Thursday, June 14, 2012

Swiss Fragility

Swiss watches aren't fragile; but can the same be said for the currency floor?

Evan recently had a post looking back on the Switzerland issue, which made me look back at our previous discussion.  In the spirit of Evan's proposition reflection, I'm also going to elaborate on a few of my arguments and look at where they're still true (often in different ways) as well as proven wrong (no surprises).

As I noted in my addition to the previous post, a lot of my previous arguments about the instabilities of a currency peg are not as relevant to the Swiss situation. The Swiss intervention is a floor on the exchange rate, and not a fixed peg. The SNB policy is to intervene on foreign exchange markets so that the Franc/Euro exchange rate can not go below 1.20. In effect, the 1.20 Franc/EUR exchange rate is the strongest the SNB will allow the Franc to be. As a result, the Swiss are not dealing with inflationary pressures as a result of the currency floor, and there's no tension between the domestic monetary policy goal of stabilizing NGDP and the external goal of limiting real appreciation.

Moreover, I do agree with Evan's broad argument that the power of expectations substantially reduces, but does not eliminate, the need for the SNB's forex interventions. However, I'm not quite as optimistic as Evan on the longer term effects of the exchange rate floor. I can't deny that the economic indicators look good: 2.8% annualized GDP growth, 3.2 percent unemployment, etc. But in spite of these positive signs, I want to make a general comment about discussions about expectations and cautionary notes on the fragility of such an arrangement.

First is that expectations cannot work in every instance. My arguments on the instability of currency pegs are a perfect example. Pegs can't function on expectations because speculators can push the bank off the edge because there's not an infinite supply of foreign reserves to defend the peg. While the SNB's policy is not a peg, this comparison leads to an important theorem for expectations arguments. Expectations can only be used as an argument if the policy would work even without expectations. In the case of the currency floor, the forex intervention would stop appreciation even if speculators didn't have expectations of future policies. The SNB could just keep on printing francs and depreciate the currency. If the speculators didn't take into account that they were overpaying for francs, they would be naturally selected out of the market by speculators who would bet that the SNB's floor will hold. Thus, if the SNB commits, they can hold the exchange rate greater than or equal to the 1.2 floor. Speculators, knowing this, would then stop speculating. In the case of unconventional monetary policy, asset purchases and lower IOR in and of itself can raise growth through portfolio balance and hot potato effects. Expectations can help augment these policies because the market would do the work for the central bank, increasing monetary policy's effectiveness. In each of these circumstances, expectations make the policy more powerful, but in and of themselves cannot make impossible policies possible. Expectations are powerful, but we need to make sure to use a rigorous set of criteria before we start applying the argument everywhere.

Second, such large scale asset purchases on the part of the SNB to maintain the peg represent an important form of instability. While I was not prescient enough to make the argument in full, I did comment on the danger of such a large balance sheet filled with foreign currency denominated assets. As Evan notes, a lot of forex intervention from the SNB is the result of absorbing demand for a safe haven currency. Thus, unless policy changes, we should expect the balance sheet to get larger and larger.

As the size of the balance sheet increases, we have to start worrying about fragility, or any ripple effects across international financial markets. To illustrate the problem, imagine a "fundamental" rate path that charts the value of the franc in the absence of a floor from the SNB. This fundamental value would also take into account the demand for a safe haven asset. Given the power of the floor, we can safely assume, currently, the fundamental value is below the floor.

However, if, for whatever reason, the fundamental value shoots above the floor (when the orange part goes above the blue), we can anticipate a sudden large amount of selling and revaluing of assets. All of a sudden, what was once at a fixed exchange rate is now moving. Because the SNB's balance sheet is so large, this shift could have massive implications for the balance sheets of other governments.

The most pernicious part of this shift would be that, given the suppressed volatility from the floor, there would be no way for the SNB to tell when we should start to get worried. Even if the fluctuation was pure static, like that shown in the graph, there would be a massive signal confusion problem. Policy makers could guess, but prediction is often quite difficult. It would also be unlikely that other indicators, such as unemployment or inflation, would be update quickly enough to keep up with changes in forex flows. This represents a key form of fragility that we need to worry about in all expectations regimes. If, for some reason, the regime changes, many arrangements that depended on the previous regimes have to be recreated, with sometimes drastic consequences.

To get around such issues, the exchange rate needs to be less about maintaining some floor and more about maintaining a favorite target, such as forecasts of NGDP growth. This way, the market is allowed a natural level of volatility, and policymakers can use this volatility to gauge the state of demand for the currency. Additionally, this natural, day to day volatility would caution those who hold the Franc from depending too much on its 1.2 floor. Alternatively, increasingly draconian capital controls could be used to change the fundamentals of holding currency in Switzerland, thereby limiting capital inflows while allowing volatility in the exchange rate. However, such an approach is subject to leaks, and given the manual smuggling of currency, is likely to fail.

Third, and as a corollary to the second point, the question we need to ask whenever we talk about regimes that fix certain values is "what are the costs?" As Miles Kimball noted, central banks can do a lot of things, so our attention needs to shift to the collateral effects of such actions. And as Bastiat reminds us, we need to look at that which is hidden in addition to the more obvious effects. While a lot of the more mainstream discussions focus on welfare costs of volatility, I worry more about fragilities in the spirit of Taleb or fault lines in the spirit of Rajan. I've commented on this line of thought in the context of NGDP targeting (here and here), and I think it has an important role in any kind of system that suppresses volatility, such as the exchange rate peg.

I can't quibble about the SNB's effective block of the Franc's appreciation, but we must qualify its success. Beware manufactured stability, especially when it creates fragilities and uncertainties that we aren't prepared for.

No comments:

Post a Comment