Fixed income, Fed, and CBRT

THE SPREAD BETWEEN US 2-year and 10-year notes was 0.37 exactly a year ago. Now it stands at 1.47. This kind of normalization entails a strong expectation of genuine recovery, especially after the Biden bill passed. It also underlines a reflation scenario. With fiscal packages amounting to USD 5tn, nearing a quarter of the U.S. 2020 GDP, such anticipation is only normal given that the Fed balance sheet has also expanded from USD 4.312bn to USD 7.58tn in exactly one year. The sell-off in risky assets, i.e. stocks mainly and EM currencies, was triggered by the above 40 basis points increase in the U.S. 10-year note since early February. All major central banks balance sheets have expanded, however, and this is one of the reasons why the dollar didn’t depreciate much. It is also a sweeping argument for why it isn’t appreciating rapidly either. The playing field is even, so to speak, but the American reflationary recovery is the main story going forward. Finally, markets are pricing in a vaccine-driven quick recovery also.


As spreads narrow, for instance between 2- and 10-year U.S. Treasuries, concerns about the sustainability of American growth have been constantly voiced. There is predictive power in the term structure, or so the profession thinks, and there are studies that show the predictive power in question hasn’t diminished as of late. Therefore, the current yield curve normalization should also be equally indicative, and markets think the pending recovery will be huge, in the order of 5-6% GDP growth. So what about the U.S. spread? Well, for Turkish assets it carries a message.

Do we have an insight as to what may happen next? Past data show causality running from the Fed, and more often than not, a spread that either increases steadily compared to the past – as it happened after 2010 compared to the previous window of 7 years – or remains almost stable. Regarding Turkish Eurobonds and TRY-denominated fixed-income assets what this implies is clear: as the Fed’s future plan unfolds, Turkish assets have little chance of narrowing the spread. In other words, the episode between 2009 and 2011 may prove to be the golden age: the spread between Turkish Eurobonds and U.S. 10-years remained contained within the disciplinary box of 5.60-5.90 percentage points. It is to some extent that episode that breaks “stochastic dominance”. From now on, even spread-preserving behavior will dictate a commensurate increase in Turkish asset yields. Hence, we tend to think the market has already priced in only part of what is to come.


The other issue for Turkey is obviously CDS behavior and carry trade returns. An important problem in finance has been correlation within a portfolio; that is among the asset returns that form it. In this respect, using long-memory data sets would be of no avail, since such a practice only, perhaps, augments the parameter space since co-variation matrices can grow very large. Gaussian copulas can to some extent be used to separate marginal distributions from their joint dependency. Here the main issue is correlation. A CBRT study demonstrated that CDS behavior can be traced to macroeconomic policies, credit ratings and carry trade returns, which imply that there are discernible co-movements in the EM CDS performance patterns. Furthermore, movements along the tails can be symmetric. Hence good and bad news and events may have the same impact on CDS portfolios. As in many equity market booms observed in the past, it is not bilateral financial movements and trades that explain clustering in CDS performances. There are groups that show common dependence on both tails and averages. So, CDS volatility is also what we would have expected should the Fed tighten. Yet we aren’t there yet, and there is still a window of opportunity. If only reflation turns into a massive surge in U.S. inflation should we expect tightening, and that will definitely bear on the CDS – but not now.


The Fed is crucial, but also consider this. In many New Keynesian DSGE models, common monetary policy shocks and pairwise cross-country correlations are omitted, some would say. Hence, there exists a global component that spills over other countries as both core economies and the global economy as such become financially more integrated. Shocks that appear local may in fact be caused by global factors. First, there is a common Euro-Dollar component to such effects. The coefficient estimates of the shares of economies’ overall financial integration accounted for by the US or the euro area are both positive. The common component in the monetary policy shock estimates contains both a US and a euro area component. Looking at the Fed only is not enough.

Second, the cross-country interaction between economies’ gross foreign asset and liability position relative to GDP can be proxied at times by the interactions between portfolio, foreign direct and other investment relative to GDP. Monetary policy shock estimates obtained by adding a common global component are larger for economies for which ‘other investment’, which includes bank loans accounts, form a larger share in their overall gross foreign asset and liability position. Of course, the category of ‘other investment’ also includes items unrelated to bank loans, for example trade credit and advances, special drawing rights or currency and deposits. The ratio of non-resident bank loans relative to GDP as an alternative and possibly more accurate measure of the importance of cross-border banking linkages suggest the higher such loans account for a higher portion of total funding the higher the spill-over effect.


Yes, but on what metric are they more vulnerable? The main linkage passes through non-core financing. While this is true, ‘other capital flows’ do explain a lot. There is probably no significant effect coming from general portfolio investments, investments by non-residents, on domestic lending by high non-core liability local banks. It is the loan channel directly, which is captured by other capital flows that make the difference. Large and high non-core liability banks lend more as they can find overseas funding more, and that’s it. If not, lending becomes problematic. The spill-over effect is larger compared to the past, and the sensitivity of bank lending on such conditions is also higher. If there is second-order stochastic dominance of the Fed rates on Turkish bond yields – a plausible hypothesis backed by many, but it depends on the window you open according to my computations – a more important ‘dominance’ channel comes from the ‘other capital flows’. It isn’t just the Fed, and it isn’t just the rate of interest, either here or there.


All arguments were in favor of at least a 100 bps hike. Inflation is on the rise. Given the prospect for recovery at least on the basis of positive base effects after the vaccine – currently this is rather enigmatic- and against the backdrop of soaring oil prices and food prices, a rate hike would have been prudent. It would also reinforce the credibility that has been slowly building up anew. Moreover, dollarization is very hard to reverse on short notice. Further to that, people think inflation is in fact higher than the official print. Also, the U.S. 10-year note’s yield is on the rise. And the precious FX reserve buffer has eroded. And the head-on inflation outlook isn’t improving.


The response to all these was not only perfectly in line with expectations but it surpassed the median forecast. This is a wise policy move, and it is timely. The raise should suffice to show all players that the CBRT is an active watch guard on the exchange rate front. I am one of those who think the policy rate ought to be kept high – in the vicinity of 17% (+/- 1) – for the best part of the year if not in the vicinity of 20%. All rate cuts would be premature unless vaccination works globally and in Turkey, and the tourism season opens with flying colors and much needed hard currency pours in at breakneck speed. The CBRT has come in with full force indeed. The decision and the accompanying message would help sharply build up the CBRT’s credibility.

Leave a Reply

Your email address will not be published.