Back to gold?
[This weekend, I am attending a conference that examines the gold standard. Here are my thoughts going into the conference.]
People occasionally ask me whether it would make sense to go back to the gold standard. Most economists think that this would be a bad idea. I agree, but not necessarily for the reasons that most other economists would cite.
It’s hard to debate this issue on purely theoretical basis, as much of the debate ends up being about whether the historical record of the gold standard is superior to that of fiat money. That turns out to be an extremely difficult question to answer, for all sorts of reasons. And even if we could answer this question, we’d face another question: Would a gold standard in the 21st century perform as well as the 19th century version?
And before these questions can be answered, we face an even trickier question: What do we mean by a gold standard? What is fiat money? History provides examples of both good and bad gold standards, as well as good and bad fiat money. Which systems should we compare?
My preferred definition of a gold standard is one where currency can be converted into gold at a fixed nominal price, in a wide range of leading developed economies. By that definition, the world was on a gold standard from 1879-1914, 1926-33 and (perhaps) approximately 1950-68. That last period is especially iffy, as Americans were not allowed to freely convert dollars into gold. I include Bretton Woods here, however, because some proponents of the gold standard cite is an example of how fixing the price of gold can prevent extreme inflation. We all know what happened after 1968, when the gold price peg ended.
Note that my preferred definition of a gold standard is not my preferred gold standard. In my preferred gold standard, the government would merely define the unit of account as a fixed quantity of gold, and then do nothing. For example, “The US dollar is one gram of gold”. That’s all. No central bank, no government currency issue, no regulation of banking, etc. That sort of international gold standard never existed. If that sort of system is viewed as the theoretical ideal, one might say that 1879-1914 was an 80% gold standard, 1926-33 was a 60% gold standard, and 1950-68 was a 20% gold standard.
One problem I have with some gold proponents is that they cite how the $35/oz gold price peg prevented runaway inflation until it was abandoned in 1971, and then disavow any role of gold in the severe deflation of 1929-33. Each argument has some merit considered in isolation, but when viewed together these two claims make little sense. You can’t have it both ways, taking credit for a 20% gold standard and then saying a 60% gold standard isn’t really a gold standard. Even worse, many gold proponents cite 1971 as the end of the fixed price of gold. But a fixed free market price of gold is the sine qua non of a gold standard, and that ended in March 1968. After the market price of gold started rising, the $35 official price was completely meaningless. (I believe the official price today is $42.22/oz.) After March 1968, central banks could “freely” convert dollars into gold in much the same sense that in the late 1980s the Japanese could “freely” sell cars in America under Reagan’s “voluntary” export restraint program.
BTW, gold proponents should prefer to use 1968 rather than 1971 as the ending date for the gold standard, as it actually makes their argument stronger. Inflation was getting much worse during that 3 1/2 year period.
So what’s the strongest argument in favor of a gold standard? The most persuasive arguments that I have seen do roughly the following:
1. They concede that the system only works well if most important countries adopt it. In recent decades, the purchasing power of gold has been extremely unstable. Any single country returning to gold would only be able to modestly reduce that instability. Thus we would have to hope for a truly international system.
2. They do not compare the gold standard to fiat money. They do not compare the best version of the gold standard to the best version of fiat money (which is inflation/NGDP targeting). Rather they often compare the best version of the gold standard (1879-1914) to all of fiat money. That includes the poorly performing unanchored system of 1968-90, and also the period of implicit or explicit 2% inflation targeting (1990-2022.) In my view, it would be more logical to either include the poorly performing interwar gold standard, or exclude the fiat money system before inflation targeting was adopted. I can’t speak for other economists, but when I say that I prefer that we stay on fiat money, I am not suggesting that the fiat system of 1968-1990 was better than the so-called “classical” gold standard. I’m saying the best of fiat is better than the best of gold, and that the entire fiat system in the US is better than the entire gold standard.
3. Gold proponents tend to highlight the metrics by which gold looks good, and ignore those by which fiat money does better. Under the international gold standard, the long run rate of inflation was roughly zero. In addition, the price level a few decades out could be predicted with some degree of accuracy. However, there was a great deal of year-to-year inflation volatility. In addition, the price level followed roughly a random walk. That means the near zero average inflation of 1879-1914 was partly (not entirely) coincidence. Prices trended lower during 1879-1896 and trended higher from 1896-1914. And even within those sub-periods, there was substantial year-to-year fluctuation in the rate of inflation.
Now I’ll make some empirical claims that gold proponents may reject. I believe the post-1990 regime of 2% inflation targeting produced a better outcome than even the best version of the international gold standard. We do have more inflation (2% on average, vs. zero), but that’s because policymakers decided that 2% trend inflation was preferable. There are good arguments both ways on that point, but to me it’s roughly a wash. The welfare difference from 0% and 2% trend inflation are trivial (if anything, I slightly prefer 2%). I also believe that year-to-year volatility of inflation was less under 2% inflation targeting, although the poor quality of older price indices makes that a bit debatable. And I believe that with 2% inflation targeting people are better able to forecast where the price level will be 20 years in the future, as compared to the international gold standard. Once again, that claim is debatable, but I think I’m right. So in terms of the sort of nominal stability that is important for social welfare, I believe inflation targeting does a bit better. (All my views are provisional, based on 1991-2020. If the Fed doesn’t get this current inflation under control then I may change my mind.)
You can also compare the two systems using other criteria, such as the business cycle, but we don’t have very reliable data on real output stability from the 19th century, and in any case the economic system was so different that we have no way of knowing if any differences are due to money and not some other factor like the shift from farms to factories to services, or changes in wage flexibility, unemployment comp., etc. It would be like saying, “Fiat money has produced better telephones than did the gold standard.” It’s better to stick to nominal stability, the one thing monetary policy can clearly affect.
4. Gold proponents say that some of our problems under the gold standard were due to bad banking regulations. I think that’s true, and it’s an underrated point that is overlooked by gold’s critics. On the other hand, if we adopt an international gold standard then we’d like it to be robust enough to survive bad banking regulation.
5. Gold proponents often point to the gold standard’s ability to constraint governments, to prevent them from engaging in policies that make the value of money unstable. But when asked to account for the extreme instability in the value of money during 1926-33, they (correctly) point to government meddling in the monetary system. I don’t know how you can have it both ways. If you assume the sort of good government that would allow a theoretically pure gold standard to run without interference, wouldn’t that sort of government also be able to do effective inflation targeting, perhaps at zero percent inflation (if that’s your preference?) Historically speaking, gold standards don’t seem to constrain bad governments.
6. On a related point, it’s not clear how we should think about wartime. Proponents of the gold standard cite price stability data from peacetime, excluding periods such as 1861-79 and 1914-26. In one sense that seems fair, as key countries were not on gold during those periods. But that raises the question of what do gold proponents favor during wartime? If they believe the gold standard system cannot be blamed for the extreme price level instability during and after war, then presumably they favor some alternative policy. But what is that alternative policy? Staying on gold? What if that causes a country to be unable to raise enough revenue to win the war? Return to gold at a high price, in order to prevent postwar deflation? Maybe, but that sort of policy is actually far more difficult than it looks.
One big problem with the “look at history” argument for a gold standard is that we don’t have many good examples of gold standard regimes doing well during major wars. It’s fair to say that the gold standard shouldn’t be blamed for 1861-79 and 1914-26, but it’s also true that we have no evidence that things would have been better (in an overall welfare sense, admittedly prices would have been more stable) if countries had remained on gold and refrained from selling central bank gold reserves during wartime. Gold proponents are excluding periods where running a successful gold standard would have been especially challenging.
7. Gold proponents deny that a gold standard would lead to more mining of gold (which might be socially wasteful), correctly pointing to the rise in the real price of gold after 1970. They attribute this increase to the fact that private gold demand increased as a hedge against rising inflation.
8. The purchasing power of gold has been extremely unstable in recent decades. Gold proponents respond by pointing to the relative stability of the purchasing power of gold during 1879-1914, and suggest that the recent instability is due to the fact that the world is not on a gold standard. As far as the 1970s is concerned, I agree. See point #7. But I don’t believe that is true of more recent gold value fluctuations.
During the 2000s, the relative price of gold skyrocketed (see above). If this had occurred when the gold standard was in place, then there would have been a massive fall in the global price level, and perhaps another Great Depression. Gold proponents sometimes suggest that the increase in gold prices reflected people buying gold as a hedge against inflation. I do buy that argument for the 1970s, but not for the 2000s. There was very little inflation during the 2000s, and the modest long-term nominal interest rates suggest very little fear of high future inflation. Instead, I’d point to the rapid rise in gold demand in important developing countries such as China and India, each of which has a population comparable to the entire western world.
Do I have evidence for this claim? Yes, it wasn’t just gold. The enormous economic boom in Asia drove up the relative prices of a wide range of commodities during the 2000s, not just gold. In some respects, the 2000s were like the 1870s and 1920-33, when rising demand for gold caused gold’s value (purchasing power) to rise sharply. In the 1870s and 1920s it was many countries joining the gold standard and building or rebuilding their gold stocks. In the 2000s, the same would have occurred as China and India effectively joined the world economy. Unlike in the 1870s and 1920s, we didn’t see a big deflation because the increase in the value of gold was accommodated by a higher nominal price. But under a gold standard the nominal price is fixed and changes in the value of gold require a change in the overall price of goods and services.
9. Research by Barksy and Summers suggests that the Gibson Paradox (the tendency for the price level to be positively correlated with nominal interest rates under the gold standard) was due to gold demand rising when nominal interest rates fell. Recall that under the gold standard, the nominal interest rate is the opportunity cost of owning gold. People demanded more gold when nominal rates fell, the value of gold rose, and the price level fell. Given that interest rates now fall to zero during recessions, there is a greater danger of massive gold hoarding during the 21st century than during the 19th century.
Can a gold standard work well in the 21st century? Perhaps if at least most of these occur:
1. Almost all major countries agree to join.
2. There are no more China shocks (which is plausible).
3. There are no more world wars (which is plausible).
4. We avoid banking crises by adopting a completely laissez-faire banking system.
5. Wage flexibility returns to 19th century levels as minimum wage laws, labor union laws, etc., are abolished.
6. Central banks are abolished and governments don’t meddle in the system by varying their demand for gold reserves.
7. Governments run responsible fiscal policy, as deficits could not longer be monetized.
8. Governments credibly promise never to leave the gold standard during a recession, as fear of devaluation can trigger massive gold hoarding, turning a recession into a depression.
9. Interest rates return to more “normal” levels, well above zero.
I understand the argument against my proposal for NGDP level targeting; there’s only a 1% chance the US government would adopt the system and stick to it. My response is that there’s less than a 1% chance that the world’s major governments would agree on an international gold standard and somehow do the various things above needed to make it work.