Follow the Money

Turkey’s currency initially rallied after the central bank raised interest rates yesterday. Perhaps a bit more orthodoxy is all it will take to restore a modicum of stability to Turkey’s markets. Then again, the lira’s rally didn’t last long. Even if Turkey firmly commits to a somewhat more orthodox monetary policy, Turkey retains substantial vulnerabilities: Turkey’s current account deficit was quite large even before oil rose to $80, and Turkey imports a ton of oil (and natural gas). Turkey has a dollarized financial system and trades heavily with Europe. It thus doesn’t benefit much from the rise in imports that Trump’s fiscal stimulus has generated, but gets hurt by higher interest rates on its dollar borrowing. Turkey has a decent stock of external debt, and much of that is denominated in foreign currency. Turkey’s banks lend domestic dollar deposits to Turkey’s firms, who have more foreign currency debt than they have external debt. And, Turkey could use a few more reserves.  Turkey's reserves fall far short of covering its external financing need over the next year. There are three important things to know about Turkey’s foreign exchange reserves. A lot of Turkey’s reserves are in gold and thus not in dollars or euro. Turkey’s debts though don’t settle in gold. A lot of Turkey’s reserves are borrowed from the domestic banks, who can meet their reserve requirement on lira deposits by posting either gold or foreign exchange to the central bank.  Most of the gold, and a decent chunk of the foreign exchange, is effectively borrowed. That limits the pool of funds available to intervene directly in the foreign exchange market (though it also provides the banks with a bit of a buffer). Turkey’s reserves don’t come close to covering its maturing external debt, let alone its external financing need, no matter how you cut it. Turkey has about $180 billion in external debt coming due (around $100b from the banks, $65b from firms).* It has a $50 billion (give or take) current account deficit. That’s a one-year external financing need of close to 30 percent of Turkey’s pre-depreciation GDP, against 12 percent in total reserves and about 10 percent of GDP in foreign exchange reserves. And the bulk of Turkey’s maturing external debt is denominated in a foreign currency—it thus is a claim on reserves that cannot be depreciated away. On classic measures of external vulnerability, Turkey—like Argentina–– is much more under-reserved than the IMF’s reserve metric would suggest (M2 and exports aren’t large versus GDP, so they pull the reserves Turkey’s needs to meet the IMF's standard well below maturing short-run debt). And even if it is graded on the IMF’s generous curve, Turkey falls short (See paragraph 19 of the IMF’s latest staff report). These vulnerabilities aren’t new. Turkey has long looked vulnerable to an Asian style financial crisis—one triggered by a loss of access to bank funding and a bank-corporate doom loop from the private sector’s foreign currency debts. And it has some of Argentina’s old vulnerabilities too, with roughly $200 billion in domestic foreign currency deposits (data from Turkish Central bank)) in addition to $400 billion plus in external debt (see this classic book on emerging market crises for background on the Asian and Argentine crises). These long-standing vulnerabilities haven’t triggered a full-on crisis yet. Turkey’s domestic deposit base and its external funding has historically been fairly sticky.   It has been surprisingly resilient in the past. Yet there are reasons to think Turkey faces a more difficult challenge now. U.S. rates are rising when oil is going up, and that’s a bad combination for an oil importer with lots of dollar debt. While Turkey has long looked bad on classic indicators of external vulnerability, it now is starting to look really bad—short-term debt has jumped a bit relative to GDP (thanks mostly to a surge in corporate borrowing, which shows up in the “trade credit” line) and the external funding need is now close to three times liquid (non-gold) foreign exchange reserves. Turkey though differs from Argentina in one critical respect. Its government hasn’t been the biggest external borrower, and it doesn’t have the biggest stock of foreign currency debt in the economy. That honor goes to Turkey’s firms, who have almost $340 billion in foreign currency denominated debt ($185 billion is owed to domestic creditors, and $150 to external creditors—$110 in loans and $40 in trade credit [source]). The quality of their hedges will be tested: I never have been convinced all foreign currency debt is really backed by export receipts. What really makes Turkey interesting, though, is its banks. They have a rather fascinating balance sheet and engage in some fairly creative forms of financial intermediation, with a bit of regulatory help. The core problem Turkey’s banks face is simple. Turks want to hold a large chunk of their savings in dollars. And foreigners want to lend to Turkey in dollars (or euros). But Turkish households want to borrow in lira (in fact the banks cannot lend to households in foreign currency—a sensible prudential regulation). So the Turkish banking system has a surplus of foreign currency funding—and a shortfall in lira funding. There is an easy way to see this. Look at the loan to deposit ratio in foreign currency. It’s clearly below 1, about 0.8. And then look at loan to deposit ratio in lira. It is well above 1—it is now about 1.4. (See the IMF’s 2016 staff report [Paragraph 46], or the most recent financial stability report of the Central Bank of Turkey, starting on p. 50. Chart III.2.4 on p. 51 has the loan to deposit ratio by currency).** So how do the banks transform dollars into lira? A couple of tricks: They swap a lot of dollars into lira. Fair enough. But the tenor of the swaps is fairly short (see box III.2.1 of the central bank’s financial stability report . The “lira” can run even if the dollars raised by selling longer-dated bonds cannot.) And the central bank lets the banks meet their reserve requirement in lira by posting foreign exchange or gold at the central bank (so gold deposits in effect fund lira lending, indirectly). This means the banks have a decent buffer of foreign exchange—which they can draw on if their creditors start to withdraw funding. One secret source of strength of the system is that the banks themselves have a fairly large stockpile of foreign exchange deposits that matches their large short-term external debts.*** Makes for a strange system. So long as the domestic hard currency deposits don’t run, the banks ultimate funding need is in lira. In addition to transforming foreign exchange funding into lira lending, the banks do a lot of classic intermediation—borrowing short-term (there aren’t lots of sources of long-term lira funding) to fund lira denominated installment loans and mortgages. The banks consequently are exposed to an interest rate shock—in much the same way the U.S. savings and loans were back in the 1970s (they funded long-term mortgages with short-term deposits). This has a plus—raising domestic interest rates sharply would quickly slow bank lending and reduce demand, helping to close the current account deficit. But it also encourages a lot of creativity on the part of the central bank, which knows—I think—that the banks ultimately rely on it for lira funding and are vulnerable to an interest rate shock. And historically at least, the regulators have often preferred to use macroprudential limits to cool the economy rather than rate hikes, though it isn’t clear if that would be enough right now. Bottom line: Turkey cannot really use its reserves to try to defend the lira. Selling off some of its already limited reserves to cover an ongoing current account deficit would create the perfect conditions for a run on the banks’ foreign currency liquidity to develop. A free fall in the lira would cause corporate distress, even if it would take a lot to really threaten the government’s solvency. And raising rates sharply would squeeze the banks ability to lend—slowing the economy, but helping to close the current account deficit. Pick your poison. * Thanks to $120 billion or so in short-term external debt and the scheduled roll off of a fraction of its long-term claims. ** The CBRT (emphasis added): “Depositors’ FX deposit preferences and the change in favor of the TL in the loan composition of banks led to a widening in the gap between the TL and FX L/D ratios. The difference between TL and FX L/D ratios indicates that banks need TL liquidity. As a result of depositors’ FX deposit preferences and banks’ TL liquidity needs increased FX swap transactions with foreign residents. Therefore, the amount, maturity, cost and counterparty structure of FX swap transactions have recently become important with respect to monitoring the liquidity risk of banks.” *** It helps that domestic depositors tend to switch out of lira and into domestic foreign currency deposits in periods of stress. The banks can absorb a loss of external funding for a while (they likely have something like $50 billion in liquid assets at the central bank, and presumably could borrow against their gold too) but not a simultaneous loss of external funding and a run on their domestic dollar deposits.  

Probably only connoisseurs of IMF Access Policy debates understand the importance of being judged “sustainable with a high probability” by 19th Street. That’s what unlocks the keys to the kingdom, so to speak. Sustainable, with a high probability, countries can borrow large sums from the IMF without having to worry about locking in their private creditors in some way. What happens when a country that is sustainable, but not with a high probability, needs large sums? Well, it is complicated. I suspect I could litigate the precise meaning of the early 2016 access policy decision with the best of them, and I am not sure. The Fund broadly expects private creditors to maintain their exposure, but that goal can be achieved in a lot of different ways (selling some bonds to the private IMF?), and it may not always be necessary.* I have little doubt that the Fund considers Argentina to be sustainable with a high probability (the last Article IV signaled more concern about the value of the exchange rate than about debt sustainability), and thus eligible for exceptional access. The U.S. certainly does. I don’t disagree—public debt to GDP isn’t that high (yet). But I do think it is a closer call than many. Largely because of a variable that isn’t the focus of the new access policy. Namely, exports, or the lack of them. In 2017, exports were only a little above 10 percent of Argentina’s GDP. As result, Argentina’s external debt—the debt it owes to the rest of the world—is pretty high relative to Argentina’s limited exports. External debt of thirty-five percent of GDP should be a bigger concern in an economy that exports 11-12 percent of its GDP (and relies on the almighty soybean, often processed, and other agricultural products for about a half of its exports) than in an economy like Germany, or Korea. The same point applies to Argentina’s 5 percent of GDP current account deficit. It is large relative to Argentina’s export base. Argentina thus isn’t a simple liquidity crisis—a crisis caused more by too few reserves than by too much overall debt. And Argentina also isn’t (yet) a country that obviously has too much overall debt. But it is a country that is adding to its (external) debt at too rapid a pace.** Argentina thus falls into the category of countries that are “sustainable if they can successfully adjust.” The Fund thinks that Macri’s government can carry off the needed fiscal adjustment, and that financing to allow a smooth path of adjustment will make the needed changes socially and politically sustainable. The Fund also believes—see the Article IV—that a gradual tightening of fiscal policy will allow a looser monetary policy and thus help bring the peso and the current account down over time. Fair enough. Ultimately, this is the kind of judgement call that the IMF has to make. But the path to sustainability strikes looks difficult. Argentina’s external debt isn’t really 35 percent of its GDP. That’s the ratio at an over-valued exchange rate. With a 25 percent depreciation, external debt rises to close to 50 percent of GDP (and Argentina now pays on average about 6 percent on its external bonds, — so interest payments aren't small relative to Argentina’s export base). Exports would also obviously go up versus GDP if the peso depreciates, so the ratio of external debt and debt service to exports wouldn’t change).***  I did a quick estimate of what happens to Argentina’s debt to GDP path if:  it reduces its external borrowing to $20 billion a year (likely by reducing its current account deficit, but it could come through a rise in foreign direct investment); it borrows $30 billion in reserves from the IMF (e.g. an IMF program) and; the peso depreciates by 25 percent (lowering GDP in dollar terms commensurately).   This is a bit of ballpark math—but hopefully it is a reasonable bit of ballpark math.**** Thanks to the need to borrow reserves, external debt to GDP rises to close 60 percent of GDP at the end of 2019, against exports of around 15 percent of post-depreciation GDP and reserves of around 15 percent of GDP. A bad harvest, a large rise in the dollar (as U.S. treasury yields rise to pull in the funds the U.S. needs and prevent the economy from over-heating in the face of unneeded fiscal stimulus), or bad dynamics around the peso and the growing stock of short-term peso debt that Argentina needs to roll all could throw the needed adjustment off track. Argentina needs enough foreign demand for its bills to allow the central bank to lower rates. But not so much that the inflow of funds bids the peso up and prevents adjustment in the current account. Argentina went into its 2001 crisis with an external debt to GDP ratio of 60 percent and a current account deficit of around 3 percent. So Argentina is on a trajectory that will potentially get it close to quite problematic debt levels.  The main difference is that back in 2001 Argentina’s currency board—and its highly dollarized banking system (see the case study on Argentina here)—made depreciation impossible without a crisis. The fact that the peso now floats gives Argentina a chance to put its external debt on a more sustainable long-term trajectory without creating a domestic banking crisis or throwing the economy into deep recession. But I do think that the Fund needs to look a bit beyond public debt when it thinks about an economy like Argentina. The combination of a low level of domestic savings and a low level of exports have long been Argentina’s Achilles heel. */ The exceptional access policy decision for cases in the grey zone: ”It would be appropriate for the Fund to grant exceptional access so long as the member also receives financing from other sources during the program on a scale and terms such that the policies implemented with program support and associated financing, although they may not restore projected debt sustainability with a high probability, improve debt sustainability and sufficiently enhance the safeguards for Fund resources … Directors noted that, in applying this more flexible standard in circumstances where debt is assessed to be sustainable but not with high probability, there would be a range of options that could meet the prescribed requirements. There would be no presumption that any particular option would apply. Rather, the choice would depend on the circumstances of the particular case … If the member has lost market access and private claims falling due during the program would constitute a significant drain on available resources, a reprofiling of existing claims would typically be appropriate …. In this context, the scope of debt to be reprofiled would be determined on a case-by-case basis, recognizing that it would not be advisable to reprofile a particular category of debt if the costs for the member of doing so—including risks to domestic financial stability—outweighed the potential benefits.” Not clear? Good. There should be a bit of flexibility. The staff paper behind the exceptional access policy decision can be found here [PDF]. **/ Though between the growing stock of central bank paper—Lebacs—and short-term domestic law dollar borrowing—Letes—Argentina also does have too much short-term debt. ***/ Compared to the Fund, I obviously put more weight on external debt relative to exports than public debt to GDP. To make this difference concrete, think of the difference between Japan—a high domestic public debt country that is a net creditor to the world, and Argentina. Japan is a net creditor to the world and its government’s solvency improves with a yen depreciation (as Japan’s government has lots of foreign currency-denominated assets, and its debts are all in yen). Argentina is a net borrower from the world, and primarily in foreign currency, so its solvency deteriorates when the peso depreciates. Countries like Russia and Brazil are more resilient because their governments are net long in foreign currency, as their central banks have more reserves than the government (and the big state firms) have foreign currency debt. ****/ I didn't model peso and foreign currency debt separately, so I implicitly assumed that foreign investors would make up any losses from the peso's depreciation through high interest rates. As most of the peso debt seems short-term, that is likely to be close to true. I also would recommend the external debt sustainability analysis in the IMF's latest staff report (buried a bit in Annex I, which starts on p. 41); the real depreciation shock modeled on p. 48 scares me, but in the model it comes after several more years of current account deficits).  

Korea has indicated that it will, very gradually, start to disclose a bit more about its direct activities in the foreign exchange market. The Korean announcement presumably was meant to front-run its currency side agreement with the United States. Optics and all—better to raise your standard of disclosure unilaterally and then lock in your new standard in a trade deal than the reverse. The problem is…Korea’s actual disclosure commitment is underwhelming. If this is all the U.S. is getting out of the side agreement, it is a bad deal. It sets too low a bar globally, and fails to materially increase the amount of information available to assess Korea’s actions in the market. Many emerging markets disclose their purchases and sales (separately) monthly, with a month lag. India for example (see the RBI monthly data here [table 4] and here). That should be the basic standard for any country that wants a top tier trade agreement with the U.S. Remember, the agreement is about disclosure only—it isn’t a binding commitment not intervene. What has Korea agreed to? A lot less. For the next year or so, it will disclose its net intervention semi-annually, with a quarterly lag. That means data on Korea’s purchases next January this won’t be available until the end of September, and January’s purchases will be aggregated with the purchases and sales of the next five months—blurring any signal.* Korea will start to disclose quarterly with a quarter lag at the end of 2019. But that’s still a long lag. Intervention in January 2020 wouldn’t be disclosed until the end of June 2020. Moreover, quarterly disclosure of net purchases doesn’t provide much information beyond what is already disclosed in the balance of payments (BoP). The BoP shows quarterly reserve growth, which combines intervention and interest income, with a quarter lag. Now is it true that quarterly intervention with a quarter lag was the standard in the TPP side agreement. But the Trump Administration has claimed that TPP was a bad deal, and they would do a better deal. They don’t seem to have gotten that out of Korea. And currency intervention should have been a real focus in the negotiations with Korea. There is no doubt Korea intervenes, at times heavily. And I am confident that the absence of any currency discipline in the original KORUS has had a real impact. Korea, in part through intervention, has kept the won weaker than it was prior to the global crisis. And the won’s weakness in turn helped raise Korea’s auto exports, and thus contributed to the increase in the bilateral deficit that followed KORUS. To be sure, Korea’s German style fiscal policy has also contributed to Korea’s overall surplus. But that isn’t something that realistically can be addressed in a trade deal. Moreover, the failure to get a higher standard than TPP undercuts the Trump administration’s argument for bilateral deals. A big, multi-country deal can in theory be held up by a few reluctant countries. Singapore, for example, has made no secret of its opposition to a high standard for the disclosure of foreign exchange intervention (see end note 4 in the currency chapter of the draft TPP agreement; I assume exceptions to quarterly disclosure didn’t arise by accident). Singapore also discloses comparatively little about the activities of the GIC. And a bilateral deal in theory also could address country-specific currency issues—like the activities of Korea’s large government pension fund. A reminder: Korea’s government-run pension fund is building up massive assets, placing a growing share of those assets abroad and reducing its hedge ratio (it is now at zero, or close to it). This at times has looked a bit like stealth intervention. And it certainly has an impact on Korea’s external balance—structural, unhedged outflows of well over a percentage point of Korea’s GDP have helped Korea to maintain a sizeable current account surplus with less overt intervention. And I worry that the pension fund’s balance sheet will in the future provide Korea with an easy way to skirt the new disclosure standard—particularly if Korea would be at risk of disclosing a level of intervention that might raise concerns about manipulation. Suppose the Bank of Korea bought a bit too much foreign exchange in the first two months of a quarter. The Korean government could encourage the pension fund to buy a couple of billion more in foreign assets in the third month of the quarter, and meet that demand through the sale of foreign exchange from the intervention account. Voila, less disclosed intervention. Remember, sales don’t need to be disclosed separately. A bilateral deal in theory could have included commitments disclose the pension service’s foreign assets, and its net foreign currency position vis-à-vis the won (e.g. its hedges, if any). It thus could have set a standard not just for disclosure of direct intervention, but also for disclosure by sovereign wealth and pension funds. The side agreement on currency with Korea consequently looks to be to be a missed opportunity for sensible tightening of disclosure standards, on an issue that really matters for the trade balance.     Now for some super technical points. The intervention data should not precisely match the reserves data. When Korea buys foreign exchange, it sometimes then swaps the foreign exchange with the domestic banks for won. This lowers the net amount of foreign exchange the central bank ends up directly holding, while creating a future obligation to buy back the dollars swapped for won. This shows up in the central bank’s reported forward book. Total intervention thus may exceed the change in reserves in the balance of payments. However, it can be inferred from the combined increase in reserves and forwards—and Korea currently releases both its forwards monthly and its balance of payments data monthly. As a result its intervention can be inferred from these monthly numbers—quarterly data with a quarter lag will add very little. The buildup of government assets abroad—non-reserve government assets that is—now accounts for a significant share of the net outflow associated with Korea’s current account surplus. And with higher oil prices set to lower Korea’s surplus further, that share will grow. As a chart of the net international investment position shows, the rise in the foreign assets by the National Pension Service now accounts for the bulk of the rise in the total foreign assets of the government of Korea. On a flow basis, outflows from insurers are now more important than the pension outflow—however the insurers, unlike the NPS, supposedly hedge. 3. The increased scrutiny of Korea’s management of the won that has come with the negotiation of the currency chapter—and the risk Korea could be named in the foreign exchange report—has had some positive effects. It didn’t keep Korea from intervening pretty massively to block won appreciation in January at around the 1060 mark (and I suspect Korea has bought at a few other times in the first quarter as well). But it does seem to have encouraged the Koreans to take advantage of dollar rallies to sell won and thus hold their net purchases down. In 2015 and 2016 the Koreans didn’t tend to sell dollars unless the won was approaching 1200 (an extremely weak level). In the past few months they have been selling on occasion at around 1100 (or at least not rolling over some maturing swaps and thus delivering dollars to the market). The won’s trading band has been pretty tight. I just think the block at 1060 should disappear.   */ as I understand it, Korea won’t ever disclose its intervention this January—the first disclosed data will be for the second half of 2018.