econometrics

Demystifying dummy variables

pnrj statistics , , , ,

Nov 5, JDN 2458062

Continuing my series of blog posts on basic statistical concepts, today I’m going to talk about dummy variables. Dummy variables are quite simple, but for some reason a lot of people—even people with extensive statistical training—often have trouble understanding them. Perhaps people are simply overthinking matters, or making subtle errors that end up having large consequences.

A dummy variable (more formally a binary variable) is a variable that has only two states: “No”, usually represented 0, and “Yes”, usually represented 1. A dummy variable answers a single “Yes or no” question. They are most commonly used for categorical variables, answering questions like “Is the person’s race White?” and “Is the state California?”; but in fact almost any kind of data can be represented this way: We could represent income using a series of dummy variables like “Is your income greater than $50,000?” “Is your income greater than $51,000?” and so on. As long as the number of possible outcomes is finite—which, in practice, it always is—the data can be represented by some (possibly large) set of dummy variables. In fact, if your data set is large enough, representing numerical data with dummy variables can be a very good thing to do, as it allows you to account for nonlinear effects without assuming some specific functional form.
Most of the misunderstanding regarding dummy variables involves applying them in regressions and interpreting the results.
Probably the most common confusion is about what dummy variables to include. When you have a set of categories represented in your data (e.g. one for each US state), you want to include dummy variables for all but one of them. The most common mistake here is to try to include all of them, and end up with a regression that doesn’t make sense, or if you have a catchall category like “Other” (e.g. race is coded as “White/Black/Other”), leaving out that one and getting results with a nonsensical baseline.

You don’t have to leave one out if you only have one set of categories and you don’t include a constant in your regression; then the baseline will emerge automatically from the regression. But this is dangerous, as the interpretation of the coefficients is no longer quite so simple.

The thing to keep in mind is that a coefficient on a dummy variable is an effect of a change—so the coefficient on “White” is the effect of being White. In order to be an effect of a change, that change must be measured against some baseline. The dummy variable you exclude from the regression is the baseline—because the effect of changing to the baseline from the baseline is by definition zero.
Here’s a very simple example where all the regressions can be done by hand. Suppose you have a household with 1 human and 1 cat, and you want to know the effect of species on number of legs. (I mean, hopefully this is something you already know; but that makes it a good illustration.) In what follows, you can safely skip the matrix algebra; but I included it for any readers who want to see how these concepts play out mechanically in the math.
Your outcome variable Y is legs: The human has 2 and the cat has 4. We can write this as a matrix:

\[ Y = \begin{bmatrix} 2 \\ 4 \end{bmatrix} \]

reg_1

What dummy variables should we choose? There are actually several options.

 

The simplest option is to include both a human variable and a cat variable, and no constant. Let’s put the human variable first. Then our human subject has a value of X1 = [1 0] (“Yes” to human and “No” to cat) and our cat subject has a value of X2 = [0 1].

This is very nice in this case, as it makes our matrix of independent variables simply an identity matrix:

\[ X = \begin{bmatrix} 1 & 0 \\ 0 & 1 \end{bmatrix} \]

reg_2

This makes the calculations extremely nice, because transposing, multiplying, and inverting an identity matrix all just give us back an identity matrix. The standard OLS regression coefficient is B = (X’X)-1 X’Y, which in this case just becomes Y itself.

\[ B = (X’X)^{-1} X’Y = Y = \begin{bmatrix} 2 \\ 4 \end{bmatrix} \]

reg_3

Our coefficients are 2 and 4. How would we interpret this? Pretty much what you’d think: The effect of being human is having 2 legs, while the effect of being a cat is having 4 legs. This amounts to choosing a baseline of nothing—the effect is compared to a hypothetical entity with no legs at all. And indeed this is what will happen more generally if you do a regression with a dummy for each category and no constant: The baseline will be a hypothetical entity with an outcome of zero on whatever your outcome variable is.
So far, so good.

But what if we had additional variables to include? Say we have both cats and humans with black hair and brown hair (and no other colors). If we now include the variables human, cat, black hair, brown hair, we won’t get the results we expect—in fact, we’ll get no result at all. The regression is mathematically impossible, regardless of how large a sample we have.

This is why it’s much safer to choose one of the categories as a baseline, and include that as a constant. We could pick either one; we just need to be clear about which one we chose.

Say we take human as the baseline. Then our variables are constant and cat. The variable constant is just 1 for every single individual. The variable cat is 0 for humans and 1 for cats.

Now our independent variable matrix looks like this:

\[ X = \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} \]

reg_4
The matrix algebra isn’t quite so nice this time:

\[ X’X = \begin{bmatrix} 1 & 1 \\ 0 & 1 \end{bmatrix} \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} 2 & 1 \\ 1 & 1 \end{bmatrix} \]

\[ (X’X)^{-1} = \begin{bmatrix} 1 & -1 \\ -1 & 2 \end{bmatrix} \]

\[ X’Y = \begin{bmatrix} 1 & 1 \\ 0 & 1 \end{bmatrix} \begin{bmatrix} 2 \\ 4 \end{bmatrix} = \begin{bmatrix} 6 \\ 4 \end{bmatrix} \]

\[ B = (X’X)^{-1} X’Y = \begin{bmatrix} 1 & -1 \\ -1 & 2 \end{bmatrix} \begin{bmatrix} 6 \\ 4 \end{bmatrix} = \begin{bmatrix} 2 \\ 2 \end{bmatrix} \]

reg_5

Our coefficients are now 2 and 2. Now, how do we interpret that result? We took human as the baseline, so what we are saying here is that the default is to have 2 legs, and then the effect of being a cat is to get 2 extra legs.
That sounds a bit anthropocentric—most animals are quadripeds, after all—so let’s try taking cat as the baseline instead. Now our variables are constant and human, and our independent variable matrix looks like this:

\[ X = \begin{bmatrix} 1 & 1 \\ 1 & 0 \end{bmatrix} \]

\[ X’X = \begin{bmatrix} 1 & 1 \\ 1 & 0 \end{bmatrix} \begin{bmatrix} 1 & 1 \\ 1 & 0 \end{bmatrix} = \begin{bmatrix} 2 & 1 \\ 1 & 1 \end{bmatrix} \]

\[ (X’X)^{-1} = \begin{bmatrix} 1 & -1 \\ -1 & 2 \end{bmatrix} \]

\[ X’Y = \begin{bmatrix} 1 & 1 \\ 1 & 0 \end{bmatrix} \begin{bmatrix} 2 \\ 4 \end{bmatrix} = \begin{bmatrix} 6 \\ 2 \end{bmatrix} \]

\[ B = \begin{bmatrix} 1 & -1 \\ -1 & 2 \end{bmatrix} \begin{bmatrix} 6 \\ 2 \end{bmatrix} = \begin{bmatrix} 4 \\ -2 \end{bmatrix} \]

reg_6

Our coefficients are 4 and -2. This seems much more phylogenetically correct: The default number of legs is 4, and the effect of being human is to lose 2 legs.
All these regressions are really saying the same thing: Humans have 2 legs, cats have 4. And in this particular case, it’s simple and obvious. But once things start getting more complicated, people tend to make mistakes even on these very simple questions.

A common mistake would be to try to include a constant and both dummy variables: constant human cat. What happens if we try that? The matrix algebra gets particularly nasty, first of all:

\[ X = \begin{bmatrix} 1 & 1 & 0 \\ 1 & 0 & 1 \end{bmatrix} \]

\[ X’X = \begin{bmatrix} 1 & 1 \\ 1 & 0 \\ 0 & 1 \end{bmatrix} \begin{bmatrix} 1 & 1 & 0 \\ 1 & 0 & 1 \end{bmatrix} = \begin{bmatrix} 2 & 1 & 1 \\ 1 & 1 & 0 \\ 1 & 0 & 1 \end{bmatrix} \]

reg_7

Our covariance matrix X’X is now 3×3, first of all. That means we have more coefficients than we have data points. But we could throw in another human and another cat to fix that problem.

 

More importantly, the covariance matrix is not invertible. Rows 2 and 3 add up together to equal row 1, so we have a singular matrix.

If you tried to run this regression, you’d get an error message about “perfect multicollinearity”. What this really means is you haven’t chosen a valid baseline. Your baseline isn’t human and it isn’t cat; and since you included a constant, it isn’t a baseline of nothing either. It’s… unspecified.

You actually can choose whatever baseline you want for this regression, by setting the constant term to whatever number you want. Set a constant of 0 and your baseline is nothing: you’ll get back the coefficients 0, 2 and 4. Set a constant of 2 and your baseline is human: you’ll get 2, 0 and 2. Set a constant of 4 and your baseline is cat: you’ll get 4, -2, 0. You can even choose something weird like 3 (you’ll get 3, -1, 1) or 7 (you’ll get 7, -5, -3) or -4 (you’ll get -4, 6, 8). You don’t even have to choose integers; you could pick -0.9 or 3.14159. As long as the constant plus the coefficient on human add to 2 and the constant plus the coefficient on cat add to 4, you’ll get a valid regression.
Again, this example seems pretty simple. But it’s an easy trap to fall into if you don’t think carefully about what variables you are including. If you are looking at effects on income and you have dummy variables on race, gender, schooling (e.g. no high school, high school diploma, some college, Bachelor’s, master’s, PhD), and what state a person lives in, it would be very tempting to just throw all those variables into a regression and see what comes out. But nothing is going to come out, because you haven’t specified a baseline. Your baseline isn’t even some hypothetical person with $0 income (which already doesn’t sound like a great choice); it’s just not a coherent baseline at all.

Generally the best thing to do (for the most precise estimates) is to choose the most common category in each set as the baseline. So for the US a good choice would be to set the baseline as White, female, high school diploma, California. Another common strategy when looking at discrimination specifically is to make the most privileged category the baseline, so we’d instead have White, male, PhD, and… Maryland, it turns out. Then we expect all our coefficients to be negative: Your income is generally lower if you are not White, not male, have less than a PhD, or live outside Maryland.

This is also important if you are interested in interactions: For example, the effect on your income of being Black in California is probably not the same as the effect of being Black in Mississippi. Then you’ll want to include terms like Black and Mississippi, which for dummy variables is the same thing as taking the Black variable and multiplying by the Mississippi variable.

But now you need to be especially clear about what your baseline is: If being White in California is your baseline, then the coefficient on Black is the effect of being Black in California, while the coefficient on Mississippi is the effect of being in Mississippi if you are White. The coefficient on Black and Mississippi is the effect of being Black in Mississippi, over and above the sum of the effects of being Black and the effect of being in Mississippi. If we saw a positive coefficient there, it wouldn’t mean that it’s good to be Black in Mississippi; it would simply mean that it’s not as bad as we might expect if we just summed the downsides of being Black with the downsides of being in Mississippi. And if we saw a negative coefficient there, it would mean that being Black in Mississippi is even worse than you would expect just from summing up the effects of being Black with the effects of being in Mississippi.

As long as you choose your baseline carefully and stick to it, interpreting regressions with dummy variables isn’t very hard. But so many people forget this step that they get very confused by the end, looking at a term like Black female Mississippi and seeing a positive coefficient, and thinking that must mean that life is good for Black women in Mississippi, when really all it means is the small mercy that being a Black woman in Mississippi isn’t quite as bad as you might think if you just added up the effect of being Black, plus the effect of being a woman, plus the effect of being Black and a woman, plus the effect of living in Mississippi, plus the effect of being Black in Mississippi, plus the effect of being a woman in Mississippi.

 

Intellectual Property, revisited

pnrj core principles, development economics, public policy , , , , , , , , , , ,

Mar 12, JDN 2457825

A few weeks ago I wrote a post laying out the burden of proof for intellectual property, but didn’t have time to get into the empirical question of whether our existing intellectual property system can meet this burden of proof.

First of all, I want to make a very sharp distinction between three types of regulations that are all called “intellectual property”.

First there are trademarks, which I have absolutely no quarrel with. Avoiding fraud and ensuring transparency are fundamental functions without which markets would unravel, and without trademarks these things would be much harder to accomplish. Trademarks allow a company to establish a brand identity that others cannot usurp; they ensure that when you buy Coca-Cola (R) it is really in fact the beverage you expect and not some counterfeit knockoff. (And if counterfeit Coke sounds silly, note that counterfeit honey and maple syrup are actually a major problem.) Yes, there should be limits on how much you can trademark—no one wants to live in a world where you feel Love ™ and open Screen Doors ™—but in fact our courts are already fairly good about only allowing corporations to trademark newly-coined words and proper names for their products.

Next there are copyrights, which I believe are currently too strong and often abused, but I do think should exist in some form (or perhaps copylefts instead). Authors should have at least certain basic rights over how their work can be used and published. If nothing else, proper attribution should always be required, as without that plagiarism becomes intolerably easy. And steps should be taken to ensure that if any people profit from its sale, the author is among them. I publish this blog under a by-sa copyleft, which essentially means that you can share it with whomever you like and even adapt its content into your own work, so long as you properly attribute it to me and you do not attempt to claim ownership over it. For scientific content, I think only a copyleft of this sort makes sense—the era of for-profit journals with paywalls must end, as it is holding back our civilization. But for artistic content (and I mean art in the broadest sense, including books, music, movies, plays, and video games), stronger regulations might well make sense. The question is whether our current system is actually too strong, or is protecting the wrong people—often it seems to protect the corporations that sell the content rather than the artists who created it.

Finally there are patents. Unlike copyright which applies to a specific work of art, patent is meant to apply to the underlying concept of a technology. Copyright (or rather the by-sa copyleft) protects the text of this article; you can’t post it on your own blog and claim you wrote it. But if I were to patent it somehow (generally, verbal arguments cannot be patented, fortunately), you wouldn’t even be able to paraphrase it. The trademark on a Samsung ™ TV just means that if I make a TV I can’t say I am Samsung, because I’m not. You wouldn’t copyright a TV, but the analogous process would be if I were to copy every single detail of the television and try to sell that precise duplicate. But the patents on that TV mean that if I take it apart, study each component, find a way to build them all from my own raw materials, even make them better, and build a new TV out of them that looks different and performs better—I would still be infringing on intellectual property. Patents grant an extremely strong notion of property rights, one which actually undermines a lot of other, more basic concepts of property. It’s my TV, why can’t I take it apart and copy the components? Well, as long as the patent holds, it’s not entirely my TV. Property rights this strong—that allow a corporation to have its cake of selling the TV but eat it too by owning the rights to all its components—require a much stronger justification.

Trademark protects a name, which is unproblematic. Copyright protects a work, which carries risks but is still probably necessary in many cases. But patent protects an idea—and we should ask ourselves whether that is really something it makes sense to do.

In previous posts I’ve laid out some of the basic philosophical arguments for why patents do not seem to support innovation and may actually undermine it. But in this post I want to do something more direct and quantitative: Empirically, what is the actual effect of copyrights and patents on innovation? Can we find a way to quantify the costs and benefits to our society of different modes of intellectual property?

Economists quantify things all the time, so I briefly combed the literature to see what sort of empirical studies had been done on the economic impact of copyrights and patents.

Patents definitely create barriers to scientific collaboration: Scientific articles with ideas that don’t get patented are about 10-20% more likely to be cited than scientific articles with ideas that are patented. (I would have expected a larger effect, but that’s still not trivial.)

A 1995 study found that creased patent protections do seem to be positively associated with more trade.

A 2009 study of Great Britain published in AER found it “puzzling” that stronger patents actually seem to reduce the rate of innovation domestically, while having no effect on foreign innovation—yet this is exactly what I would have predicted. Foreign innovations should be largely unaffected by UK patents, but stricter patent laws in the UK make it harder for most actual innovators, only benefiting a handful of corporations that aren’t even particularly innovative.

This 1996 study did find a positive effect of stronger patent laws on economic growth, but it was quite small and only statistically significant when using instrumental variables that they couldn’t be bothered to define except in an appendix. When your result hinges on the use of instrumental variables that you haven’t even clearly defined in the paper, something is very fishy. My guess is that they p-hacked the instruments until they got the result they wanted.

This other 1996 study is a great example of why economists need to listen to psychologists. It found a negative correlation between foreign direct investment and—wait for it—the number of companies that answered “yes” to a survey question, “Does country X have intellectual property protection too weak to allow you to transfer your newest or most effective technology to a wholly-owned subsidiarythere?” Oh, wow, you found a correlation between foreign direct investment and a question directly asking about foreign direct investment.

his 2004 study found a nonlinear relationship whereby increased economic development affects intellectual property rights, rather than the other way around. But I find their theoretical model quite odd, and the scatter plot that lies at the core of their empirical argument reminds me of Rexthor, the Dog-Bearer. “This relationship appears to be non-linear,” they say when pointing at a scatter plot that looks mostly like nothing and maybe like a monotonic increase.

This 1997 study found a positive correlation between intellectual property strength, R&D spending, and economic growth. The effect is weak, but the study looks basically sound. (Though I must say I’d never heard anyone use the words “significant at the 24% level” before. Normally one would say “nonsignificant” for that variable methinks. It’s okay for it not to be significant in some of your regressions, you know.)

This 1992 paper found that intellectual property harms poor countries and may or may not benefit rich countries, but it uses a really weird idiosyncratic theoretical model to get there. Frankly if I see the word “theorem” anywhere in your empirical paper, I get suspicious. No, it is not a theorem that “For economies in steady state the South loses from tighter intellectual property rights.” It may be true, but it does not follow from the fundamental axioms of mathematics.

This law paper is excellent; it focuses on the fact that intellectual property is a unique arrangement and a significant deviation from conventional property rights. It tracks the rise of legal arguments that erroneously equate intellectual property with real property, and makes the vital point that fully internalizing the positive externalities of technology was never the goal, and would in fact be horrible should it come to pass. We would all have to pay most of our income in royalties to the Newton and Faraday estates. So, I highly recommend reading it. But it doesn’t contain any empirical results on the economic effects of intellectual property.

This is the best paper I was able to find showing empirical effects of different intellectual property regimes; I really have no complaints about its econometrics. But it was limited to post-Soviet economies shortly after the fall of the USSR, which were rather unique circumstances. (Indeed, by studying only those countries, you’d probably conclude that free markets are harmful, because the shock of transition was so great.)

This 1999 paper is also quite good; using a natural experiment from a sudden shift in Japanese patent policy, they found almost no difference in actual R&D. The natural experiment design makes this particularly credible, but it’s difficult to generalize since it only covered Japan specifically.

This study focused in particular on copyrights and the film industry, and found a nonlinear effect: While having no copyright protection at all was harmful to the film industry, making the copyright protections too strong had a strangling effect on new filmmakers entering the industry. This would suggest that the optimal amount of copyright is moderate, which sounds reasonable to me.

This 2009 study did a much more detailed comparison of different copyright regimes, and was unable to find a meaningful pattern amidst the noise. Indeed, they found that the only variable that consistently predicted the number of new works of art was population—more people means more art, and nothing else seemed to matter. If this is correct, it’s quite damning to copyright; it would suggest that people make art for reasons fundamentally orthogonal to copyright, and copyright does almost nothing useful. (And I must say, if you talk to most artists, that tends to be their opinion on the matter!)

This 1996 paper found that stronger patents had no benefits for poor countries, but benefited rich countries quite a large amount: Increased patent protection was estimated to add as much as 0.7% annual GDP growth over the whole period. That’s a lot; if this is really true, stronger patents are almost certainly worth it. But then it becomes difficult to explain why more precise studies haven’t found effects anywhere near that large.

This paper was pretty interesting; they found a fat-tailed distribution of patents, where most firms have none, many have one or a few, and a handful of firms have a huge number of patents. This is also consistent with the distribution of firm revenue and profit—and I’d be surprised if I didn’t find a strong correlation between all three. But this really doesn’t tell us whether patents are contributing to innovation.
This paper found that the harmonization of global patents in the Uruguay Round did lead to gains from trade for most countries, but also transferred about $4.5 billion to the US from the rest of the world. Of course, that’s really not that large an amount when we’re talking about global policy over several years.

What does all that mean? I don’t know. It’s a mess. There just don’t seem to be any really compelling empirical studies on the economic impact of copyrights and patents. The preponderance of the evidence, such as it is, would seem to suggest that copyrights provide a benefit as long as they aren’t too strong, while patents provide a benefit but it is quite small and likely offset by the rent-seeking of the corporations that own them. The few studies that found really large effects (like 0.7% annual GDP growth) don’t seem very credible to me; if the effect were really that large, it shouldn’t be so ambiguous. 0.7% per year over 25 years is a GDP 20% larger. Over 50 years, GDP would be 42% larger. We would be able to see that.

Does this ambiguity mean we should do nothing, and wait until the data is better? I don’t think so. Remember, the burden of proof for intellectual property should be high. It’s a fundamentally bizarre notion of property, one which runs against most of our standard concepts of real property; it restricts our rights in very basic ways, making literally the majority of our population into criminals. Such a draconian policy requires a very strong justification, but such a justification does not appear to be forthcoming. If it could be supported, that 0.7% GDP growth might be enough; but it doesn’t seem to be replicable. A free society does not criminalize activities just in case it might be beneficial to do so—it only criminalizes activities that have demonstrable harm. And the harm of copyright and patent infringement simply isn’t demonstrable enough to justify its criminalization.

We don’t have to remove them outright, but we should substantially weaken copyright and patent laws. They should be short-term, they should provide very basic protection, and they should never be owned by corporations, always by individuals (corporations should be able to license them—but not own them). If we then observe a substantial reduction in innovation and economic output, then we can put them back. But I think that what defenders of intellectual property fear most is that if we tried this, it wouldn’t be so bad—and then the “doom and gloom” justification they’ve been relying on all this time would fall apart.

Yes, but what about the next 5000 years?

pnrj book reviews, core principles , , , , , , , , , , , , , , , , ,

JDN 2456991 PST 1:34.

This week’s post will be a bit different: I have a book to review. It’s called Debt: The First 5000 Years, by David Graeber. The book is long (about 400 pages plus endnotes), but such a compelling read that the hours melt away. “The First 5000 Years” is an incredibly ambitious subtitle, but Graeber actually manages to live up to it quite well; he really does tell us a story that is more or less continuous from 3000 BC to the present.

So who is this David Graeber fellow, anyway? None will be surprised that he is a founding member of Occupy Wall Street—he was in fact the man who coined “We are the 99%”. (As I’ve studied inequality more, I’ve learned he made a mistake; it really should be “We are the 99.99%”.) I had expected him to be a historian, or an economist; but in fact he is an anthropologist. He is looking at debt and its surrounding institutions in terms of a cultural ethnography—he takes a step outside our own cultural assumptions and tries to see them as he might if he were encountering them in a foreign society. This is what gives the book its freshest parts; Graeber recognizes, as few others seem willing to, that our institutions are not the inevitable product of impersonal deterministic forces, but decisions made by human beings.

(On a related note, I was pleasantly surprised to see in one of my economics textbooks yesterday a neoclassical economist acknowledging that the best explanation we have for why Botswana is doing so well—low corruption, low poverty by African standards, high growth—really has to come down to good leadership and good policy. For once they couldn’t remove all human agency and mark it down to grand impersonal ‘market forces’. It’s odd how strong the pressure is to do that, though; I even feel it in myself: Saying that civil rights progressed so much because Martin Luther King was a great leader isn’t very scientific, is it? Well, if that’s what the evidence points to… why not? At what point did ‘scientific’ come to mean ‘human beings are helplessly at the mercy of grand impersonal forces’? Honestly, doesn’t the link between science and technology make matters quite the opposite?)

Graeber provides a new perspective on many things we take for granted: in the introduction there is one particularly compelling passage where he starts talking—with a fellow left-wing activist—about the damage that has been done to the Third World by IMF policy, and she immediately interjects: “But surely one has to pay one’s debts.” The rest of the book is essentially an elaboration on why we say that—and why it is absolutely untrue.

Graeber has also made me think quite a bit differently about Medieval society and in particular Medieval Islam; this was certainly the society in which the writings of Socrates were preserved and algebra was invented, so it couldn’t have been all bad. But in fact, assuming that Graeber’s account is accurate, Muslim societies in the 14th century actually had something approaching the idyllic fair and free market to which all neoclassicists aspire. They did so, however, by rejecting one of the core assumptions of neoclassical economics, and you can probably guess which one: the assumption that human beings are infinite identical psychopaths. Instead, merchants in Medieval Muslim society were held to high moral standards, and their livelihood was largely based upon the reputation they could maintain as upstanding good citizens. Theoretically they couldn’t even lend at interest, though in practice they had workarounds (like payment in installments that total slightly higher than the original price) that amounted to low rates of interest. They did not, however, have anything approaching the levels of interest that we have today in credit cards at 29% or (it still makes me shudder every time I think about it) payday loans at 400%. Paying on installments to a Muslim merchant would make you end up paying about a 2% to 4% rate of interest—which sounds to me almost exactly what it should be, maybe even a bit low because we’re not taking inflation into account. In any case, the moral standards of society kept people from getting too poor or too greedy, and as a result there was little need for enforcement by the state. In spite of myself I have to admit that may not have been possible without the theological enforcement provided by Islam.
Graeber also avoids one of the most common failings of anthropologists, the cultural relativism that makes them unwilling to criticize any cultural practice as immoral even when it obviously is (except usually making exceptions for modern Western capitalist imperialism). While at times I can see he was tempted to go that way, he generally avoids it; several times he goes out of his way to point out how women were sold into slavery in hunter-gatherer tribes and how that contributed to the institutions of chattel slavery that developed once Western powers invaded.

Anthropologists have another common failing that I don’t think he avoids as well, which is a primitivist bent in which anthropologists speak of ancient societies as idyllic and modern societies as horrific. That’s part of why I said ‘if Graber’s account is accurate,’ because I’m honestly not sure it is. I’ll need to look more into the history of Medieval Islam to be sure. Graeber spends a great deal of time talking about how our current monetary system is fundamentally based on threats of violence—but I can tell you that I have honestly never been threatened with violence over money in my entire life. Not by the state, not by individuals, not by corporations. I haven’t even been mugged—and that’s the sort of the thing the state exists to prevent. (Not that I’ve never been threatened with violence—but so far it’s always been either something personal, or, more often, bigotry against LGBT people.) If violence is the foundation of our monetary system, then it’s hiding itself extraordinarily well. Granted, the violence probably pops up more if you’re near the very bottom, but I think I speak for most of the American middle class when I say that I’ve been through a lot of financial troubles, but none of them have involved any guns pointed at my head. And you can’t counter this by saying that we theoretically have laws on the books that allow you to be arrested for financial insolvency—because that’s always been true, in fact it’s less true now than any other point in history, and Graeber himself freely admits this. The important question is how many people actually get violence enforced upon them, and at least within the United States that number seems to be quite small.

Graeber describes the true story of the emergence of money historically, as the result of military conquest—a way to pay soldiers and buy supplies when in an occupied territory where nobody trusts you. He demolishes the (always fishy) argument that money emerged as a way of mediating a barter system: If I catch fish and he makes shoes and I want some shoes but he doesn’t want fish right now, why not just make a deal to pay later? This is of course exactly what they did. Indeed Graeber uses the intentionally provocative word communism to describe the way that resources are typically distributed within families and small villages—because it basically is “from each according to his ability, to each according to his need”. (I would probably use the less-charged word “community”, but I have to admit that those come from the same Latin root.) He also describes something I’ve tried to explain many times to neoclassical economists to no avail: There is equally a communism of the rich, a solidarity of deal-making and collusion that undermines the competitive market that is supposed to keep the rich in check. Graeber points out that wine, women and feasting have been common parts of deals between villages throughout history—and yet are still common parts of top-level business deals in modern capitalism. Even as we claim to be atomistic rational agents we still fall back on the community norms that guided our ancestors.

Another one of my favorite lines in the book is on this very subject: “Why, if I took a free-market economic theorist out to an expensive dinner, would that economist feel somewhat diminished—uncomfortably in my debt—until he had been able to return the favor? Why, if he were feeling competitive with me, would he be inclined to take me someplace even more expensive?” That doesn’t make any sense at all under the theory of neoclassical rational agents (an infinite identical psychopath would just enjoy the dinner—free dinner!—and might never speak to you again), but it makes perfect sense under the cultural norms of community in which gifts form bonds and generosity is a measure of moral character. I also got thinking about how introducing money directly into such exchanges can change them dramatically: For instance, suppose I took my professor out to a nice dinner with drinks in order to thank him for writing me recommendation letters. This seems entirely appropriate, right? But now suppose I just paid him $30 for writing the letters. All the sudden it seems downright corrupt. But the dinner check said $30 on it! My bank account debit is the same! He might go out and buy a dinner with it! What’s the difference? I think the difference is that the dinner forms a relationship that ties the two of us together as individuals, while the cash creates a market transaction between two interchangeable economic agents. By giving my professor cash I would effectively be saying that we are infinite identical psychopaths.

While Graeber doesn’t get into it, a similar argument also applies to gift-giving on holidays and birthdays. There seriously is—I kid you not—a neoclassical economist who argues that Christmas is economically inefficient and should be abolished in favor of cash transfers. He wrote a book about it. He literally does not understand the concept of gift-giving as a way of sharing experiences and solidifying relationships. This man must be such a joy to have around! I can imagine it now: “Will you play catch with me, Daddy?” “Daddy has to work, but don’t worry dear, I hired a minor league catcher to play with you. Won’t that be much more efficient?”

This sort of thing is what makes Debt such a compelling read, and Graeber does make some good points and presents a wealth of historical information. So now it’s time to talk about what’s wrong with the book, the things Graeber gets wrong.

First of all, he’s clearly quite ignorant about the state-of-the-art in economics, and I’m not even talking about the sort of cutting-edge cognitive economics experiments I want to be doing. (When I read what Molly Crockett has been working on lately in the neuroscience of moral judgments, I began to wonder if I should apply to University College London after all.)

No, I mean Graeber is ignorant of really basic stuff, like the nature of government debt—almost nothing of what I said in that post is controversial among serious economists; the equations certainly aren’t, though some of the interpretation and application might be. (One particularly likely sticking point called “Ricardian equivalence” is something I hope to get into in a future post. You already know the refrain: Ricardian equivalence only happens if you live in a world of infinite identical psychopaths.) Graeber has internalized the Republican talking points about how this is money our grandchildren will owe to China; it’s nothing of the sort, and most of it we “owe” to ourselves. In a particularly baffling passage Graeber talks about how there are no protections for creditors of the US government, when creditors of the US government have literally never suffered a single late payment in the last 200 years. There are literally no creditors in the world who are more protected from default—and only a few others that reach the same level, such as creditors to the Bank of England.

In an equally-bizarre aside he also says in one endnote that “mainstream economists” favor the use of the gold standard and are suspicious of fiat money; exactly the opposite is the case. Mainstream economists—even the neoclassicists with whom I have my quarrels—are in almost total agreement that a fiat monetary system managed by a central bank is the only way to have a stable money supply. The gold standard is the pet project of a bunch of cranks and quacks like Peter Schiff. Like most quacks, the are quite vocal; but they are by no means supported by academic research or respected by top policymakers. (I suppose the latter could change if enough Tea Party Republicans get into office, but so far even that hasn’t happened and Janet Yellen continues to manage our fiat money supply.) In fact, it’s basically a consensus among economists that the gold standard caused the Great Depression—that in addition to some triggering event (my money is on Minsky-style debt deflation—and so is Krugman’s), the inability of the money supply to adjust was the reason why the world economy remained in such terrible shape for such a long period. The gold standard has not been a mainstream position among economists since roughly the mid-1980s—before I was born.

He makes this really bizarre argument about how because Korea, Japan, Taiwan, and West Germany are major holders of US Treasury bonds and became so under US occupation—which is indisputably true—that means that their development was really just some kind of smokescreen to sell more Treasury bonds. First of all, we’ve never had trouble selling Treasury bonds; people are literally accepting negative interest rates in order to have them right now. More importantly, Korea, Japan, Taiwan, and West Germany—those exact four countries, in that order—are the greatest economic success stories in the history of the human race. West Germany was rebuilt literally from rubble to become once again a world power. The Asian Tigers were even more impressive, raised from the most abject Third World poverty to full First World high-tech economy status in a few generations. If this is what happens when you buy Treasury bonds, we should all buy as many Treasury bonds as we possibly can. And while that seems intuitively ridiculous, I have to admit, China’s meteoric rise also came with an enormous investment in Treasury bonds. Maybe the secret to economic development isn’t physical capital or exports or institutions; nope, it’s buying Treasury bonds. (I don’t actually believe this, but the correlation is there, and it totally undermines Graeber’s argument that buying Treasury bonds makes you some kind of debt peon.)

Speaking of correlations, Graeber is absolutely terrible at econometrics; he doesn’t even seem to grasp the most basic concepts. On page 366 he shows this graph of the US defense budget and the US federal debt side by side in order to argue that the military is the primary reason for our national debt. First of all, he doesn’t even correct for inflation—so most of the exponential rise in the two curves is simply the purchasing power of the dollar declining over time. Second, he doesn’t account for GDP growth, which is most of what’s left after you account for inflation. He has two nonstationary time-series with obvious exponential trends and doesn’t even formally correlate them, let alone actually perform the proper econometrics to show that they are cointegrated. I actually think they probably are cointegrated, and that a large portion of national debt is driven by military spending, but Graeber’s graph doesn’t even begin to make that argument. You could just as well graph the number of murders and the number of cheesecakes sold, each on an annual basis; both of them would rise exponentially with population, thus proving that cheesecakes cause murder (or murders cause cheesecakes?).

And then where Graeber really loses me is when he develops his theory of how modern capitalism and the monetary and debt system that go with it are fundamentally corrupt to the core and must be abolished and replaced with something totally new. First of all, he never tells us what that new thing is supposed to be. You’d think in 400 pages he could at least give us some idea, but no; nothing. He apparently wants us to do “not capitalism”, which is an infinite space of possible systems, some of which might well be better, but none of which can actually be implemented without more specific ideas. Many have declared that Occupy has failed—I am convinced that those who say this appreciate neither how long it takes social movements to make change, nor how effective Occupy has already been at changing our discourse, so that Capital in the Twenty-First Century can be a bestseller and the President of the United States can mention income inequality and economic mobility in his speeches—but insofar as Occupy has failed to achieve its goals, it seems to me that this is because it was never clear just what Occupy’s goals were to begin with. Now that I’ve read Graeber’s work, I understand why: He wanted it that way. He didn’t want to go through the hard work (which is also risky: you could be wrong) of actually specifying what this new economic system would look like; instead he’d prefer to find flaws in the current system and then wait for someone else to figure out how to fix them. That has always been the easy part; any human system comes with flaws. The hard part is actually coming up with a better system—and Graeber doesn’t seem willing to even try.

I don’t know exactly how accurate Graeber’s historical account is, but it seems to check out, and even make sense of some things that were otherwise baffling about the sketchy account of the past I had previously learned. Why were African tribes so willing to sell their people into slavery? Well, because they didn’t think of it as their people—they were selling captives from other tribes taken in war, which is something they had done since time immemorial in the form of slaves for slaves rather than slaves for goods. Indeed, it appears that trade itself emerged originally as what Graeber calls a “human economy”, in which human beings are literally traded as a fungible commodity—but always humans for humans. When money was introduced, people continued selling other people, but now it was for goods—and apparently most of the people sold were young women. So much of the Bible makes more sense that way: Why would Job be all right with getting new kids after losing his old ones? Kids are fungible! Why would people sell their daughters for goats? We always sell women! How quickly do we flirt with the unconscionable, when first we say that all is fungible.

One of Graeber’s central points is that debt came long before money—you owed people apples or hours of labor long before you ever paid anybody in gold. Money only emerged when debt became impossible to enforce, usually because trade was occurring between soldiers and the villages they had just conquered, so nobody was going to trust anyone to pay anyone back. Immediate spot trades were the only way to ensure that trades were fair in the absence of trust or community. In other words, the first use of gold as money was really using it as collateral. All of this makes a good deal of sense, and I’m willing to believe that’s where money originally came from.

But then Graeber tries to use this horrific and violent origin of money—in war, rape, and slavery, literally some of the worst things human beings have ever done to one another—as an argument for why money itself is somehow corrupt and capitalism with it. This is nothing short of a genetic fallacy: I could agree completely that money had this terrible origin, and yet still say that money is a good thing and worth preserving. (Indeed, I’m rather strongly inclined to say exactly that.) The fact that it was born of violence does not mean that it is violence; we too were born of violence, literally millions of years of rape and murder. It is astronomically unlikely that any one of us does not have a murderer somewhere in our ancestry. (Supposedly I’m descended from Julius Caesar, hence my last name Julius—not sure I really believe that—but if so, there you go, a murderer and tyrant.) Are we therefore all irredeemably corrupt? No. Where you come from does not decide what you are or where you are going.

In fact, I could even turn the argument around: Perhaps money was born of violence because it is the only alternative to violence; without money we’d still be trading our daughters away because we had no other way of trading. I don’t think I believe that either; but it should show you how fragile an argument from origin really is.

This is why the whole book gives this strange feeling of non sequitur; all this history is very interesting and enlightening, but what does it have to do with our modern problems? Oh. Nothing, that’s what. The connection you saw doesn’t make any sense, so maybe there’s just no connection at all. Well all right then. This was an interesting little experience.

This is a shame, because I do think there are important things to be said about the nature of money culturally, philosophically, morally—but Graeber never gets around to saying them, seeming to think that merely pointing out money’s violent origins is a sufficient indictment. It’s worth talking about the fact that money is something we made, something we can redistribute or unmake if we choose. I had such high expectations after I read that little interchange about the IMF: Yes! Finally, someone gets it! No, you don’t have to repay debts if that means millions of people will suffer! But then he never really goes back to that. The closest he veers toward an actual policy recommendation is at the very end of the book, a short section entitled “Perhaps the world really does owe you a living” in which he very briefly suggests—doesn’t even argue for, just suggests—that perhaps people do deserve a certain basic standard of living even if they aren’t working. He could have filled 50 pages arguing the ins and outs of a basic income with graphs and charts and citations of experimental data—but no, he just spends a few paragraphs proposing the idea and then ends the book. (I guess I’ll have to write that chapter myself; I think it would go well in The End of Economics, which I hope to get back to writing in a few months—while I also hope to finally publish my already-written book The Mathematics of Tears and Joy.)

If you want to learn about the history of money and debt over the last 5000 years, this is a good book to do so—and that is, after all, what the title said it would be. But if you’re looking for advice on how to improve our current economic system for the benefit of all humanity, you’ll need to look elsewhere.

And so in the grand economic tradition of reducing complex systems into a single numeric utility value, I rate Debt: The First 5000 Years a 3 out of 5.