Continuing Conversation... Daron Acemoglu on Inequality, Institutions, and Piketty

EconTalk Extra
by Amy Willis
PRINT
Daron Acemoglu on Inequality, ... Becky Liddicoat Yamarik on Pal...

Russ Roberts welcomed MIT's Daron Acemoglu back to EconTalk this week, in an episode focused on the interplay between institutions and inequality.

Now it's your turn...Use the prompts below and share your reactions in the comments. If you use the prompts elsewhere (in a class, at the dinner table or a cocktail party), we'd love to hear about that, too.

inequality2.jpg

Check Your Knowledge:

1. What does Acemoglu mean when he speaks of the "endogenous evolution" of institutions, and how does he claim that Thomas Piketty takes insufficient account of this in his analysis?

2. Toward the end of the episode, Roberts suggests that Karl Marx (and Piketty) are both guilty of "mean reversion," or "Galton's Fallacy." What does this mean, and how does this help illustrate the importance Roberts and Acemoglu grant to institutions?


Going Deeper:

3. Acemoglu says there are four things wrong with Piketty's r > g identity, which Piketty dubs the central contradiction of capitalism. What are these four problems, and to what extent do they refute Piketty's claims?

4. Both Roberts and Acemoglu agree that focusing on the top 1% in discussions of inequality is faulty. Why? How do Acemoglu's views on inequality differ from Roberts's? With whom are you in more agreement, and why?


Extra Credit:

5. Roberts and Acemoglu have slightly different views on the contributions of American labor unions in the 20th century. In your opinion, have labor unions been a net positive or negative for workers? What evidence can you provide in support of your answer? Do you think the answer would be different depending on the time period under consideration?

Comments and Sharing



TWITTER: Follow Russ Roberts @EconTalker


COMMENTS (7 to date)
SaveyourSelf writes:

1. “Endogenous”, according to Webster-Merriam dictionary, means “caused by factors inside the organism or system.” I believe Dr. Acemoglu means there are factors that affect economically important outcomes [like growth] that come from inside a culture, over time, rather than from outside a culture, like in the case of a colony. [The term also seems to have a statistical meaning relating to how a variable is treated within a formula which I do not understand.]

“Evolve” means develop. And even though institutions can theoretically change their nature over time, “…colonial institutions, once set up, have tended to persist” (Institutions, Human Capital and Development page 2).

Dr. Acemoglu takes considerable pains to lay down an argument that Institutions are more-strongly associated with future economic outcomes than any other variable [like labor or capital] in models of growth. In “Institutions, Human Capital and Development” page 6, Acemoglu et al write,

…the impact of institutions on long-run development remains qualitatively and quantitatively robust to whether human capital is included in the regression (and treated endogenously) or historical determinants of education are directly controlled for. This evidence provides support for the view that institutions are the fundamental cause of long-run development, working not only through physical capital and TFP [Total Factor Productivity] but also through human capital.
He is arguing that labor, physical capital, and whatever TFP is, may look like they correlate with predicted growth, but their relationship to growth is primarily derived from their relationship to underlying-Institutions. Remove the capital and labor from the model, and the Institutions alone produce results as robust or even more robust than when they are included. Page 26, “These regressions therefore suggest that, although human capital and institutions are positively correlated as one would expect when institutions impact economic development via all of the key proximate determinants, there is no causal impact of human capital on institutions.”

Dr. Acemoglu then argues that Piketty does not recognize the importance of Institutions on growth and is instead relying on weak substitutes of Institutional-influence in his models. In “The Rise and Fall of General Laws of Capitalism” Page 19, Acemoglu writes, “his [Piketty’s] approach and general laws ignore both institutions and the flexible and multifaceted nature of technology shaped by institutions.”

SaveyourSelf writes:

Since Institutions are so central to Dr. Acemoglu et al’s arguments, it is worth taking a moment to consider what he means by the term. Having read both his articles and listened to the podcast, I can honestly say I’m not sure what he means. The examples he gave in the podcast made me think his institutions were: specific point sources of coercion in a society or market.

But in “The Rise and Fall of General Laws of Capitalism” page 6 he says Institutions, “…affect markets, prices, and the evolution of technology.”

He also says there is a

path-dependent process of institutional change involving the rationalization of property rights, dismantling of monopolies, investment in infrastructure, and the creation of a legal framework for industrial development, including the patent system (Acemoglu and Robinson, 2012, Mokyr, 2012).
But I’m not sure if he means property rights, the patent system, etc., are institutions or the byproducts of institutions.

In “Institutions, Human Capital and Development” page 1 he says,

Arrow and Debreuís approach to general equilibrium presumes a set of very specific institutions that specify the initial ownership of assets in society, enforce private property rights over factors of production and shares that individuals hold in firms in the economy, uphold contract and prevent monopolization of markets.
But, again, I am not sure if these are examples of institutions or products coming out of institutions.

On page 2 he says Institutions can be “extractive” or “inclusive”.

In both papers he divides Institutions in to "political" and "economic".

Finally, Merriam-Webster dictionary defines Institutions as, “a significant practice, relationship, or organization in a society or culture.”

I just have to scratch my head. A practice, relationship, or organization is a very broad set of possibilities, and Dr. Acemoglu’s sentences and examples seem to allow for any or all of them. If that is the case, then I would level the same criticism of Acemoglu’s “Institutions” as I did at Dr. Solow’s “Capital”: How can a variable so broad and inclusive possibly lead to answers to specific questions about the economy? It is rather like saying that all colors are electromagnetic waves and then trying to predict the composition of a painting using the percent change in electromagnetic waves as the only variable.

On the other hand, if by Institutions he means, “specific point sources of coercion in a society or market,” he might be on to something. We know that coercion applied to enforce property rights, contracts, and Justice generally leads to desirable economic outcomes whereas coercion used to undermine property rights, violate contracts, or injure others generally leads to undesirable economic outcomes. So if Dr. Acemoglu is thinking of Institutions as causal agents for these different types of coercion, then that might be fruitful for explaining long term economic changes.
I guess competition and access to information would have to be added in as separate variables, since they don’t seem to fit under this definition of institution. Perhaps that is why Russ, in the podcast, brought up the benefits of competition to his house-cleaners.

SaveyourSelf writes:

2. “’regression to the mean’ refers to the tendency of a variable characteristic in a population to move away from the extreme values towards the average value” (http://www.fallacyfiles.org/regressf.html). Importantly, the change in value is random. The observed pattern of change can suggest a correlation where there is none, which is a type of “Post hoc fallacy.” ( http://en.wikipedia.org/wiki/Regression_fallacy )

Around 56:22 Russ Roberts said,

Marx was writing at a time when wages were stagnant; so we saw that he presumed that tomorrow was going to be like today, which is like yesterday; and therefore, there was going to be this perpetual stagnation. And he of course was writing this on the verge, on the cusp of the greatest leap in individual living standards that human history has ever seen.

Acemoglu said this was Galton’s Fallacy.

I think Dr. Roberts was commenting on the risk of trying to predict the future based on the present or recent past, whereas Acemoglu was generalizing about Piketty’s lack of effort to reduce bias while interpreting his own data. In both cases, they are discussing statistical errors. Understanding statistical errors is important when evaluating Piketty’s claims vs. Acemoglu’s claims because Acemoglu says he does more to try and decrease the influence of statistical bias than Piketty.

SaveyourSelf writes:

5. In your opinion, have labor unions been a net positive or negative for workers?

  • The long term expected outcomes from Monopolisitic-markets—when compared with Ideal-markets—are: lower quantity, lower quality, higher prices, and flat or negative trends in innovation. Unions reduce the competition in a labor market, creating monopoly power for the union members. The outcomes are always negative, in the long run, even for the union members.

What evidence can you provide in support of your answer?
  • GM and Chrysler were union shops. An American built car was doing well to reach 100,000 miles before needing replaced. At the same period in time, Toyota was a non-union shop. It was nothing for its cars to last over 300,000 miles before needing replaced. The prices for Toyota cars was somewhat higher [which is not in line with the monopoly trend prediction above] but that reflected true differences in quality as well as taxes and subsidies and fees that favored “domestic automakers” over foreign automakers. I think any comment on the “quantity” produced would be a counterfactual argument. It felt like trends in innovation favored the foreign firms who had no government protections against competition, but this is debatable. Number of patents filed might give a more objective idea but I suspect the patent office would have favored domestic car makers over foreign auto makers as well.
  • After GM and Chrysler went through bankruptcy, the unions were severely weakened and very rapidly the quality of American made cars improved. Undoubtedly other things were going on with the bankruptcy, so not all the improvements can be ascribed to weakening of the Unions, but the changes following the reduction of monopoly powers in the American car industry fit the Ideal-Market model I used to predict those changes over a decade before the bankruptcy finally occurred. So, in my mind at least, the effects from changes in Union monopoly power I am describing here are not subject to post-hoc fallacy.

Do you think the answer would be different depending on the time period under consideration?
  • No. There is no reason monopoly would produce different outcomes in different time periods. People have not changed much in…let see…homo sapien came about 200,000 years ago. That is at least how long monopoly outcomes have been negative.


There is an argument, which Acemoglu hints at accepting in the podcast, that sometimes the best solution to a single monopoly in a market--say on the supply side--is a monopoly on the other side of the market--say the demand side--to provide a kind of balance. Unfortunately this leads to an even worse situation called “Double Marginalization” which essentially means that 2 vertical monopolies are much worse for society than one. http://www.youtube.com/watch?v=2NElV61MUDE

Michael Byrnes writes:

With regard to #4, I think there may be some reason to focus on the top 1% with regard to inequality. Not solely, but to some extent.

Here is my argument.

On this podcast as well as the one with Piketty, Roberts made the point that if wealth is being concentrated at the top as a result of wealth created by those at the top (say, by making iPhones or other products consumers prize), then it is not a problem at all. I think he's correct that this, in and of itself, is not a concern (rather it is the opposite of a concern). And one need only look at all of the varied products affordable to us to see evidence of this.

However, Acemoglu raised what I think is a legitimate reason for concern about wealth concentration - even though he and Roberts are correct that the focus should not be exclusively in the top 1% and Roberts is correct that wealth concentration is benign to the extent it is driven by wealth creation. This was Acemoglu's "political" rationale for concern about inequality - that concentration of wealth may lead to (and/or result from) concentration (and misuse) of political power.

In this regard, one of the things I wonder about is the extent to which political power is used to "protect" concentrated wealth. Roberts has done several podcasts on the topic of bailouts (including his solo tour-de-force on "Gambling with other people's money").

I think - and I can't recall offhand whether he or any guests have touched on this issue - that actual bailouts are only the tip of the iceberg. To me, bailouts can be thought of as payouts of an insurance policy against financial losses - the public being the unwitting insurer in this case.

Insurance companies can afford to offer insurance because only a few of those who buy policies actually experience a covered loss. So the insurer can provide insurance to many, pay the claims of a few, and profit. And the policies themselves, most of which do not result in payments to the insured, provide value - else no one would buy them. (I am happy to pay for my term life policy, for example, even though I don't want my beneficiaries to ever collect on it!)

Anyway, my point is that to understand the benefits of insurance, one must look beyond the people who experience losses that are covered. Those people are like the tip of the iceberg - most insurance buyers don't collect on their policies, but nevertheless they receive real value in exchange for their premiums. In many cases they can make different life decisions, and perhaps live better lives as a result of being insured.

So it must be with bailouts. Those that we actually see and know about may just be the tip of the iceberg - we can't necessarily know from looking at them how many corporations and how much wealth we are (unknowingly and without compensation) insuring. But those who are protected can make different decisions, take different risks, and experience different outcomes as a result of that protection. If those risks "pay off", then there will never be a bailout, there won't be an obvious "hand of the government" in the resultant wealth, but that wealth would still be "tainted". And bailouts are just one way that concentrated wealth could be protected at our expense.

The other important point it is not necessarily enough to say, "Steve Jobs created the iPhone, thus he deserved to be wealthy." I mean, that's obviously true, but Jobs and Apple were enriched not just from their own productivity (creating and marketing iPhones), but also from restraining the productivity of others (winning patent battles with Samsung). There's no good reason to think Apple doesn't deserve to be one of the richest corporations in the world, but it's hard to know how much of its wealth is deserved vs. unearned. To the extent that it is the latter, that is worthy of concern.

SaveyourSelf writes:

@Michael Byrnes – Great post.

4. Russ did a very nice piece on this very question after the Piketty interview Afterthoughts on Piketty. In it, he says increases in the wealth at the top of the earnings distribution could reflect globalization, in which case such an increase is very desirable. The higher income, in that scenario, is a reflection of the increased standard-of-living a huge number of people are experiencing. He also recognizes that increases in the standard of living at the top of the income distribution could reflect coercive rent’s stolen from large numbers of people with the aid of Government, in which case the increase would reflect an actionable problem. So, for Roberts, the cause of the change matters. In either case though, he worries that the added revenue to a minority group will create incentives for politicians to align their actions to benefit that minority over the rest of the population, even in the absence of any direct malfeasance. In his words, “As the rich pay a bigger share of the tax burden, politicians are more likely to take care of them.” This problem, he reasons, would likely get worse if the taxes on the wealthy are made even more progressive.

Dr. Acemoglu weighs in on this question at minute 51 in this week’s podcast. He begins with the simple observation that South Africa was one of the least equitable societies in the world, but showed a decline in the concentration of wealth in the top 1% even as apartheid was expanding. Thus, he does not think it is a very good indicator of inequality. He also seconds the concern about government capture “if somehow money becomes more and more dominant in politics.” But ultimately he views inequality measurements useful only in so far as they trigger questions—or are proxies for—underlying political, social, and economic activities. He indicated that good political, social, and economic activities do exist but are the exception in human history. After all,

Nobody actually has an interest in a competitive labor market. Everybody would like to be a monopolist. Everybody would like to form coalitions with their friends and the people with whom they have some overlapping economic interest. And, with those coalitions, force...noncompetitive elements. (22:11)

It seems to me that both men are arriving at the same conclusions—albeit one from the microeconomics point of view; the other from a macroeconomics point of view. I agree with them, but approach from another direction. Questions of inequality are usually designed to engender covetous-feelings. Nothing good is ever built on a foundation of envy. They are baiting a crowd to riot. They are convincing the majority that it is ok to steal from a minority. They are laying the groundwork for the Holocaust. A more useful way to look at the most-wealthy, I think, is to ask: “what are they doing differently than others” and “what must they sacrifice to achieve those outcomes.” And if those questions reveal that they are stealing from others to make themselves wealthy, then the next logical question is “what, in their environment, allows them to get away with theft repeatedly” which, I believe, is the same question as "where do they derive their monopoly power" which is also the same as “what in their environment fails to meet the assumptions of a Free-Competitive-Informed-Just Market”.

john penfold writes:

The discussion and web posts are impressive. Yet it seems to me we are looking for the keys to our questions about income inequality, growth etc. only where the light shines, in the data. Data is extensive, important to use and is used impressively, but is a tiny piece of reality. Most reality doesn’t show up as data and that which does, shows up after the fact and in averages which have removed most detail that might be useful. Well that’s what economists do and must, to be recognized and to discern what we can know with some sense of confidence, but the best insights about how an economy works and where those institutions come from we got from Hayek and Smith who don’t use data or others like Mancur Olsen who use what they must to be taken seriously, but actually offer insights that do not depend on data. A lot of data is like a thermometer that tells us if the body is sick, but doesn't really tell us why. Balance of payments data can tell us something is wrong, aggregate income data that something has changed either up or down, but not what or why.

We have stressed over market imperfections, enduring pockets of poverty, periodic unemployment, and abuses for over a century throughout the west. In this country alone we have spent 20 trillion on poverty programs just since the 60s which has made folks worse off in all but the numbers themselves. We built gargantuan regulatory apparatus which we can agree is captured and probably does more harm than good because they tend to weaken market disciplines and the adaptation and learning markets give rise to. We’ve piled trillions into so called developing countries and the really successful ones have received the least or none at all. Does this data not tell us anything? When we look back across civilizations is there any reason to consider growth and equality as normal? We can count relatively prosperous and equal places on one hand. The default position is centralization by a tiny elite, stagnation and widespread poverty. So why do we look for ways to centralize, collect data from which most reality is excluded, and then use it to push resources around according to the lights of the elite of the day. I’d say reread “The Fatal Conceit” and start over.

Comments for this podcast episode have been closed
Return to top