During early March, when the current pandemic started to spread globally, Taiwan’s central bank for the first time publicly responded to concerns about its FX derivative interventions raised by our joint work with Brad Setser last year. In this context, it seems appropriate to revisit this matter again. There will be three broad sections:
- The publication of the hitherto only individually published chapters as one comprehensive paper, intended to ease navigation of the argument presented last year.
- A new Addendum chapter was added to the paper and is also printed in full below. It recaps the official response to the paper since initial publication last October through April 2020.
- The actual core of this post, which starts with data updates of the CBC’s and lifers’ activities since last October, before then pondering how the CBC’s new transparency, the richness of TWD offshore and the renewed occurrence of late-day depreciations of the currency may provide insights into the CBC’s policy direction going forward.
If the pdf is not displayed below, please press reload.
ADDENDUM
(APRIL 2020)
The concerns presented in this paper were highlighted in the U.S. Treasury’s delayed autumn FX report released in January 2020, with Treasury reiterating calls for greater transparency regarding Taiwan’s FX intervention disclosures.
In response, the CBC issued a somewhat perplexing message, stating it ”has never used swaps to intervene in the currency market”. This was not a core claim this paper put forth, rather chapter IV. suggests the intervention occurs conventionally in the FX spot market, with the newly accumulated FX reserves removed from the balance sheet afterwards via swap transactions. In either case, the result is the same: FX intervention beyond what is currently visible in the official figures.
Interestingly, Treasury’s call for more transparency seems to have sparked a discussion among Taiwan’s central bankers and academics about the need to align Taiwan’s FX disclosures with international standards.
On March 11, 2020 the CBC, for the first time, released detailed disclosures of its total foreign currency liquidity. As of the end of February, it holds foreign currency liquidity worth USD 617.7bn composed of:
- USD 479.7bn of FX reserves;
- USD 99.1bn of FX swaps, i.e. FX reserves lent against local currency;
- USD 32.8bn of foreign-currency deposits in domestic banks and
- USD 6.1bn of foreign-currency loans to domestic banks.
The close to USD 140bn of newly reported FX exposures validate the reasoning in chapter IV. and the modelling in chapter IV. Considering that banks will in all likelihood use deposits and loans received from the CBC to provide FX swaps to lifers, the CBC ultimately is the largest counterparty to lifers’ FX hedges, accounting for approximately 60% at the end of 2019.
We applaud Taiwan’s decision to align its disclosure standards with international norms. Further we hope this increase in transparency will not be a one-off event but will imply the newly released information to be included in regular monthly data releases going forward. Lastly, to ensure an accurate understanding by international partners of past interventions, a release of historical information on the size of the CBC’s swap book and deposits with banks would surely help.
Given the current global health crisis, it is not the time to strongly adjudicate the implications of Taiwan’s excessive past interventions in FX markets. Time and resources are more urgently needed elsewhere and international cooperation should be the order of the day
Taiwan’s renewed characterization of its transactions in FX swaps as mere liquidity provision and part of standard monetary policy should be continued to be criticized. Provision of dollar liquidity via swaps can play a role in maintaining financial stability, as the CBC has argued, but such efforts need to be disclosed and at no time should the provision of off-balance sheet swaps be allowed to obscure the actual size of the central bank’s intervention in the foreign exchange market.
New disclosures, pricey hedges, late-day jumps & a new policy direction?
With the formalities of the CBC’s public response through April 2020 discussed above, the way is almost clear to delve deeper into what has happened in Taiwan’s FX market since last summer. This will include a quick detour, highlighting a higher frequency measure of its swap book the CBC now discloses, before returning to update the datasets and modelling created in the paper, including the models tracking the CBC’s FX exposures and lifers’ activity.
With this as baseline, two rather difficult problems are addressed. Firstly, the significant increase in costs of hedging FX risk via non-deliverable forwards, which, in somewhat complicated ways, may provide the best perspective on the CBC’s future management of its currency. Secondly, the curious phenomenon of end-of-day depreciations of the TWD.
The resultant picture is still somewhat fuzzy, yet there are indications that Taiwan’s authorities are confronting the difficult task of appreciating its undervalued currency, while at the same time diligently dealing with the financial stability risks exerted by FX exposures assumed by the private sector in recent years.
High frequency swap book disclosures
After the CBC’s initial disclosures in March, which referenced end of February values, the CBC in June published its Annual Report for 2019. Included in the ‘Foreign Exchange Management’ section is the same set of information as provided in March, this time for year-end 2019.
In addition, the CBC now also releases monthly numbers on its swap exposures on its website. Like the initial disclosure, this is again a welcome development and largely solves the transparency gap relative to other nations. However, information on the CBC’s USD deposits or loans with local banks appears not to be part of this release. This omission thwarts confident high frequency assessments of its FX interventions as, via such undisclosed deposits, it remains possible for the CBC to provide FX swaps to lifers indirectly. This would occur by domestic banks on the receiving end of such deposits using these funds to deliver USD to lifers via FX swaps. This mechanism would be indistinguishable from cases where the CBC enters FX swaps itself, in terms of the opacity of its true FX interventions, as well as the amounts of FX hedges received by lifers ultimately attributable to the CBC.
To date, the CBC’s swap exposures are broadcast separately from its FX reserve statements. While not absolutely essential, it would appear sensible to package these numbers into a single release. The IMF’s IRFCL framework could serve as an already widely accepted template, with the added benefit of providing guidance on how to account for foreign currency deposits central banks may hold with their local banking system.
CBC’s modelled FX exposures since mid-2018
The original paper contained data through mid-2019, which allowed modelling of exposures up to summer 2018. Now almost a year on, these estimates can also be rolled forward by an equivalent timeframe. Going forward, the CBC’s official disclosures clearly receive preferential status over model estimates, yet these may remain useful in providing impartial confirmation.
The tracking journey again begins with the CBC’s monthly balance sheet, particularly its ‘Net other assets’. Despite its new disclosures, the CBC continues to book its FX PnL from FX reserves and swap exposures in this contra-account, with the more reasonable equity series continuing its repeated intra-year rise and year-end fall.
Plugging these values into the ‘USD-EUR’ model illustrates that these two currencies remain the CBC’s core PnL drivers, with r2 exceeding 0.95 as of the latest measurements. USD coefficients (calculated in TWD bn) have remained fairly flat. EUR coefficients in contrast have increased considerably, tripling since their trough in 2019. Despite a concurrent increase in the size of confidence intervals, it appears the CBC has reallocated1 portions of its FX reserves into EUR (or currencies correlated therewith) during 2018 and early 2019.
Currencies other than USD & EUR, after correlated effects are stripped off, are the final step required to conclude the modelling. The latest value is for May 2019, coming in at USD 629bn. Official datapoints for the CBC’s total FX exposures are also shown, the first comprehensive one of which is for year-end 2019, taken from the CBC’s annual report. These total USD 621.7. Combined, these observations indicate the CBC’s total FX exposures reached a plateau in early 2019, moving sideways through last year, before growth continued owing to FX reserve accumulation this year. The stability of the CBC’s swap book and local USD deposits is confirmed by the demand dynamics of lifers discussed in the next section.
Lifer developments
Lifers sold policies worth USD 92bn during the past year ending May 2020, with total assets at USD 996bn now just shy of the USD 1tn mark. Foreign assets held increased by USD 72bn, a portion of which stems from valuation gains due to decline of yields at the long end of the corporate curve last summer. The larger share however was the result of fresh direct cross-border purchases, outright purchases & rollover of Formosa bonds and domestically listed, USD-purchasing bond ETFs. These in total accounted for around USD 50bn. Sales of policies in 2020 were somewhat slower than normal, in part explained by heavy selling of Taiwanese equities by foreigners during the first four months. This repatriation of funds formed the counterbalance to the still ample current account surplus, thus relieving pressure on locals to allocate foreign currency abroad.
In terms of the distribution of FX risk assumed, lifers open FX position (as a share of total FX assets) saw fresh heights at the end of 2019, closing the year at 24.4%, before declining marginally to 23.6% as of Q1 2020. In nominal terms, this translates into FX exposures worth USD 145bn at the end of May 2020, or 2.3 times the size of industry-wide equity book value.
With the TWD’s steady rise since last summer, FX losses inflicted on life insurers are already constituting a sizeable drag on profitability. From September 2019 through June 2020, the currency appreciated by 5.5% against USD, causing losses (assuming full mark-to market accounting) of around USD 7bn across the industry, or 11% of year-end 2019 book value.
Lifers post an ROE of between 8-10% per annum before accounting for FX gains or losses. Again assuming full reflection of FX impacts in financial accounts, FX losses consume the largest share of profits related to regular insurance activities. That the relatively modest FX appreciation so far is already producing substantive losses is testament to the extent of private sector FX risk taking and highlights the difficulties of letting the currency freely float towards fair value.
Turning to other ways of FX risk layoff, the share of FX policies reached a new high point this year as well, relieving lifers of 27% of total FX risk taken. Due to the relative growth of lifers’ own risk taking and FX policy sales, the share of classical derivative hedges declined to a decade low of 49%. Nevertheless, the continued increase in industry size has kept the nominal size of FX hedges entered at its high plateau of USD 270 – 300bn, first reached in 2018.
Pricey NDF hedges, onshore swap restrictions & deliberate limits to arbitrage
Perhaps the most striking change in Taiwan’s FX space since last summer is the tremendous widening in the offshore x-ccy basis, turning hedging via NDFs into a very costly endeavor.
Stating this fact is trivial and requires not much more than a look at a screen. Explaining why this phenomenon is occurring is already harder, though (to a large degree) possible. The key next question—why authorities encourage this present situation—is much tougher and necessarily involves speculation; it may though provide a glimpse into the future of the CBC’s exchange rate management. This section will work towards addressing this last question, which beforehand requires examining what exactly is happening and why regular arbitrage forces are not working as they are supposed to.
The backstory & a spike
From 2013 through mid-2019, non-deliverable forward USD/TWD contracts had traded roughly in line (though remained more volatile) with onshore forwards, at an annualized negative x-ccy basis of around 100bps at the 3m tenor for both. Since then, the offshore TWD has richened significantly vs. onshore markets, thus reaching negative basis values exceeding -400bps, last seen during 2009/10.
Back then, FX exposures by Taiwanese institutions looked quite different. Lifers’ step-up in acquisitions of foreign debt was only in its infancy, implying much lower hedging needs overall, as well as little reliance on the offshore market given persistently higher and more volatile hedging costs during the noughties. Further, NDF markets were purely the domain of global banks, with Taiwanese banks restricted by regulations to conduct NDF business with domestic clients. That changed in 2014, with overseas branches of Taiwanese banks allowed to trade NDFs with domestic clients. Over time, lifers began to increase NDF hedging, particularly at times when more attractive than onshore markets, such as in late 2015 or early 2019.
NDF use across lifers is diverse, ranging from close to zero (mostly lifers part of a holding company with a bank subsidiary, i.e. with direct access to the CBC’s swap book) to up to 40% for stand-alone institutions. Industry-wide averages are hard to assess confidently given varying levels of disclosure. As of summer 2019, NDF hedging likely accounted for ~20% (~USD 50bn) of all classical FX hedges entered by lifers.
Financial account restrictions & NDF markets
The divergence in the on- and offshore basis since last year raises fundamental questions about financial account restrictions in force and the CBC’s attempts to seal off its domestic market.
Non-deliverable markets usually do not exist in countries with a fully liberalized financial account: users, whether for real-money or speculative purposes, can freely trade forwards and swaps in the desired locality itself, or rely on intermediation by global banks, which in turn possess unrestricted access to the end market in question. In cases such as Taiwan, which seek to restrict ‘speculative activities’ in order to dampen FX volatility, trading of certain instruments onshore is restricted to specific actors or purposes. This is enforced by requiring detailed documentation of trading activities as well as the need for an initial sign off by authorities to be granted the right to conduct business in a certain market segment. To stand a chance of being effective, this usually requires restrictions on both FX spot and derivative markets, and thus possibly limits arbitrage activities between market segments.
Even in cases where there exists detailed public knowledge of the set of financial account restrictions in place, their practical effectiveness (or porosity) is oftentimes only truly discernible by observing pricing differences in on- and offshore markets over a period of time.
The past half-decade has given the impression that Taiwan was on its way to reach a balance between demand and supply in both markets segments, given their similar pricing. This year’s price action upsets such thinking and begs the question why such large pricing differences can persist.
How arbitrage works … and when it does not
Perhaps the easiest way to approach this question is to consider which actors could take actions to align prices and, more importantly, what hinders them currently in Taiwan.
The first group are ‘pure financial arbitrageurs’, whose only incentive is to profit from price discrepancies. Such actors would include global banks and hedge funds which would, as the most immediate and solely derivative arbitrage, sell expensive TWD via NDFs and acquire TWD in the onshore forward market. A more balance sheet-intensive trade of the same flavor would again sell TWD against USD via NDFs, but create the offset position on-balance sheet, by translating USD to TWD and holding these funds in Taiwan for the term of the trade.
The first type of trade needs to be (close to) impossible by definition, otherwise the necessity for a split between onshore forwards and NDFs would be superfluous2. In practice, this is enforced by heavy restrictions on onshore use of FX derivatives by foreigners for speculative and arbitrage purposes. Even foreign investment funds hedging the currency exposure of (by authorities welcomed) equity holdings in Taiwan are usually only able to do so via NDFs.3
The second form of arbitrage (i.e. involving balance sheet space) is severely hindered by restrictions on foreigners’ deployment of TWD funds within Taiwan, where money market type investments (including nearly all bond types) are capped at 30% of all inward remittances by foreigners. Since some form of risk-less position4 within Taiwan is required to counterbalance the TWD sale via NDFs, such restrictions meaningfully impair the feasibility of convergence trades between NDF and onshore markets.
Taiwanese banks find themselves in a similar situation as the players just discussed. As these are however essential to Taiwan’s economy, outright bans on certain instruments are unsuitable. Instead, the size of such banks’ onshore derivative hedges5 to offset NDF exposures is limited to 20% of the ‘authorized foreign exchange position’. This supposedly refers to a bank’s ‘net FX position’, which is already rather small given the leverage characteristics of banking institutions. Once a bank reaches its allowed position limit, its ability to help engineer balance between on- and offshore segments by pure arbitrage is exhausted.
‘Real economy’ arbitrageurs form another category which could engage in activities leading to convergence. These are typically debt issuers, which opportunistically switch funding currencies, but ultimately swap back into their home currency. In Taiwan’s case, this could be corporations issuing USD bonds, converting these funds into TWD and hedging the FX risk by shorting the TWD in the rich NDF market. There are numerous reasons why such actors do in reality not have the clout to play any larger role. Taiwan’s corporates have essentially no USD bonds outstanding, complicating pushing large amounts of volume onto, to date, uninitiated investors. The number of corporations able to price multi-billion dollar deals is limited. The NDF market is almost entirely focused on the front-end (≤ 1y in tenor), so that term hedging USD issuance as in a cross-currency swap seems virtually impossible.
Lifers: the basis driver & tacit regulations
The only major player left to examine is the most crucial one here: lifers themselves. In order to maximize profits, insures should over time gravitate towards the cheapest type of FX hedge. Currently this would imply rolling all expired NDF hedges over into the onshore market, where pricing is much more favorable. In time, as the demand for offshore hedges drops, pricing should converge again.
Opinions might be split about lifers’ (over-)reliance on overseas markets, but there is no doubt insurers are savvy enough to choose their hedges appropriately. This in mind, the progressively more negative offshore basis is already somewhat suspect, as lifers’ retreat should, at a minimum, cushion any outside basis-widening forces.
High frequency information on lifers’ NDF activity is again an item in the ‘hard-to-obtain’ basket. One of the best indications is the difference in hedges reported by lifers themselves and a monthly number published by the CBC. As described in chapter II. of the paper, it appears the CBC definition does not include offshore hedges, while the lifer disclosures do. In the past, the gap between these series was in line with periodic indications of individual lifers. Since last summer, the gap between these measures has widened by almost another 5 percentage points, while lifers’ overall reported hedge ratio has stayed relatively put at around 50%.
On the presumption that this interpretation of the gap is correct, lifers have established NDF hedges worth about USD 30bn since last summer. These account for practically all new FX hedges put on during this timeframe, despite the vastly higher costs offshore hedges currently entail relative to their onshore siblings.
Further evidence in support of this thesis comes from TWD NDF trades reported by swap execution facilities. These only cover a (typically representative) portion of global NDF volumes, i.e. transactions with at least one counterparty subject to U.S. jurisdiction. The volume there observed is in line, if not marginally higher, than early 2019 levels. Granted, there exists NDF activity outside of lifers, yet these are the key players and directly (or indirectly, as intermediaries trade among each other to ultimately supply hedges) responsible for the largest trading volume. Their withdrawal for the onshore market should thus be visible; such a decline not being evident strengthens the plausibility of the prior evidence.
The presented argument quite certainly establishes that lifers have acted contrary to their supposed self-interest, raising the question what sort of regulatory constraints are in force to cause such behavior.
One possibility is that the existing set of rules just occasionally—more or less accidentally—forces lifers to act against their best interests. For instance, it has been suggested lifers, after an initial bond purchase, must designate whether they want to classify such a new acquisitions as FX-hedged, in which case the right for on-shore hedging is granted. If non-hedged is selected, bond can still be hedged dynamically, but only offshore. So it may seem plausible that previous offshore hedges (or unhedged exposures) are harder to transform into onshore hedges.
Should this be the case though, the vast differences in pricing on- and offshore should provide a strong incentive to use every possible measure to facilitate onshore hedging. Most simply, this could include selling bonds hedged via NDFs prematurely, repurchasing them and then re-classify as onshore hedged.
Similarly, Taipei-listed ETFs investing in USD-denominated bonds may not—should they want to FX hedge—have in all circumstances full flexibility in their choice of FX hedging device, with regulators potentially requiring extra time to approve onshore swap quotas.
In either case, if there was a joint will by regulators and lifers/investors, these hurdles to onshore hedging would seem surmountable.
The only explanation that remains is that the CBC deliberately restricts legitimate (i.e. non speculative) domestic investors access to onshore FX derivative markets. Last month, this interpretation of why the offshore basis remains deeply negative was confirmed by excellent reporting in Taiwan’s United Daily News newspaper. Life insurers supposedly enlisted the life insurance association for help in communicating their request to the CBC, asking to be granted larger quotas for hedging FX in the onshore swap market. The Financial Supervisory Commission (FSC) in turn indicated it will assess such proposals and raise these with the central bank if found reasonable. The head of the FSC recently confirmed these events in another interview.
Exploring the CBC’s considerations for imposing onshore quotas
It is possible that this FSC-intermediated discussion between lifers and the CBC will result in higher quotas, based on the assumption that the CBC was ‘just not sufficiently aware‘ of lifers’ hedging difficulties. Lifers’ outreach, following this train of thought, should then provide enough impetus for the central bank to take remedial measures.
Such an assumption, however, would seem rather naïve in light of the CBC’s past dexterity in its handling of its FX swap exposures.
As a review, the CBC is the key supplier of FX hedges in the onshore swap market, either directly or via FX deposits with local banks (which allows these to swap-lent USD to lifers), accounting for more than 50% of hedges provided to lifers. The large USD deposit base of households and corporations allows Taiwanese banks to provide the remaining hedges, with some non-financial corporations and foreign institutions also playing a (smaller) role. The major influence of the CBC is also seen in the behavior of the onshore x-ccy basis. It did also widen during the turbulent weeks in March, but remained at acceptable levels throughout, simply because the key hedge provider was never at risk of pulling back.
The fact that the CBC has in the past almost willingly supplied whatever quantities requested by lifers (and thereby kept onshore hedging costs bounded) should not obscure the fact that the CBC is the ultimate pricing backstop of the market. That it historically has not relied on price signals explicitly to balance the market should be seen as deliberate rather than accidental. In fact, in its annual reports the CBC always states that a reason for its swap exposures is the provision of reasonably priced hedges for lifers. This in mind, it seems extremely unlikely that the CBC is oblivious to lifers needs.
What may then account for the CBC’s reluctance to grant further onshore hedging, and is it possible to construct a reasonable framework consistent with its actions?
Viewed from a very high altitude, a logically acting country’s decision to subscribe—for prolonged6 periods of time—to a neo-mercantilist growth model relies on two implicit (and hence often unstated) assumptions:
- The belief that more economic value can be attained in the long run if a (large) share of domestic production is not used for domestic purposes (i.e. consumption or investment to increase production capacity) now, but rather is exported and exchanged for foreign financial assets. An undervalued exchange rate is one mechanism by which policy makers can implement such policies, keeping exporters extraordinarily competitive, while lowering consumers purchasing power abroad.
By axiom, a central bank actively maintaining an undervalued currency is willingly incurring latent FX losses, when the currency appreciates in the (sometimes distant) future. Still, such a choice can make sense in situations when authorities can correctly weigh future returns on financial investments abroad against the value of current consumption or the efficiency (or lack thereof) of domestic investment.
The classic example, which Taiwan matches to some degree, is a demographically driven excess of production capacity over desired domestic consumption. If potential domestic investment projects are deemed inadequate and the return on foreign assets is sufficiently high, then a central bank keeping its exchange rate undervalued may be the optimal decision.
- Weighed against such calculations, which in reality are met much less often than actual FX interventions might suggest7, must be the international response. Prolonged currency undervaluation may rightly be construed as unfair competitive advantage, gained at the expense of other nations.
It seems appropriate to classify Taiwan as having decided in favor of both of these above points during the past 20 years. Cumulative FX intervention today stands at USD ~630bn (or more than 100% of GDP), an amount far in excess of what could be ascribed to safety effects FX reserves can provide. Such intervention does not occur automatically, so that a cost-benefit analysis (not unlike that sketched out above) should be assumed.
As to the second criterion, Taiwan has in the past averted difficult conversations by not disclosing its swap-related FX interventions. Further helping was the increased FX risk taking by Taiwanese households, corporations and lifers during recent years, which eased the extent of FX intervention the central bank had to carry out.
The increased transparency around its FX swap exposures clearly alters calculations around FX interventions previously hidden by swap transactions. A return to the up to USD 50bn annual run rate in swap increases seen at times during the past decade seem hard to imagine. Further, the disclosures equalize the accounting treatment of carrying FX assets on its books vs. being swapped with lifers; either method now is disclosed, thus eliminating the concealment benefit swaps provided historically.
Changes in Taiwan’s policy makers’ views concerning the attractiveness of mercantilist policies is harder to disentangle, nonetheless a couple of observations can be offered.
- The TWD’s strengthening since last summer shows the CBC is not willing to employ its intervention tools to the maximum degree possible, even despite the FX losses lifers will incur as consequence.
- The governments passing of a repatriation bill in July 2019, incentivizing domestic investment by providing generous conditions for returning funds from abroad, runs counter to the playbook of keeping FX undervalued to disincentivize local consumption or investment. The implied consequence being that returns on domestic investment are now deemed higher than in the past.
- Demography is playing some role in explaining Taiwan’s excess production capacity (and hence savings), which in turn allows the CBC to affect its allocation by way of its exchange rate management. In the coming years, Taiwan will slowly exit the demographic sweet spot, with retirements increasing relative to the share of the high savings age cohorts. This would suggest a more balanced trade account, a more market determined and higher trading exchange rate, in order to maximize the purchasing power of consumers. A withdrawal of CBC intervention keeping TWD undervalued may thus only be a matter of time. As consequence, front-loading some appreciation to discourage further build-up of FX imbalances by the private sector would appear reasonable.
- The risk-free interest rate differential relative to the U.S. has collapsed, lowering the respective return on recycled current accounts surpluses. Spread markets still offer ex-ante attractive returns; the COVID-related uncertainties though exceed those stemming from mere regular recessions. Whether current spreads thus provide sufficient compensation for increased downgrades and defaults remains to be seen.
- The events in Hong Kong this year, and the generally more assertive mainland China, poses questions about how Taiwan might best ensure its independence from an economic perspective. A nation endowed with ample liquid funds (in the form of FX reserves) presents, all else equal, a more attractive target for a hostile M&A takeover. At the same time, mercantilist policies will not help in securing Western support. In contrast, increasing production capacity and bolstering its place in supply chains Western partners rely on seems more promising.
Devising a currency appreciation path: FX risk & lifers’ business model
Should the reading of these pointers be indeed correct, the next question is how to design the transition from the current ‘foreign asset gathering’ model to a more balanced state. The long-standing reliance on the present model renders this trickier than starting from scratch. Two of the central challenges to be considered follow, in particular the way in which they may be connected to the onshore FX hedging restrictions.
- Slowly severing the symbiotic role lifers have played in channeling savings abroad. In lifers, the CBC had a natural counterparty to its swap exposures, thus allowing it to keep parts of if its FX interventions undisclosed. With this relationship superfluous now, not to mention the financial stability risks emanating from lifer-related FX exposures, it appears their macroeconomic centrality is set to decline.
Lifers ascent also formed the natural counterpart to large current account surpluses. It was and remains the easiest vehicle for the broader population to access foreign fixed income markets. Households’ decision to acquire policies and lifers’ decision to purchase USD-denominated debt securities were not directly forced by policy makers, yet the lack of domestic alternatives and actual (as well as projected future) provision of relatively cheap FX hedges made this choice quite inevitable. Ways to limit further growth of the lifer-recycling model range, in theory, from imposing hard caps on the size of the industry on one hand, to mere reliance on verbal encouragement not to acquire lifer products investing abroad on the other. These respective ends of the spectrum seem either too draconian, undermining free market forces, or too soft, while also revealing the intentions of policy makers.
A better way seems to be to use the variable directly under the CBC’s control: the pricing and volume of onshore FX hedges. Lifers’ business model heavily relies on these hedges for keeping overall hedging costs bounded. Further, open FX positions lifers presently hold were easier to justify historically on the assumption that the central bank would remain generous in their hedge provision in the future. In consequence, higher, more market-like hedging costs would lower the attractiveness of channeling funds abroad, all without the imposition of explicit regulations. One way to reach this goal would be a slow retreat by the CBC from its provision of FX swaps, conferring more and more weight to the marginal private sector hedge providers.
A negative side effect of this method though is the indiscriminate rise in hedging costs: already existing positions are equally penalized as new hedges. Thus, increasing demand for onshore swaps by a small hand of actors could result in higher hedging costs even for institutions which only acquired FX-hedged debt in the past, while officially sanctioned by the CBC.
The solution to this inconvenience leads back to the restrictions of onshore FX hedging. This allows the separate pricing of existing and new hedges: the former continuing at present prices directly with the CBC, the latter only possible in the offshore market, where regulations to arbitrage practically ensure their costs to be substantially higher, thus disincentivizing continuation of lifers’ recycling of current account surpluses. - Concerning the financial stability risks stemming from private sector FX risk taking, any currency appreciation should be engineered to occur in an orderly fashion, so not as to invite unnecessary positive feedback loops expediting appreciation. There are three sources of private FX risk taking: lifers outright assumption of FX exposures, USD policies sold by lifers and USD deposits by corporations & households, which as of May account for USD 145, 165 and 230bn respectively, or ~90% of GDP.
All of these categories would be negatively affected by FX appreciation, with lifers’ exposures clearly the most vulnerable due to the limited equity cushion. As shown earlier, even the slow glide path higher during the past year caused meaningful FX losses. Any quicker move upwards would not only consume insurance-related profits, but eat straight into equity, with recapitalization risk looming at the horizon. By keeping currency appreciation modest, but stretched over a longer timeframe, FX losses can be balanced by insurance profits, soothing solvency concerns.
Pricing TWD NDF contracts in light of onshore swap restrictions
The last question in this section concerns the level at which the offshore TWD market will equilibrate and a conceptual mapping of the demand and supply side. This question is quite separate from the prior speculation of CBC motives—it merely assumes restrictions are in place.
The demand side, going long TWD is clear: lifers themselves and, to a much lesser degree, FX-hedged debt ETFs acting as conduit for them. The motivation, earning returns above what is on offer in domestic markets, is also self-explanatory. A detailed guide for how to accurately assess FX-hedged debt investment can be found in this prior post, and filling in the relevant numbers for Taiwan is trivial.
As of July 2020, the ACM and Kim & Wright term premia models for the U.S. sovereign curve both estimate the excess return for assuming duration risk at the 10y tenor at -90bps. 30y USD corporate credit of A-BBB split rating quality—the core of lifers’ holdings—trades at a spread of about 180 bps above Treasuries. The offshore x-ccy basis at 3m and 12m tenors is -460 and -360 bps respectively. Unless lifers hold extremely out of consensus views about the possibility of negative interest rates and/or the default risk credit markets currently price, it must be assumed they implicitly hope for a normalization of hedging costs in the long term. Holding a negative-carry position over a year can be done; doing so over several years is close to impossible.
A normalization in hedging costs is hard to envisage currently given the regulatory restraining of arbitrage forces and the CBC’s restriction on onshore hedging. Unless authorities change tack on these fronts, lifers appear stuck with hedges so expensive that already low risk premia positions turn outright uneconomic. Should this state of affairs continue, over time, cutting of offshore hedges (and the associated uneconomic exposures) would be expected, thereby relieving pressure on the basis and laying the groundwork for a move to more reasonable values.
Who will, in the meantime, supply lifers hedges via NDFs. The difficult-to-implement and traditional long cash, short forwards arbitrage was discussed before. With it unavailable to most, the basket of potential arbitrageurs must be broadened.
For instance, given the relative ease of accessing Taiwanese stocks, foreign equity investors are offered the extra payment the negative x-ccy basis presents on top of any returns earned by FX-hedged investments on the TWSE. Anecdotally, double digit negative basis point values appear insufficient to generate much foreign interest; values of 3% and above (as currently the case) realistically might cause some reallocations and should provide an outer bound. Of course, any such equity inflows will put upward pressure on the exchange rate and might necessitate targeted interventions by the CBC, even if it might agree with it directionally.
Alternatively, and if arbitrage flows including equities are insufficient, something interesting happens: the pricing of non-deliverable forwards departs from the traditional framework based on interest rate differentials plus a cross-currency basis (which reflects balance sheet costs of arbitrageurs as well as the (un-) ease of putting on such trades) to a setting based instead on the expected future exchange rate. Normally, the effect of any directional outright trading in FX forwards is quickly neutralized by arbitrage forces. With this not possible here, and markets perhaps reasoning in favor of a stronger TWD for reasons stated above, the negative x-ccy basis is repurposed as reflecting expected currency appreciation pressure beyond those implied by interest rate differentials.
Importantly, in this framework, the future price of currency is no longer governed by arbitrage from relative value trading, but subject to the whims of not necessarily risk-neutral actors with time varying risk preferences. Higher volatility in the basis would be expected in such a case.
The curious case of late-day TWD depreciation
Taiwan’s FX market provides another puzzle: prolonged periods of time during which spot TWD weakens consistently during the last hour of Taipei trading. Market efficiency should normally take care of such overly simplistic anomalies, and that it seemingly does not here poses again the question of what is going on. It might not be possible to answer this question satisfactorily, but the same inquisitive path as with the hedging question above can provide partial insight. So, the steps ahead are once more describing what is going on data-wise, surveying (fortunately numerous) explanations reported over the years, to then lastly scrutinize possible underlying motivations.
Varying degrees of jumps & the importance of the price source
FX markets are 24-hour markets, and so in theory trading is possible at any time. But as with stock markets, which can also be traded outside official market hours, most trading is focused on the business hours of the geographic area serving as host country for a given currency.
This is the case in Taiwan, where the morning session commences in earnest at 09:00 local time (GMT+8), with high volume trading for about two hours, followed by a slowing of activity into 12:00, when a two-hour lunch break kicks in. The resumption of trading at 14:00 often sees the highest relative activity, with stable volume into the official close of onshore trading at 16:00. Thereafter, trading moves on westwards. Interbank volumes tend to decline continually, until New York awakens and lifts trading activity for another two to three hours. Volumes though remain clearly below the heights reached during Taiwan business hours, before then dropping off steeply to await another Taipei morning.
There are two main venues for onshore FX trading in Taiwan, Taipei Forex Inc. and Cosmos Foreign Exchange International Co. Ltd. The former of these conducts the larger share of trading, is usually the entry point for the CBC when conducting FX interventions and its intra-day fixing reports every 15 minutes (from 09:15 through 16:00) are seen as the clearest indication of current market pricing. The 16:00 close indication Taipei Forex Inc. provides also serves as the daily exchange rate displayed on the CBC’s website.
Curiously enough, this most esteemed of pricing sources is subject to the greatest amount of late-day shenanigans. The phenomenon itself is very quickly explained: During some, usually long consecutive stretches of time, the TWD depreciates meaningfully during the last hour of onshore trading, with the largest portion occurring in the final minutes. This leads to TWD closing unusually often at intraday lows. The depreciation always proves temporary though, with trading after 16:00 occurring at pre-jump levels. This phenomenon dates back into the mid-2000s, was most pronounced during the 2015-16 period and has resurfaced since late 2019.
The one-month time period from mid-June through mid-July 2020 is representative of times when the jump phenomenon occurred and shall thus serve as example here.
In terms of pricing sources, Taipei Forex Inc. is not the only one available, and broadening the horizon really pays off in this case. Figure 7 shows a violin plot of three pricing sources for the last hour of onshore trading during the specified one-month period.
Taipei Forex Inc. on the left shows USD strengthening during every single observation and its magnitude, at on average of 0.4%, is substantial. The middle shows Bloomberg’s Generic Composite rate (BGN). It too sees statistically significant depreciation of the TWD during the last hour of trading, but its magnitude is only about half that of Taipei Forex Inc. Data for the right-most violin is sourced from a stream of executable dealer-to-client quotes aggregated from a number of European & US banks. The jump phenomenon is entirely absent in this, for actual clients, most representative price source. Its zero mean and normally behaving higher moments make it the only section reflecting characteristics of an efficiently priced market.
Deliverable as well as non-deliverable forwards across tenors follow the path of the last, executable price source, with no jumps visible during the last hour of onshore trading.
More could be said right now about the likely differences in these pricing sources; the excellent reporting over the years, recited in the next section, is so vivid though that little further comment is necessary.
A survey of reporting on late-day jumps during the past decade
“The way the central bank intervenes is quite unique. It buys dollars in massive quantities from state-owned banks during the last 15 minutes of the onshore five-hour trading session. It typically makes only one or two trades, but their size is large enough to bring the value of the NTD against the U.S. dollar back down to a level somewhere very near to the previous day’s close.”
“The tricky part is that investors literally can’t sell the U.S. dollar at those “closing” prices. The banks that are in on the trade with the central bank don’t pass the prices through to the market because market expectations – as reflected in the offshore nondeliverable forwards market – are that the NTD will continue appreciate.”
“In short, the central bank’s high jinks have managed to slow the headline rate of appreciation, but the closing prices are in effect fictional.“
Aries Poon in the Wall Street Jounral, Nov 15, 2010
“In the last half-hour of currency trading almost every day in Taiwan, state-backed lenders have the market largely to themselves and set the closing exchange rate at a level determined by the central bank.”
“The abrupt declines are part of the central bank’s arsenal of unofficial tools to keep currency speculators at bay and support exporters. The authority manipulates the opening and closing prices, discourages trading in the final half hour and limits daily remittances, according to traders.”
“The traders said CBC told them in July 2013 to stop taking client orders to buy the currency after 3:30 p.m., a policy that has remained in place since. The authority engineers sudden declines in the spot rate by having two of the eight state-backed banks conduct trades under its instruction, they said.”
“Just one out of six traders said they used the closing price on Taipei Forex to record their own mark-to-market positions.”
Justina Lee and Argin Chang at Bloomberg, Jul 01, 2015
“The Taiwan dollar has erased gains with remarkable regularity in afternoon trading over the past weeks, but the predictable moves remain off-limits to traders.”
“Despite its predictability, the late afternoon moves remain inaccessible as a bet for traders as Taiwan’s central bank strongly discourages them from buying or selling the U.S. dollar after 3:30 p.m., according to two traders who asked not to be named as they aren’t authorized to comment publicly on the market. The two traders said the monetary authority is dictating the closing price.”
“[…] the central bank responded last week that it merely “reminded” banks to finish trading before the last half hour of the session to avoid low liquidity after retail banks had closed for the day and to maintain stable foreign exchange.”
Cindy Wang and Argin Chang at Bloomberg, Dec 30, 2019
“Taiwan’s central bank appears to be taking steps to rein in the Taiwan dollar toward the end of the local trading day, as the currency tests the limits of policy makers’ comfort zone.”
“Currency traders attributed the daily retreats to the actions of the central bank. They asked not to be identified because they are not authorized to comment publicly on foreign exchange markets.”
Argin Chang, Raymond Wu and Samson Ellis at Bloomberg, Jun 16, 2020
The CBC, in setting its desired closing exchange rate, seems to rely on one of the oldest tricks in the business: restricting the market access of most players and printing a small number of then high-impact trades.
This explanation fits in well with the varying degrees of late-day jumps the different pricing sources show. Taipei Forex Inc, as the venue where actual intervention occurs and which serves as the CBC’s published end of day value, is clearly most affected. Its closing value is the variable directly targeted and seems to be set by the exchange rate of the small number of trades between the CBC and banks, or alternatively by transactions between banks which received instructions from the CBC.
Bloomberg’s BGN stream is very unlikely to cover these actual transactions with central bank involvement, given its goal of providing indicative and executable bid and ask quotes. Still, it captures some of the upward move, possibly the result of quotes banks involved in the CBC’s machinations display to each other. The relative seclusion of Taiwan’s onshore FX markets may complicate outsider’s accurate portrayal of market activity. Perhaps for this reason, Bloomberg, while it calculates TWD BFIX rates, does not publish USD/TWD fixes on its public website, indicative of a lack of outside interest in another benchmarking service for this cross.
The fact that executable spot prices for clients, as well as the forward complex, is not affected by the late-day phenomenon is also readily explained. They are not directly targeted and with everyone aware of the impermanence of the 16:00 print, execution for regular market participants is only possible at the pre-jump levels.
Examining motivations: Inertia…
Why does the CBC at a minimum endorse, but more likely create the late-day jumps? As any price-conscious investor—and central bank—knows, the price impact costs of larger transactions can be meaningfully reduced if spread out over time.
So a first question would be how much trading volume truly underlies the last minute trading between the CBC and banks when setting the closing price. It is a somewhat common (mis-)perception that great volume is necessary for large price moves. The restriction of access by other market participants makes it possible though that only small amounts of trading might be involved in the price setting of the close. Thus, it is in theory possible that the setting of the exchange rate is uncorrelated to the degree of total FX intervention, which can occur at anytime during the day8.
For instance, late-day depreciations were quite large during 2015 and early 2016 (averaging almost 1% during autumn 2015) , a time during which FX reserve growth was much less swift than during 2009 or 2010, which saw practically no late-day depreciations. The same applies to the 2016-2018 period, which saw slightly faster growth in FX reserves and appears to have been the sweet spot for the CBC’s then ‘hidden’ FX interventions via its swap book9. However, late-day depreciations then were of far smaller magnitude than during 2015 or 2016.
In terms of pricing power, even if the central bank acts only during the last hour of trading, the power dynamic is less one-sided than it may perhaps appear. Yes, if the central bank wishes to acquire large amounts of FX at times the private sector is also keen to hold on to it, large market swings would ensue. However, central bank purchases of foreign currency usually occur when the local currency is subject to upward pressure, meaning insufficient private sector supply of local currency did materialize. Banks which intermediate order imbalances thus tend to be left long FX at the end of the trading day, with the central bank the only counterparty able to square their FX exposures. In consequence, the image of central banks as pure price taker in late-day transactions appears incomplete.
To return to the initial thought, if the CBC would indeed consciously overpay this much in its FX interventions (perhaps out of intertia, so as to only intervene once a day), what purpose would this serve? Would banks really deserve such an extra subsidy for a quite pedestrian service, and could this overspending by a public body be justified to the population at large?
…signaling…
Given the inconsistencies in this ‘pure intervention’ theory of late-day jumps, what other factors could impel the CBC to nonetheless embrace these? If the volume involved in the phenomenon is in reality not necessarily a reflection of actual interventions, perhaps it is the lower TWD print itself that is of interest to authorities.
A possible answer in this direction is mentioned in the first article referenced above and also in an earlier Treasury FX report: signaling. As a show of force, in setting the end-of-day exchange rate, the CBC supposedly signals to the market its displeasure with appreciation trends. Thereby it supposedly invites other parties to assume the FX risk instead, based on the expectation that the CBC would curtail further appreciation anyway.
It is true that sentiment plays a much larger role in financial markets than the rigidity of many textbooks might suggest. Nevertheless, the above argument seems not quite coherent. Yes, central banks can indeed mightily influence the trajectory of markets, but whether a frequently occurring swift up-down reversal in the USD/TWD rate (which is not even accessible to virtually all market participants) can serve as true signal of some policy intention seems questionable.
…reporting dates…
Another possibility—quite familiar from other markets—is volatility ahead of financial reporting dates. Financial institutions typically have quarterly public disclosure requirements, with the monthly timeframe also important for macroeconomic and certain regulatory accounting purposes. In a benign sense, firms may engage in window-dressing activates ahead of such dates, which in itself may give rise to anomalies ahead of these periods. Less benign, some firms (or authorities) may wish to influence market pricing to achieve a more favorable reporting basis. Since both Taiwan’s private and official sector own large amounts of foreign assets, there may be an incentive to increase their valuation in local currency via targeted depreciations against foreign currencies ahead of reporting dates.
Empirically, such considerations are not corroborated by the data. Quarter- and month-end datapoints show no more tendency of jumps than the days preceding or following these. The same applies to any seasonality at the monthly timeframe or day-of-the week effects, meaning there exist none of statistical significance.
In order to get a better grasp of potential drivers of the CBC’s late-day depreciations, recent price history is now also taken into account. While the prediction intervals are wide, the CBC shows a tendency for more late-day depreciations at times when TWD is appreciating, confirming anecdotal evidence of leaning against the wind.
Unsurprisingly, the size of late-day depreciations holds no predictive power for the direction of future USD/TWD moves.
…derivative settlements.
While the leaning against the wind effect is real, the most dominant effect hinges on the CBC’s, for a lack of a better term, current late-day mode. Once in late-day depreciation mode, the closing performance across all days, irrespective of their prior returns, is equally affected.
Given the unsatisfactory status of not knowing what factors influence the CBC’s end-of-day mode, a final argument is examined. Unusual market behavior can often be related to specific features of certain derivative instruments. Products involving knockouts, or gamma hedging more generally, are the classic examples in a non-linear setting.
In Taiwan, however, settlement of forward contracts appears a more realistic possibility. Non-deliverable forwards are a natural suspect, as their cash-settlement may provide incentive to expend some funds on a market moving spot trade, while earning a larger amount on NDF settlements resulting from such moves. NDF settlement issues have in fact made the occasional headline, however, their settlement is based on the 11:00 Taipei Forex Inc. fix, thus rendering these irrelevant to the present discussion.
On the assumption that the CBC has now disclosed all its FX exposures, it is its onshore forward book which could play a role in the late-day issue. Since in an FX swap transaction actual currencies are exchanged, banging the close before trade maturity would achieve little, since the closeout rate was already determined at the outset. However, a comparable effect is possible at trade initiation. If forward rates do not move during the late-day depreciation but spot does, this inevitably increases the hedging cost for domestic hedge acquirers.
For instance, during early afternoon Taipei time in mid-July, spot USD/TWD trades at 29.504, with onshore 3m forwards trading at 29.484. If now just prior to the close, spot depreciates to 30.00 while forwards stay unchanged, any hedger will exchange his initial TWD funds, at this from his view, worse exchange rate. If the CBC were to conduct its swap business only during the last hour of trading, it could by setting the spot exchange rate lower affect the pricing of hedges for lifers in yet another way. While speculative in nature, this would explain the recent late-day jumps, mirroring the CBC’s restrictive stance on onshore FX hedging. It would also explain why these jumps occur always late in the day and not earlier and also why they CBC’s mode tends to persist over weeks at a time.
In any case, there seems ample room for more transparency from the CBC regarding its late-day machinations. Either it is consciously overpaying (to the detriment of Taiwan’s broader population), or it is perhaps using the close of trading as a deliberate instrument to finetune the pricing of hedges it offers to lifers. In this case, monitoring its behavior for clues about the CBC’s stance on onshore hedging would gain added relevance.
Conclusion
The deeper forces at play in Taiwan’s FX space are unchanged from last October: a country with a longstanding policy of maintaining an undervalued currency, achieved first by official FX reserve accumulation and then by private sector debt acquisitions, aided in no small part by undisclosed hedges the CBC provided. The country is now coming to terms with the limits of this model, specifically the destabilizing effects of private sector FX exposures and questions of whether mercantilism is truly the most suitable economic model for Taiwan at this stage of its development.
The CBC’s apparent reluctance to provide further onshore hedges and the currency appreciation since last year—even if slow—seems a step towards a more balanced direction.
The U.S. Treasury’s next FX report could challenge this new architecture of Taiwan’s polices. The CBC’s spot interventions during the past four months were substantial and are likely sufficient to qualify Taiwan in terms of this criterion on the Treasury’s checklist. Its current account surplus remains easily above the two percent threshold and its adept handling of the COVID pandemic also ensures a trade surplus with the U.S. in excess of the USD 20bn limit. The Treasury’s ultimate publication date and its cutoff point for its review period will matter, but Taiwan seems clearly on its way onto the monitoring list and is aware of this fact, reflected in statements by its officials.
If Taiwanese authorities are really committed to work towards a more balanced state of trade and its current heavy spot interventions are only a slowing of appreciation to ward off debilitating side effects of private sector FX positions, Treasury might be advised to show leniency despite Taiwan currently checking all ‘currency manipulator’ boxes. In such a scenario, the TWD should continue to strengthen at the maximum pace possible compatible with keeping domestic FX risks at bay. An average quarterly appreciation of 1.5-2% would appear a feasible target. Within two to three years at this pace, the TWD would on a real effective valuation basis be back to early 2000s levels, a time when Taiwan’s current account surplus stood at much more acceptable, low single digit values as a share of GDP.
If present intervention were to prove (again) of the anti-competitive sort instead and Taipei officials could not provide assurances of continued, even if slow, currency appreciation, a more confrontational U.S. economic stance should not be ruled out. Taiwan’s deliberate past obfuscation of its forward book when confronted by (former) Treasury officials would weigh particularly heavily in such a situation.
It would be a great shame if Taiwan’s relationship with international partners, whose cooperation with the country seems more essential today than ever, be put at risk by the continued pursuit of mercantilist policies, which after all may at this point not be the optimal policy choice for Taiwan anyway.
Two recent events are worth highlighting in closing this piece.
On August 5th, it was reported in FX Markets that the CBC appears willing to allow lifers greater access to the onshore swap market for hedging dividends or coupons from overseas investments. While perhaps helpful to lifers at the margin, dividends and coupons represent only a small percentage share of the notional exposures underlying these cash flow streams. As long as the CBC does not grant access for hedging the underlying bond or equity exposures, its overall stance regarding onshore hedging would remain restrictive. Market pricing seems consistent with this interpretation, as neither did the onshore x-ccy basis curve widen, nor the offshore curve tighten visibly (particularly at the longer end) following the news.
It was also reported by Reuters that an emergency meeting on the strengthening of the TWD was held in mid-July, with participants from the FSC, CBC and private sector bankers. The timing of this meeting seems somewhat mysterious as the currency did not strengthen too much during the prior weeks. The focus on helping smaller and medium-sized exporters is equally confusing given that, as the article shows, the TWD had appreciated by a mere 2.1% at this point so far in 2020.
The central bank did not comment specifically, but, in other news, repeated its mantra to “intervene [in FX markets when] there is huge fund flow in or out in short period of time”. The CBC rarely defines what fund flows it is referring to and what huge means exactly. A quick look at Taiwan’s Balance of Payments shows that the current account surplus by far dominates equity inflows and the ‘net other investment’ category—the two types of flows most frequently termed ‘hot-money’ flows. As such, the CBC’s one-sided intervention practically always forms the counterpart to trade flows and not flows of a more speculative sort, for which FX intervention could be more easily justified.
1 FX reserves grew slowly during this time period, while lifers maintained their almost exclusive focus on USD-denominated debt. This implies the CBC’s excess FX exposures remained USD-denominated, leaving a reallocation of on-balance sheet into EUR as the sole alternative. >>
2 The more elaborate reason is that pressure in offshore forwards would be immediately transmitted into onshore markets by convergence trades, thus pushing the latter into the direction of offshore pressure. The deviations in the pricing of onshore FX derivatives relative to interest rate differentials would then invite arbitrage opportunities via spot trades, which could force the central bank to undertake spot interventions resultant from purely offshore pressures. Oftentimes, it is easier to control the access to FX derivative markets onshore than enforce strict financial account controls in the cash market, where myriads of methods are available to circumvent these, for instance by disguising arbitrage trades as current account transactions. >>
3 Incidentally, the aforementioned large equity selling of Taiwanese equities by foreigners during the first third of 2020 might have contributed to the widening in the offshore basis, as long USD/TWD positions (which hedged the FX risk of equity holdings) were unwound. Conclusive hedge ratios of equity investors are not available, but these anecdotally aren’t too large, especially given the tranquility of this FX pair. Reflecting this assumption is the commonly low statistical effects equity inflows have on the offshore x-ccy basis. In either case, the nominal amount of unwind trades is limited (<< USD 10bn) and can, as a point-in-time flow effect, not explain why the basis remained wide. >>
4 To fully exclude the possibility of arbitrage by foreigners, all possible risk-less bundles of exposures would have to be prohibited. The deposit and bond categories are clearly the most obvious ones, but any synthetic bundle with close to zero risk can serve as a substitute. For instance, acquiring a physical exposure to an equity index while selling a future on the same index. More abstractly, any forward sale of domestic assets, locking in a current price, has the potential to form a riskless bundle and should thus be subject to restrictions to an absolutely stringent FX regulator. As there are economic side-effects to banning activities of overseas actors in such areas, their regulation is oftentimes not enforced … and hence occasionally provides arbitrage opportunities for shrewd types. >>
5 It is important to differentiate the pure arbitrage situation discussed here from cases where banks hedge on-balance sheet FX mismatches (as shown in chapter IV.). In these instances, banks can either transact in onshore swaps or NDFs, depending on prevailing market pricing. >>
6 Prolonged here is loosely defined as accumulation of foreign exchange exposures by a central bank in excess of what may be deemed reasonable as insulation against detrimental FX volatility. >>
7 Rarely are the authorities of a country a completely omniscient and benevolent group. As result, macroeconomic policies are often shaped by considerations of specific subgroups, securing outsized (financial) gains relative to their productive potential. Such decisions often are to the detriment of achieving a society-wide optimal policy. Michael Pettis is the standard-bearer in this area and his recent book with Matthew Klein may provide a skillful synopsis of this worldview. >>
8 For instance, it is often the case in many countries for the central bank to intervene (absorb may be a more fitting term here) late in the day, when the extent of FX imbalance during business hours is clear. This is though far from a strict rule and interventions are frequently pulled forward when large order imbalance become clear earlier. >>
9 These exposures did still require an initial intervention in FX spot markets, so should also lead to late-day jumps, assuming one fully subscribes to this theory. >>
concentrated ambiguity 