August 17, 2009

Was Max Weber right… at all?

Posted in International development at 6:27 pm by Eamon Aghdasi


Max Weber, author of The Protestant Ethic and the Spirit of Capitalism, and inspiration to lazy economists everywhere

For a long time I’ve been fascinated by the idea that economic outcomes might somehow be related to cultural and ethical norms. Do societies which value work tend to be wealthier or develop faster? What about societies in which people more commonly believe in women’s freedom? What about those in which individuals look more favorably on democracy, or more critically at corruption and bribery?

Economists and social scientists have grappled with these questions for about a century. The topic goes as far back at least as Max Weber, who in defining the “Protestant ethic” argued that Protestant societies, by virtue of their religious and cultural history, were better equipped to succeed than Catholic societies. Soon after Weber, Werner Sombart published a book called The Jews and Modern Capitalism, in which he argued that Jews had qualities that allowed them to succeed in capitalistic environments.

More recently, economists have pointed at culture as a burden in some countries’ economic development. For better or worse, this cultural argument for development stagnancy has fallen flat on its face in recent years. The lone fact that Ireland and India were two common examples supporting the culture-as-economic-constraint argument now render this line of argument ridiculous. Amid India as one of today’s shining examples of emerging market performance, the phrase “Hindu rate of growth” is now used only to giggle at economists’ lack of foresight. Similarly, the once disappointing Ireland now finds itself with a higher per capita GDP level than the United Kingdom.

While many economists now focus on the role of institutions in development, few seem interested in revisiting things like culture or religion as explanations for development outcomes. My sense is that the cultural arguments have been shot down so consistently and so embarassingly over time that few academics want to risk being the next one to look stupid by bringing it up again. There’s also the unfortunate racial motivation for some of these studies in the past, which renders more modern versions of this research as suspicious.

Last year I took a look at data from the World Values Survey, a global effort to collect individuals’ thoughts and perceptions on a wide range of topics, and investigated whether or not we could find any interesting correlations with countries’ strength of the rule of law (as measured by the World Bank).

What’s the difference between this and previous research? Well, I hope a lot. On the theoretical side, I sought to answer the question of how public perceptions and values affected institutions (in this case, the rule of law) which in turn link to development outcomes. I find this a much more plausible causal argument than simply arguing something like “Protestants work harder because they believe in hard work”. But also, I don’t have any illusions of culture as an exogenous, immutable reality for countries. This is where a lot of this line of research delves into subtle racism, in my opinion, this notion that different nations have different qualities, and that’s the way they’ll always be. This makes a difference in the methodology and the numbers; it’s ridiculous to treat culture or norms as some sort of fixed effect in time series data, and makes more sense simply to examine correlations for any given year.

Here’s the methodology: I looked at four broad categories — individual identity, trust, political activity, and opinions about democracy — for 43 countries in the WVS 2005 List A dataset. Each category is made up of a handful of variables, all of which I converted to 0 or 1 dummies and collapsed into country-level weighted sums. I then ran regressions for each, with rule of law as the dependent variable, first as a univariate regression, second controlling for GDP levels, and third controlling for both GDP and the prevalence of primary education. (Here’s the list of variables.)

Here’s what the results show: There’s only one category of variable, political activity, for which we can reject the Null hypothesis of no correlation at the 5% level. (Here’s a table summarizing the statistical results.) The two variables in this category which were statistically significant were the number of people who had ever performed a political action, or who had recently performed a political action. Oddly, political action is positively correlated with a strong rule of law in a single-variable institution, but when we control for GDP and education, political action is negatively correlated. What exactly does this mean? If anything, to me it suggests that rich, educated societies with strong rule of law also tend to feature a high rate of political action among their citizens.

A few other variables show statistical significance without any control variables. These are the proportion of people answering positively to these questions or statements:

I see myself as an autonomous individual (POSITIVE)

Willingness to fight in war for the country (NEGATIVE)

People can be trusted (POSITIVE)

Do you trust people you know personally? (POSITIVE)

Do you trust people you meet for the first time? (POSITIVE)

It’s interesting how many of the trust variables work out; three of five prove statistically significant with single-variable regressions, and the extent of the association with the rule of law is bigger for those three than any other statistically significant variables. Though the relationships break down when you bring in education and GDP, this seems to support at least somewhat the sensible hypothesis that a strong rule of law is associated with a high level of trust within a society.

I also think it’s interesting that feelings of being an autonomous individual are positively correlated — most likely catching the phenomenon of a culture of autonomy in the Western societies where rule of law tends to be strong — though this says little about causality. I also think the negative association with willingness to fight in war is intriguing… though I don’t really know what that might mean.

Problems with this research: There’s a number of reasons why we can’t take these results too seriously, starting with the data itself. The World Value Survey people have over the years made a valiant effort (which I by the way think is underrated in its importance) to ask opinion-based questions of people around the world. Standardizing these across countries sounds easy, but it’s not. When presented with the hurdles of different languages, different educational levels, and different overall contexts — and while using the filter of translators and survey takers needed to surmount these hurdles — inevitably the results of questions like “How does democracy make you feel?” become difficult to compare across countries.

But besides the data, we’ve got major endogeneity problems here. First and foremost, it seems that GDP and education are highly correlated with many (or maybe all) of our WVS variables, giving us multicolinearity issues. For instance, it’s sensible to wonder if the level of education may have a direct effect on whether or not people “can be trusted” within a society, not just on the rule of law. But secondly, the paths of causality go both ways between many of the WVS variables and the rule of law.

Yes, I know this is problematic. It’s not surprising, thus, that we get such messy results when we introduce the GDP and education controls. Unfortunately, I see no practical way around this problem short of using instruments for each WVS variable that have a direct effect on rule of law but are uncorrelated with education and GDP. That makes little practical sense here, as far as I can see.

Which leads us back to the original single-variable regressions, and the same place that a lot of research into institutions finds itself. We can relatively easily figure out which things correlate with what — trust and the rule of law, for instance. Disentangling them from other phenomena, or arguing causality in one particular direction, of course, are still very sticky territory.

How do you get around these endogeneity problems (or any other problems)? Or, does it make more sense to live with what you’ve got, and make due with the correlations we see without controls?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: