Colin Lambert reviews a new paper from the Bank for International Settlements’ Monetary and Economic Department, which suggests the direct links between liquidity conditions in FX spot and swaps markets are getting stronger – and that regulation is having a serious impact on the FX market’s functioning.
At face value, the phrases “…the pricing of spot and FX swaps is intimately linked…” and “…price formation in FX swaps depends on price formation in the spot market, and vice versa” would raise eyebrows amongst FX traders. It is generally acknowledged that there is a close link when setting the spot leg of an FX swap, mainly due to the need to hedge the “spot effect” or the difference between the two virtual interest payments, however the “vice versa” in the above statement doesn’t garner much support with few spot traders or pricing engines taking into account what is happening in the swaps market.
In certain currency pairs there is an undeniable link and automated pricing engines in spot often have trouble quoting adequately in pairs such as the CAD crosses thanks to the USD/CAD leg being a one-day maturity while the other leg is most often a two-day maturity. Equally, in some emerging market pairs the same trader will manage both spot and forward books because they are a country specialist. But away from those examples? It is hard to find solid links – just look at how the businesses are often managed, with the FX swaps being part of the interest rate franchise and the spot either sitting alone or being part of a global macro type set up.
The above referenced phrases appear in a new paper from the Bank for International Settlements (BIS) Monetary and Economic Department, which finds empirical evidence that liquidity in these markets is linked and that there are spillovers between the two. Crucially, the authors, Ingomar Krohn and Vladyslav Sushko, focus on what happens around month and quarter ends – a time when liquidity conditions undoubtedly change, although largely one could argue (outside of the Fix) in the area of the cost of funding liquidity.
The paper sets out to assess liquidity conditions taking into account “the interrelation between liquidity provision in FX spot and FX swaps”. By assessing conditions in the FX swap market at the same time, the authors are breaking new ground – most previous efforts to assess liquidity have focused on spot markets only – and this allows them to examine the interaction between FX funding liquidity and more general market liquidity levels. They do this in the EUR/USD and USD/JPY pairs using quote data from Refinitiv Datascope, which raises the first potential variable – the data is not based upon actual trades, and therefore the bid-offer spread the authors use to assess liquidity conditions (the data is cleaned up and the analysis more complex than that basic measure of course) may not always reflect conditions in the market, for example some dealers could be quoting not to deal due to a variety of factors.
Notwithstanding that, the authors state: “We find that bid-ask spreads in spot and FX swaps are very highly correlated, indicating that market liquidity in spot and swaps markets is intimately linked.”
Specifically, the paper finds that a deterioration in FX funding liquidity, as measured by deviations in the covered interest parity (CIP) lead to a widening of spreads in both swaps and spot markets. This focus on month and quarter ends to study the impact of funding shocks on spot markets makes it is easy to see why there would be an impact given how a funding squeeze would impact the cost of carry to traders carrying open positions (especially Japanese retail engaged in the carry trade), which in turn leads to a clearing out of positions, volatility in spot markets and a widening of spreads.
An interesting follow up to this paper would be to analyse the relationship at other times when there is not necessarily a funding squeeze, for example in this research the last price quote in each hour is used to build the database. This may be skewed slightly (although again the database is large enough to suggest it may not be) by what happens around the major fixes of the day, especially 4pm London, when major dealers typically reduce their principal activities while they, or an agency desk at the institution, executes the heavy Fix flow. This reduction in pricing to the market during major fixes is most likely unrelated to anything in the swaps market. There is also the question of how many of the price quotes used in the data are actually providing firm liquidity.
Another factor suggesting that the authors have picked up on a specific phenomenon in how their model finds a stronger link between funding liquidity and market liquidity in USD/JPY than it does EUR/USD, as again this suggests that the carry trade may have had a significant influence on the findings.
An interesting aside in the paper is how it finds that the positive net effect of dealer competition in FX swaps has “all but disappeared” in the second sub-period studied, from July 2014 to May 2017. This would suggest that spreads have compressed to a level through which the major dealers no longer see the value in pricing to the market, although it could also highlight that the mix of interests in the market is suitably rich to ensure there is generally good two-way interest. This will be an interesting area of study as more multi-dealer platforms seek to roll out their FX swaps trading technology – generally these platforms’ selling point is how they can offer tighter spreads as part of a more efficient package overall, if the spread compression has already happened, one potential selling point is at the very best diluted.
Impact of Regulation
Perhaps the most notable finding of the report is how it highlights the impact of the capital and liquidity rules on FX swaps market liquidity, which the data suggest has become “several times larger” since 2014 thanks to G-SIBs (global systemically important banks) “window dressing” their balance sheets for quarter-end reporting. When looking at quarter-end US dollar liquidity droughts, Covered Interest Rate Parity (CIP) deviations since 2014 have become three times greater than they were during the 2011-12 European debt crisis as G-SIBs have significantly cut back on their quoting activity in FX swaps during these times.
The “window dressing” referred to is the practice of G-SIBS to reduce the size of their balance sheets over reporting dates in order to cut the potential cost of capital and liquidity requirements. It can also help keep some banks in lower tiers when it comes to helping them avoid being pushed into a higher, and as such more costly, G-SIB bucket.
When these conditions hit, part of the slack is taken up by non-G-SIB banks, the authors find, however, these banks typically quote wider and in smaller amounts, hence the deterioration in market depth. This is limited to the FX swaps market though, which does suggest that the links between the two markets are not as strong as some of the headline comments would have people think. The report notes how G-SIBs generally maintain a healthy presence in spot markets throughout most conditions.
Highlighting this impact of regulation, the research finds that prior to June 2014, liquidity barely changed during quarter ends, but FX funding conditions were usually worse. Post July 2014 (and it could be significant that the Liquidity Coverage Ratio aspect of Basel III came 100% into effect in 2015 but was announced in mid-2014), FX funding liquidity measures are “considerably worse” and the deterioration at quarter-ends much larger in relative terms, with spreads about two times wider. “Furthermore, unlike the earlier period, bid-ask spreads exhibit widening at quarter-ends for both spot and swap markets, indicating possible spillovers from FX funding to FX market liquidity at quarter-ends during the most recent period,” the paper states.
There is one slightly puzzling dichotomy in the paper where the authors at one point state: “Overall, the empirical evidence suggests that a deterioration in funding liquidity at quarter-ends can spillover to market liquidity in spot and swap market[s]. Taken together with our previous results on dealer activity, these findings suggest that the pullback by G-SIBs from dealing in FX swaps at quarter- and year-ends can have a particularly contagious implications for spot market liquidity.”
Later on in the paper, however, they also write, “While large dealers still dominate market making in spot markets at all times, and their quoting intensity is associated with improved liquidity dynamics, they have exhibited a tendency to pull back from posting price quotes in FX swaps around balance sheet reporting periods.”
While they add that “spot market liquidity appears to also suffer because liquidity conditions in spot and swap markets are tightly linked”, there is still the nagging feeling that the findings are being skewed by the use of month- and quarter-end data, which are associated with larger moves in spot markets because of the tendency of funds to rebalance their portfolios at these times – using the Fix or not. Thus, it could be argued that the real influence on spot market liquidity at quarter-end is less impacted by what happens in swaps than suggested here and that it is really more about the dynamics of an influential group of participants in the asset managers whose activities around these periods are widely known.
This is a very intriguing paper and leaves the reader wanting more, specifically, as noted, a study of the phenomena at different times as well as, if possible, a study using real trade data where possible, in order to better calculate liquidity in both markets. In day-to-day terms, the suspicion is that what is happening in swaps markets barely registers with spot traders who are more concerned with data from spot markets and their own order books to formulate a price – witness the meltdown in repo markets in September 2019 when FX swaps liquidity in dollar pairs disappeared, while the spot market continued to function normally. That there is an influence at certain times is undeniable, however, and this paper offers a real advance in the study of those links, what would be very valuable now is a follow up that can throw light on whether this study is looking at a phenomenon or a structural trend in FX markets.
The full paper can be downloaded at https://www.bis.org/publ/work836.pdf