Monday, 27 March 2017

Articles & Op-Eds

Cash and Crime

February 21, 2017

This is the third of several posts on Ken Rogoff’s The Curse of Cash. As summarized in an earlier post, Rogoff argues that banning physical cash has two major benefits: reducing crime and enabling effective monetary policy at the zero lower bound. In this post, I will address the first of these supposed benefits by considering the relationship between cash and crime. I’ll look at the relationship between cash and monetary policy in my next post.

Rogoff acknowledges some costs of banning cash (e.g., a reduction in financial privacy). As such, his consequentialist case in favor of banning cash requires that the benefits exceed these costs. He does not attempt to estimate the benefits and costs of banning cash. Rather, he suggests the benefits are so large (and costs so small) that more precise estimates are unwarranted.

I believe Rogoff exaggerates the benefits from reducing crime by banning cash in two important ways. First, he overestimates the extent to which cash is employed by criminals. Second, he overestimates the social costs of the typical crime. If (1) cash is less important in criminal transactions than Rogoff suggests or (2) the net cost to society of the typical crime is lower than Rogoff suggests, then the benefits from banning cash are lower than Rogoff suggests. If the benefits from banning cash are lower than Rogoff suggests, then it is less likely that they are sufficiently large to warrant the costs of banning cash that Rogoff accepts.

How important is cash for crime?

In an earlier post, I described Rogoff’s estimate for the extent to which cash is held by criminals in the domestic economy. Briefly, x = 1 – a – b – f – c, where x is the share of currency used by criminals in the domestic economy, a is the share of currency held abroad, b is the share of currency held by banks for legitimate purposes, f is the share of currency held by firms for legitimate purposes, and c is the share of currency held by consumers for legitimate purposes.

Rogoff discusses some interesting methods for estimating a, which he puts at 50 percent. He gets b, which is around 1 percent, from the Federal Reserve’s H.3 release (table 2). The estimates get worse from here.

For f, he relies on a simple back of the envelope calculation offered by Porter and Judson (1996). In a footnote, they write:

Most businesses need nothing more than seed cash to operate, and the total amount of such cash is not likely to be significant, as the following calculation shows. Almost 2.7 million retail establishments existed in 1992. Taking certain elements of cash use at supermarket chains as the standard for all retail establishments that year, assume that each establishment had ten cash registers (currently the median number for supermarket chains) and each register contained $200 of seed cash (the amount that at least one large supermarket chain uses for that purpose); then the total currency holdings by all retail establishments would have been only $5.4 billion, or 1.8 percent of the total stock of currency at the end of 1992. If, in addition, one business days’ worth of total consumption was always in transit to depository institutions, the total amount from both of these sources would have been only $22.3 billion, or only 7.7 percent of total currency holdings in that year.

Taking “the huge trend of decline in cash used for medium and large retail transactions over the past 20 years” that Rogoff refers to on page 49 (without actually citing any evidence) as given, we might put f in the neighborhood of 5 percent today.

  1. That’s roughly sixty percent of the currency accounted for. What about the rest? This is where things get interesting. Rogoff uses survey evidence of consumers to estimate c. The assumption is that any cash that consumers surveyed don’t fess up to holding must be held for nefarious purposes. There are at least two problems with this approach. First, cash holdings are probably largest among those on the ends of the wealth spectrum. The very poor hold nearly all of their wealth in cash because it doesn’t pay to have a bank account. Think: $4k in savings, hidden throughout the house, because that’s all that we have. The very wealthy hold nearly none of their wealth in cash—but, since they are so wealthy, “nearly none” is still a substantial amount. Think: $4k on the counter because it’s hard to know where the weekend will take us. The problem: it seems reasonable to think these two groups are less likely than those in the middle of the wealth spectrum to answer survey questions. And, if I am right about that, surveying the great middle leads us to underestimate the extent to which cash is held for legal transactions.

Second, it is problematic to assume any cash undeclared in surveys is held for nefarious purposes. As Rogoff notes in his discussion of the zero lower bound, it is costly to hold cash. For one reason, you have to keep it secure. Let me suggest that one way to reduce the costs of securing cash is to refrain from telling others that you hold a lot of cash. Likewise, some people hold cash for financial privacy. How likely is it that one valuing financial privacy will accurately report cash holdings? These concerns suggest that Rogoff’s estimate for c—between 5 and 10 percent—is too low. And, the more he underestimates c, the more he overestimates x.

Now, perhaps this is immaterial. We might agree that Rogoff’s estimate that more than a third of all US currency in circulation is used to commit crime in the domestic economy is too high. But, by how much? And, even if we doubled the estimate for c, we are still left with more than one fifth of currency in circulation held by criminals. Perhaps there is less crime to eliminate than Rogoff thinks, but still enough to warrant banning cash. That brings us to the next point.

How large are the social costs of the typical crime?

There is no denying the large net social costs of some crimes: murder, human trafficking, property theft, identity theft, etc. But Rogoff implicitly assumes that all crimes—or, at the very least, the typical crimes enabled by cash—are a net cost for society. One might reasonably argue, however, that some criminal activities have net benefits for society.

Let’s start with the simple consideration of property theft, where John steals a nice mug worth $5 from Sally. Tally it up. John gains roughly $5. Sally loses $5. And the rest of us lose because, knowing that someone might steal our mugs, we incur more costs to protect our property than we would in a world without theft (or, without as much theft). Of course, reality is more complicated. We would like to include any psychic benefits to John or psychic costs to Sally; perhaps it’s not a mere transfer of wealth from Sally to John. But, still, the basic idea is that this is a net loss to society. We would be better off in a world without theft.

But other crimes seem to lack a victim. Who is harmed when John buys marijuana from Sally? Not John, who values the pot more than the cash he gives up. Not Sally, who values the cash more than the pot she gives up. Where’s the social cost? Perhaps there are some psychic costs to third-parties who dislike the idea of others using drugs. Perhaps John doesn’t evaluate his own benefits well and purchasing marijuana actually makes him worse off. But, at the very least, I think we can agree that not all crimes are obviously a net cost to society. Indeed, to the extent that those transacting outside the law realize gains from trade, there might be a net benefit to society when some criminal transactions take place.

Rogoff seems to take it for granted that the law reflects some social calculation about the desirability of various actions; that, if some transaction has been banned, it must be bad on net for society. But that seems unlikely to be true in all cases. Surely there are some laws we would be better off without.

It also makes for some odd calculations. The transaction between Sally and John in Ohio, where purchasing marijuana is a crime, would be chalked up as a net cost to society. But this same transaction would be chalked up as a net benefit in Colorado, where marijuana is legal. Perhaps that’s right. Perhaps third parties in Ohio are more troubled by the actions of others than those in Colorado. But it seems like an odd assumption to make.

Interestingly, Rogoff acknowledges the existence of bad laws. For example, on page 76, he notes his support for allowing more legal migrants to enter advanced economies like the US. If more legal immigration is the best-case scenario, it is at least conceivable that the large amount of illegal immigration experienced in the presence of the current restrictions on immigration when cash is available is better for society than the small amount of illegal immigration in the presence of the current restrictions that would result after cash is banned. And, if that is the case, it would be wrong to count the reduction of illegal immigration that would follow from a ban on cash as a benefit to society.


Let me be clear: I am not saying that there isn’t much crime or that we should ignore the genuine costs of crime to society. There is a lot of crime and some crimes clearly make us worse off. Rather, I am saying that (1) Rogoff overestimates the extent to which cash enables crime and that (2) the presence of bad laws means that some crimes—specifically, those actions in violation of poor laws—might well be a net benefit to society. As such, the benefits of reducing crime by banning cash are lower than Rogoff implies.

Note further that I have said nothing about the assumption that banning cash will reduce crime. That, I am afraid, will have to wait for another post.