« A short history of the Canadian one per cent | Main | Short Run vs Long Run order of moves between monetary and fiscal authorities »


Feed You can follow this conversation by subscribing to the comment feed for this post.

I love love *love* that baseball bat correlation.

I would think the reason that immigrants struggle on the job market is directly tied to the fact that many jobs have a "Canadian citizens and permanent residents only" requirement.

Even once an immigrant has obtained PR, the fact they were (legally!) discriminated against while they were waiting to qualify for PR probably has a long term effect on outcomes (less Canadian experience, prolonged unemployment etc.).

I certainly agree with your deeper methodological points here though.

Evan - I'm confused. Let's think about who is in Canada and is not a Canadian citizen or permanent resident. The major categories are: students, temporary foreign workers, people on tourist visa, refugees and illegal immigrants.

The temporary foreign workers are already working, and students can work under certain conditions. I can see a case for allowing refugees to work, but is it really a good idea to allow employers to hire tourists and illegal immigrants?

Frances - The case I chiefly had in mind was that of spouses of students (I will readily admit that there possibly aren't too many of these). I would guess that spouses of temporary workers are often in a similar situation.

Now, spouses of students are generally allowed to work. But, depending on their profession, many of the jobs that they might be looking for are restricted to residents and citizens only. It is also my experience that some companies place a resident and citizen condition in their job ads even when not legally required to.

As for tourists, there are some visas such as the working holiday program visas that allow ``tourists'' to work. People on these visas face the same issues described above. These working holiday program visas can now be extended (without limit until the age of 30 (?), from memory) so that some people will be working on these visas in Canada for many years.

I have to think about this. You start by saying economics isn't about predictions and you end by saying it is, sometimes. There is a useful idea and distinction in here somewhere.

I think you also have the last two sentences all wrong. Econometrics make insights useful, econometrics itself is just a tool and it connects theory analysis intuition beliefs etc to policy, not data to policy. Data is just a tool as well. The two things that need to be connected and tested are beliefs and action. The question always and everywhere is How do we make useful choices today when we don't and can't know some important components of tomorrow.

If economists are not in predicting businesses, then what businesses are they in? Bullshitting?

Economics is a serious subject for serious purposes of making important decisions. It is all about forecasting.


This post brings to mind something I wrote a while back:

How economic policy analysis is done, and why it's not the same as forecasting

Assuming you think economics is a science, science is about observing the real world, constructing hypotheses that explain interactions, and then testing these hypotheses, especially focusing on falsifiability. If the hypothesis holds up, it is considered potentially valid. Your grad student appears to be attempting to do real science. Why shouldn't he compare his immigrant population to the non-immigrant one and see if employment outcomes can be explained by characteristics other than their immmigrant status? Science is inherently predictive. If economics is not predictive, then it certainly is not a science.

Mary: "Why shouldn't he compare his immigrant population to the non-immigrant one and see if employment outcomes can be explained by characteristics other than their immmigrant status? "

This is my point. It is sensible to use immigrant status to explain employment outcomes. Immigrant status is a perfectly sensible explanatory variable. It is not sensible to see if immigrant status can be explained by employment outcomes - unless you can find some kind of theory to explain why employment outcomes in 2014 would cause people to immigrate to Canada in 2004 (this sounds like the worst Dr Who plot line ever).

JJ: What Stephen said.

Dan: "I have to think about this. You start by saying economics isn't about predictions and you end by saying it is, sometimes. There is a useful idea and distinction in here somewhere."

Yeah, I know. The pizza chef didn't have time to do the gourmet version with homemade dough and caramelized onions, so all you've got is pita topped with the leftover spaghetti sauce that was in the fridge and some cheap mozzarella cheese and then baked in the oven for 15 minutes.

But it's still free pizza.

This is interesting, have never come across reverse regression before. I can't quite grok from the paper linked exactly what advantages reverse regression has on standard Blinder-Oaxaca decomp though.

The question of forecasting (not prediction) is more of a scientific one than an economics issue, and it's also a question of having proper intellectual discourse around the boundaries of any given forecast. We make reasonably good forecasts of demographic trends and their impact on healthcare expenditures, and there's no reason why quality methodologies can't be transferred to other areas too. This is particularly important in areas where experimental research is usually impossible.

Re the limitations of "data mining" this is absolutely true. Big data and data mining are currently the most hubristic topic in popular science, ignoring the fact that vast quantities of raw data are extremely difficult to manage. Austin Frakt penned a good series of posts about the excess enthusiasm for Big Data.


Did you ever watch that show Criminal Minds? What drove me crazy about that show was the way that the computer genius woman was able to effortlessly merge multiple data sets with nary a case of someone being entered as Nick in one data base, Nicholas in a second data base, P. Nicholas in a third, and Phillip in a fourth.

Frances: Yes, that was exactly the show I had in mind (also 24). In a nerdy way, that was all I could focus on, so I hated the show. But that show wasn't much of an exaggeration from some of the more straight-faced claims being made about Big Data in the real world.

Economics done properly is in the business of making conditional predictions – if we change this, that will most likely happen type of thing. Unconditional forecasting is not good economics, but unconditionally forecasting is where you make money. The ability to unconditionally forecast asset prices is the claim of active money managers. Of course, no one knows how to forecast asset prices unconditionally, but claiming that you can, can make you a bundle.

Unfortunately, most of your students if they stay in economics, but not in academics, will be in the business of producing unconditional forecasts. Just look at the asset price forecasts by bank economists, including the chief economists of all banks in Canada. Of course these people are really in the marketing business as opposed to economics, but all the public see is an “economist” is telling them something, whether it's the dollar or housing, is going to go up or down. Even the IMF get in on this nonsense with the recent claim in the news that Canadian housing is in for a “soft landing”. So, while good economics is about conditional forecasting, most professional economists don't, they unconditionally forecast, because that's where the money is – even if they're no better than a coin toss.

Finally, you say “For example, economists can predict that stock market bubbles will form and then burst, but they cannot predict when crashes will occur.” Actually, no they can't. Economists have a really hard time even defined a bubble, let only develop a theory of them. Prices went up and then they went down observations is no more of a theory about asset prices than your student's attempt to make immigrant status the dependent variable.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad