Is there a third way?
With or without the IMF, that is the question. We ought to note that ‘net errors & omissions’ amounted to $20 billion over the last twelve months. Now, Argentina signed an agreement with the IMF: $50 billion first, with an addendum of $6 billion afterward. Out of $56 billion, Argentina has received $21 billion so far. Another tranche of $12 billion is expected to be released soon. Hence, net errors & omissions, as it did in 2009, has effectively come to replace an IMF stand-by. Could that continue? I don’t think so.
What is going on?
Based on the relative stabilization in the financial markets, CDS and so on, I had come up with 6-6.10 USDTRY by the end of 2019 as sort of a ‘fair’ level, fair meaning a depreciation on par with the last ten years’ historic average. Inflation plus something…Any level exceeding that, especially if it is accompanied by a sharp rise in implied volatility, will mean a higher than expected inflation, less room for maneuver in monetary policy, balance sheet recession and what not. Another currency shock is the one thing that this economy can’t risk or afford. CDS is a crucial metric on this score. We have seen how it could fluctuate as an indicator of ‘composite (country) risk’. This time around, it wasn’t the CDS or anything. Because locals didn’t believe the exchange rate stabilization was here to stay, they continued piling up FX assets even when the USD/TRY rate was falling. New political risks pop up all of a sudden, and give reason to those who think investing in dollar assets is the only way to make money or at least to protect savings against inflation. Analysts may also have looked at net errors & omissions, and decided reserve depletion could be the only option in the absence of new portfolio inflows.
Dollarization or what?
Credibility is a delicate concept, and if lost it can only be gradually built up again. Also, dollarization is a stochastic process that displays memory-dependence. A full dollarization is a persistent phenomenon. It doesn’t reverse back easily. Moves in both directions along the dollarization curve take time. So, what is going on here? Two options exist. If the lira remains relatively stable in the aftermath of elections, we should see a fall in FX holdings. Otherwise, either asset substitution or FX debt is so high that corporates buy FX well before the date when their payments are due, believing that the now and then ‘current’ rates are always favorable. When the exchange rate jumps, there is almost always a fat tail to the distribution ex post, and not only that the distribution is fat tailed ex ante. In other words, people jump in at the last minute, mostly after the shock has run its course. Hence, as the lira depreciates, people begin holding more FX deposits and suchlike. That is understandable - late reaction to the movement, human psychology etc. However, now FX holdings are going up even though the lira has stabilized. Look at the scissors-like shape resident FX deposits took against the USD/ TRY lately. This isn’t only swaps. In fact, swap costs are quite high and making money out of swaps is increasingly difficult. Two explanations: either those who had already bought FX at high exchange rates are now trying to average out the cost of their FX holdings by buying more at lower rates or people don’t believe in the stability of the lira and switch to FX holdings – or both. There may also be a gradually forming dollarization tendency. The fact that money is perhaps endogenous keeps the dollarization (asset substitution ratio) stable, but dollar holdings go up in terms of magnitudes. If that is borne out by events, we might be looking at a hill-climbing Hurst coefficient, and the FX deposits/ M2 ratio could jump, indicating asset substitution.
Could rebalancing help keep the exchange rate stable?
It isn’t the trade or current account deficit per se that triggers lira weakness. It never was in fact. It is because risk premia have risen beyond any controllability, admissibility and measurability arguments. It is because exports depend on imports that asset dollarization has been going up in tandem with debt dollarization, most FX debt is unhedged, and the country needs c. $200 billion in the next 12 months to roll it over. Plus, there is political noise. This isn’t an old-style balance of payments crisis. The current situation has nothing to do with the Asian Crisis of 1997 or with the Russian Crisis of 1998. A positive non-energy non-gold current account is a good thing, but it has no bearing on currency stability as such, as we have seen time and
again. The sources of instability are different.
Can the CBRT do anything more?
Except the standard argument to the effect that inflation should be rendered low and stable, there are many things a central bank can do to prevent any shift towards dollarization. Forced de-dollarization is generally short-lived and counter-productive. Financial instability is the most fearful consequence, bringing in disintermediation, informal lending and expatriation of local capital. The government already announced a couple of years ago that a complete ban on FX borrowing for 23,000 small companies with FX debt below $15 million was due. This is not forced de-dollarization per se, but it comes close to it. Had it been spread to larger firms and larger sums it would have been interpreted in this light. However, there are no limitations for such companies. Instead, they would be required to implement hedging strategies. However, the details on hedging have not been announced yet. Tying FX income to FX debt is also a good idea, and I don’t buy the argument that such restrictions will lead to capital flight. It is just a sound, reasonable macroprudential measure. Nevertheless, it was a bit late, or too little too late.
One way of doing this – to trigger de-dollarization - is to create a sizeable wedge between reserve requirement ratios of FX and local currency deposits. FX deposits can also be discouraged since the currency composition of deposits and loans have a close relationship; at least from time to time they converge to each other. This is already done because RRR for local currency deposits and FX deposits are different; FX deposits carry 250 basis points disadvantage. Already there are measures in place. The CBRT has been trying to prolong the duration of non-core FX liabilities, for instance. I don’t think tightening net open positions is fruitful be- cause the net banking short positions are not high anyway. Toying with the liquidity coverage ratio so as to plug in a disincentive for FX liabilities can also provide a feasible tool. Deposit insurance ratios can also be altered in the same spirit. Again, I am not sure whether extending central bank operations in FX derivatives is a good idea. That would directly place central banks as market plays along the same dimension as hedge funds. However, there exists the example of Mexico, and it seems it can be done, generally speaking. However, there were already measures taken in that direction. All in all, reducing FX volatility would require many factors, some of which are within the power of the CBRT, some of which aren’t.
This is why a fully articulated new macro program is warranted
As recession continues, except subsidies and incentives of all sorts, there are very few things left to do. Monetary policy isn’t an all too powerful tool here, and budget discipline is a two-way street, at best an anchor and a sign of commitment. Furthermore, we don’t see fiscal discipline in the data yet. Monetary policy isn’t sufficient alone because money is probably endogenous in Turkey. Endogenous money means that money is created not by fiat, but by the exigencies of the financial system. Once loans have been granted and deposits have increased so core liabilities match to a large degree core assets, central banks accommodate demand for central bank money so as to guarantee solvency and manage systemic liquidity. In Turkey, the system is bank-based, and banks have also rather substantial non-core liabilities. Both core and non-core liabilities create assets. Money responds to asset creation; causality runs from assets (loans) to monetary aggregates. This is one view, and the evidence seems to shore up this conjecture. Hence, the central bank can only control the interest rates but not the supply of reserves. In other words, the central bank can either control interest rates or the quantity of its liabilities. In that case, barring the impact of cross-border capital flows on the currency and bond markets, the central bank can in principle use its interest rate as an exogenous policy tool, but not its supply of money. There is also the view that the central bank may not fully accommodate reserves demanded by commercial banks. If there are liquidity constraints, banks could overcome such constraints imposed via liability management. The liquidity preference view is the third approach, which questions the demand-driven money supply approach. Because economic agents have different liquidity preferences about the amount of money they wish to hold, if the supply of deposits is insufficient to meet the demand for loans, individual preferences will change relative interest rates to fill the gap by increasing the supply of deposits and reducing the demand for loans. This hardly fits the Turkish case though, where deposits are almost always in short supply and the game is not a stage game whereby banks compete for deposits at time t=1 and lend at time t=2 accordingly.
While there are weak signs that in Q2 GDP growth could turn to positive, the future is unclear. Because any currency shock could trigger balance sheet recession of a sticky nature, there will have to be a drastic intervention to real sector balance sheets. This means recapitalization + debt restructuring. Because banks can’t incur the whole burden, there will have to an SPV, debt issuance and securitization. This isn’t unchartered territory, but it has to be undertaken with care. The help of favorable base effects and the quasi-automatic business cycle frequency-led improvement in economic activity alone may not wash. So, is there a third way indeed?