Identifying Causal Relationships vs. Ruling Out All Other Possible Causes

Portrait of Artistotle (Source: Wikimedia Commons.)

I was in Washington last month to discuss my work on food prices, in which I look at whether food prices cause social unrest, at an event whose goal was to discuss the link between climate change and conflict.

As many readers of this blog know, disentangling causal relationships from mere correlations is the goal of modern science, social or otherwise, and though it is easy to test whether two variables x and y are correlated, it is much more difficult to determine whether x causes y.

So while it is easy to test whether increases in the level of food prices are correlated with episodes of social unrest, it is much more difficult to determine whether food prices cause social unrest.

In my work, I try to do so by conditioning food prices on natural disasters. To make a long story short, if you believe that natural disasters only affect social unrest through food prices, this ensures that if there is a relationship between food prices and social unrest, that relationship is cleaned out of whatever variation which is not purely due to the relationship flowing from food prices to social unrest. In other words, this ensures that the estimated relationship between the two variables is causal. This technique is known as instrumental variables estimation.

Identifying Causal Relationships vs. Ruling Out All Other Causes

As with almost any other discussion of a social-scientific issue nowadays, the issue of causality came up during one of the discussions we had at that event in Washington. It was at that point that someone implied that it did not make sense to talk of causality by bringing up the following analogy:

Imagine a house on stilts. Mold has been accumulating on the stilts for a number of years. One day, a hurricane comes along, and the house collapses. Can we really say that the hurricane caused the house to collapse when the mold has been eating away at the stilts for a long time?

I am paraphrasing, but the idea was that it was impossible to talk of causality given that most things had both proximate causes and distal causes — the hurricane and the mold, respectively, in the example above.

When the house-on-stilts analogy was brought up, many shrank and responded that establishing whether x causes y is not the same as saying “x is the only cause of y.”

Likewise, establishing that food prices cause social unrest is not the same thing as saying that food prices are the only cause of social unrest. Of course there are other factors.

When food prices rise sharply, people might be rioting in Lagos, but they are unlikely to riot in Milwaukee, which tells us that there must be something else going on.

But the fact that something else might going on does not mean that we cannot ask whether x causes y. When we talk about identifying causal relationships, we are not talking about ruling out all other possible causes. Sometimes a cigar is just a cigar, and when we ask whether x causes y, what we are really asking is “Does x cause y?,” and not “Is x the only cause of y?” The former is answerable; the latter is akin to asking about the unmoved mover of Aristotle’s Metaphysics.


  1. andreas beger

    When people bring up alternative causes of an event, isn’t it often motivated by some implicit question about substantive effects?

    For example, with the hurricane and mold story, and assuming that somehow we found that both mold and hurricanes seem to have a causal relationship with house collapse, how important is it if hurricanes (or mold for all I know) explain house collapses better than some other factor?

  2. Pingback: 10 Monday PM Reads | The Big Picture
  3. Chad Shipmaker

    Good to see you at the event!

    I understand the broader point you are making about causality – but I’m not sure our colleague in DC who used the metaphor of the mold and hurricane would agree with the way you’ve represented his argument. I thought that rather than saying it was “impossible to talk of causality” he was actually just being more nuanced about the causality itself. His work convincingly showed (and predicted) that food prices had created the conditions for conflict (mold) but noted that for conflict to occur there also needed to be a “trigger” (the hurricane).

    Going in a completely different direction – I personally would not agree with the notion that natural disasters only affect social unrest through food prices. You also need to account for the other impacts that disasters frequently have on populations – beyond loss of life there is often a deterioration of social cohesion due to a breakdown of community ties, disruption of order, family structure, loss of housing, opportunity, education, etc. Look at the components of ‘social cohesion’ and almost every one is likely to be adversely impacted by a natural disaster.

    Not sure what that means for your research but let me know next time you’re up in DC if you want to chat!

  4. Marc F. Bellemare

    Thanks for your comment, Chad. Maybe our colleague intended it that way, but that is certainly not what I (and a few otherss I talked to afterwards) understood.

    Your point about the exclusion restriction is well taken and on a long enough time scale, I completely agree with you. That is why I use monthly data — in such a short period of time, it becomes less likely that those other channels affect social unrest, whereas markets respond relatively quickly.

  5. Chad Shipmaker

    Understood on the first point.

    On the second, we’ll have to agree to disagree. A Typhoon, Tsunami, or Earthquake often impact social cohesion quickly… not just months. Try applying your assumption to Haiti for example. Did the earthquake only impact social cohesion in Haiti through food prices? Not sure you can ignore the massive loss of life? etc, etc, etc.

    I also wonder about the sensitivity of using the FAO food price index for a measure of volatility generally. Most traders and farmers make decisions by following individual commodities daily (or hourly, or less), while the FAO index aggregates a basket of commodities for monthly data. In my experience volatility is a problem day to day, not month to month, so wouldn’t the FAO index hide a lot of the actual volatility?

    I would be interested to see the outcome of your research using the daily closing prices of major grains. Have you tried that? (curious, not trying to make more work!)

  6. Marc F. Bellemare

    Actually, on the second point, I do not agree to disagree. None of the natural disasters you mention are among the ones I retain for analysis (I retain those that constitute shocks to the supply and demand of food, e.g., droughts, episodes of extreme temperature, floods, insect infestations, storms, volcanic eruptions, and wildfires).

    Moreover, you might recall from your reading of the paper that I use worldwide data. This means that the identification strategy goes something like (i) natural disaster in country A, (ii) shock to the supply and/or demand of food propagated through world food markets, and (iii) social unrest in country B as a consequence of an exogenous change in the price of food. Rarely (if ever) in the data will you have a natural disaster in country A followed by social unrest in country A. So the short time scale and the geographical dispersion of the causal chain aids the identification. Thanks, however, for highlighting that I need to do a better job of explaining that.