group are roughly the same. If the 2 groups are completely different, then the control group cannot serve as a valid proxy for what would happen to the experimental group if they didn’t receive the treatment. Usually, we can get relatively similar control and experimental group by randomizing the members of each group.  

What kind of experiments can we run in order to identify the causal link between 2 variables? Economists tend to use the following 3 types of experiments:

• Laboratory experiments: These experiments are similar to experiments in physics or health science. They are run in a laboratory with willing human participants (usually students). These are the easiest experiments to design. However, they suffer in their applicability to the real world: is an experiment run in a laboratory over a couple of hours really transferable to the real world? In health sciences or physics, probably. Maybe it’s a bit harder to believe in the case of economic studies where we measure the response of the experimental group to different incentives.
• Field experiments: Are experiments that are designed by scientists, but instead of being run in a laboratory setting, the experiments are run in the real world. We might decide to offer different incentives for people to get their kids vaccinated in different villages in a developing country. We can then measure the impact of the incentives on the vaccination rate in the different villages. There is little doubt that the results from these experiments are applicable to the real world. They are, however, much more costly than laboratory experiments to run.
• Natural experiments: Different levels of governments consistently change public policies. These policy changes can be thought as experiments that the government is running all the time. Obviously, the elected government doesn’t see these policy changes as an experiment but simply the correct policy to implement. However, economists can use these small policy changes as experiments to identify the causal impact of the policies by using clever experimental design. In Canada, Quebec is a gold mine for economists trying to identify the causal impact of policies. Quebec usually implements social policies at a different time than other jurisdictions. As such, it can often be used as a source of natural experiments (For example, Quebec started it’s subsidized day care program in 1997, while most other Canadian jurisdictions didn’t even contemplate such a program. We can use Canada as our control group and Quebec as our experimental group.).

Unfortunately, Economists are not always able to run controlled random trials (experiments) as easily as scientists in other hard sciences. There is a plethora of issues that plagues the design of controlled random trials in economics. We present 2 here: self-selection bias and moral issues:

• Self-selection bias: In many experiments that naturally arise in economics, the participants are able to self-selected into the treatment group or the control group. Suppose that you want to identify the impact of a training program on the probability of unemployed workers to find a new job. It is possible that only motivated unemployed workers apply to your program and that unmotivated workers do not apply to the program. When we measure the impact of the program on participants, we won’t really know if any significant increases in the probability of finding a job was due to the program or the unobserved variables like motivation.
• Moral issues: Moral issues can often prevent economists from running experiments. We would like to know the impact of poverty on various outcomes, for example. It is immoral to force people to live in poverty just to measure such impacts though. This often limits the scope of the type of research that economists can run in the laboratory or in the field.

Empirical economics (running well designed experiments to evaluate the causal impact of a policy) takes a central role in modern economic research. Historically, economists themselves didn’t believe that the experimental method could be applied in economics. However, things have changed a lot since:

In this course we will  focus mostly on theoretical problems. The reason is twofold:

• In order to understand the empirical research that is conducted in economics a strong foundation in both statistics and economic theory is required.
• Theoretical models can give us useful intuitions that can be useful in multiple fields (business, finance etc.)

Choices and Costs

We have limited resources and unlimited wants; we have to choose how to use our limited resources! In fact, an alternative definition of economics could be: “The science that studies how individuals/societies make choices.” Fundamental to the economic thought process is the idea that every time you make a choice, you give up something. There is a cost to every single one of your decisions. For example, you are currently reading this textbook, you gave something up in order to read it. Perhaps you could be playing your favorite computer game, or having a beer with friends at the local pub. The cost of your decisions is the opportunity cost of your decision:

“The opportunity cost is the best forgone alternative.”

This definition implies that the opportunity cost is not an amount of money. The opportunity cost is what you gave up in order to get something. This cost will vary for each reader. If you did not buy (or do) something, what would you do instead?

The opportunity set is a graphical representation of 3 concepts that have been introduced thus far: Choice, Limited Resources and Opportunity Cost. To make this graphical representation, we simply graph the bundles that are attainable by a consumer/firm. Everything that is achievable is in the opportunity set and everything that is unachievable is outside of the opportunity set. Let’s demonstrate this concept with two examples.

Production Possibility Frontiers are usually represented with a concave function. As more of a product is produced, the opportunity cost of that product increases: When a society is producing 0 units of wheat, the cost of producing wheat is relatively low. Workers with the best skills to produce wheat and the most favorable land to produce wheat can be shifted to wheat production. As more arable land and skilled workers are used to produce wheat, the country eventually runs out of workers with the relevant skills and/or arable land. The country may be able to produce more wheat, but the opportunity cost rises.