Wednesday, August 24, 2011

Insuring $12 billion in gold

MR pointed to a fantastically intriguing article yesterday on Venezuela's intentions to transport $12 billion in gold-- 211 tons of it-- back to Venezuela, presumably Caracas. Before we get into it, this is a great phrase:

"...the market in physical gold is tiny, and largely comprised of nutcases."

Not sure if that became the case when Chavez got involved or if he just reinforced that characteristic. Anyway, experts estimate that this could take 40 trips. (Sorry, one more diversion-- but who is an expert in transporting 211 tons of gold?) Naturally, insuring this transfer is something that needs to be considered. The article claims that no company would be willing to take that contingency onto its books (though I think that's more a function of dealing with Venezuela than the sheer value of gold, though both certainly do matter).

As a side note: If Venezuela considers itself socialist-- either in practice or doing what they can to get there-- doesn't the government play the role of providing insurance? Why hasn't anyone brought this point up? Insuring your own transfer makes about as much sense as anything Chavez does anyway.

Anyhow, I think people are thinking of this as a financial issue and not quite enough as a practical issue. Does there exist a risk premium by which, say, Lloyd's would be willing to take this on? I'm sure there is-- in the only previous instance of even beginning to approach this level of gold transfer, the rate came to 3.3%. Chavez' rate would undoubtedly be higher since there's a decidedly non-zero chance that he'd be involved in any nonsense that arises in transporting all that gold. Then again, it's presumably an international deal-- between Venezuela and an insurance company-- so the insurance company can decide not to pay and there's really no legal recourse towards claiming a settlement.

But it would seem that a superior risk/reward tradeoff could be achieved by simply arranging a very large number of transfers-- like in the thousands? I didn't see anything in the article saying that Chavez needed all of the gold quickly, so time's not a large problem. Let's say the going risk premium is 10% for Chavez to keep the numbers easy. That means he's paying $1.2 billion in premiums. What if you didn't pay any premiums and made any of the following arrangements:

- 1 million transfers of $12,000 in gold. Plus: That's a small amount of gold and could be transported in just about any way possible-- civilian aircraft, boats, anything. At current prices, that's what, about 6 or 7 ounces. Minus: That's a whole lot of transfers; coordination costs would be high.

- 100,000 transfers of $120,000 in gold. It's still a fairly small amount of gold-- 60 or 70 ounces, maybe 4 or 5 pounds-- so much of the benefits remain from above while the number of transfers is reduced by an order of magnitude.

- 10,000 transfers of $1.2 million in gold. At this point, that's a considerable sum of money and a lot of transfers to arrange-- the coordination costs here probably outweigh the benefits from spreading the money around.

I think the middle one provides the best characteristics-- yes, it's a lot of transfers, but you'd need more than 10,000 of those transfers to encounter trouble in order to outpace the risk premium (I know I came up with 10% to keep things easy, and that's where 10,000 comes from, but even at 3.3% you'd need 3,300 thefts.) So why not break it up and take your chances with not being robbed thousands of times? You're effectively insuring yourself this way (not in the facetious way I alluded to above). Plus, incentives are high to make certain nothing bad happens to the gold since it's coming right out of your pocket every time the gold never makes it.

And yes, I realize that hiring shady people may be a problem-- but if that's the concern, you can't automatically assume that away at any point in the process. So that impacts all plans. In fact, if you assume that shady people are more likely to get involved when there's more money to be had, then it would be MORE of an issue in large-scale transfers like the ones described in the article.