Fra Barclays
Referendum countdown
Turkey is heading for a public referendum on the presidential system on 16 April, this Sunday. The recent polls suggest a pick-up in momentum in favor of “yes” (Figure 1) and this is also evident in the latest surveys of pollsters such as Metropoll and Gezici, which show “yes” at around 53% as opposed to earlier surveys of below 50%. The “yes” and “no” outcomes still appear to be close on average, however, and the associated margin of error (2-3% according to pollsters) along with the large share of “undecided” voters underscores the binary nature of the referendum outcome.
Achieving 50%+ support for the “yes” campaign (AKP-MHP) might look relatively straightforward at first sight, given a combined voter base of 61% (November 2015 election results) and the almost perfectly aligned rhetoric and policies of the parties. Recall that the transformation of the political landscape before and in the aftermath of the November 2015 elections led to a firm macro-level consolidation of Turkish politics along two lines: the “nationalist front” (mostly AKP and MHP voters) and the “social democrat front” (mostly CHP and HDP voters).
However, polls suggest a less comfortable race for the “yes” campaign: i) a large number of MHP voters seem to still be unconvinced by the “yes” campaign, and ii) the true color of the “undecided voters” is hard to decipher: pollsters say voters are increasingly refraining from revealing their preference due to the mood created by the extraordinary state of affairs, and “undecided” voters could be skewed towards “no”. Metropoll argued that more than 75% of undecided need to vote “no” for it to win; while historically, undecided voters have tended to either vote for the status quo or not participate in elections. The low polling response rate due various factors (such as peer pressure in the South East and the Black Sea regions according to some pollsters) is yet another complication that could potentially be distorting the poll results.
Nevertheless, momentum has picked up in favor of “yes” based on the most recent polls, and this has also been echoed by the political expert media commentary. Among the key catalysts, experts note the following.
- The effectiveness of President Erdogan’s campaign to push the “yes” votes of the AKP electorate higher (from the 80% level to the 90% level), as well as its impact on convincing more of the MHP electorate to vote “yes”.
- The tailwind provided by escalating tensions with the Netherlands and the EU, which is likely to help consolidate the nationalist vote.
- The positive impact of the improving economic sentiment recently on the “yes” campaign (Figures 2 and 3).
- The tactical missteps of the main opposition party CHP (i.e. comments by party officials) influencing undecided AKP voters in favor of “yes”.
Overall, it is hard to simplify the referendum to a probability exercise of an early election under yes/no outcomes, given that it is far more complex and the implications are unclear (both scenarios entail an early election possibility, in our view). While markets will likely perceive near-term political risks (i.e. an early election) to be lower in the scenario of a “yes” outcome, medium-term concerns about policy uncertainty and institutional strength are likely to remain; the Venice Commission opined that constitutional changes will remove checks and balances, leading to weaker institutions. The influence of presidential advisors on policymaking and the transformation to a “new economic model” would likely be solidified under a presidential system accompanied by an accommodative monetary policy bias, a potentially looser fiscal stance and an increase in quasi-fiscal spending focused on infrastructure projects via the sovereign wealth fund (SWF). We believe that, even in the event of a “yes” outcome, the likelihood of an early election in Q4 2017/Q1 2018 is still non-negligible as President Erdogan may choose to bring forward presidential elections (from August 2019).
* * *
Market implications for FX, local rates and sovereign credit
The market appears largely positioned for an outcome consistent with polls that suggest a “yes” result is most likely. In FX option markets, for example, the volatility-adjusted premium for USDTRY calls versus puts has recently dropped to multi-year lows and below-average kurtosis suggests little demand for negative tail-event protection (Figure 4). In bonds, TurkGB risk premia remain extremely low and currently offer less return than USTs on a hedged basis (Figure 5). Finally, in EM credit, Turkey YTD has outperformed the broader Bloomberg Barclays USD EM sovereign index, partially reversing the c.10% underperformance in 2016 (in total return terms).
YES: A “yes” outcome would likely result in a broad-based, yet potentially short-lived, relief rally
Despite the market’s anticipation of a “yes” outcome, we think the associated reduction in near-term political uncertainty would likely still deliver some relief rally, allowing a temporary reprieve for the TRY and a steeper curve in anticipation of a “gradual” unwinding of tight liquidity policy.
In FX, still-large TRY political risk premia and undervaluation suggest room for appreciation following a “yes” outcome. While our estimate of the lira’s political risk premia has reduced from 15pp at the end of January, it remains relatively large at 8pp (Figure 6). Furthermore, our short-term Financial Fair Value (FFV) model suggests a 4% undervalued TRY against the USD (Figure 7).
We believe risk-reward argues for being long TRYZAR targeting January highs of 3.90 with a stop-loss at 3.67 for a reward to risk ratio of 3:1 (spot reference: 3.73). We prefer this to short USDTRY as South Africa’s similarly low risk-adjusted real interest rate differentials and heightened political risk should provide a degree of protection in the event of a “no” outcome. The trade also remains positive carry.
In rates, very low bond risk premia suggest a rates rally following a “yes” is likely to be concentrated at the front end of the yield curve as market participants will likely price a gradual unwinding of the CBT’s liquidity tightening measures. As such, we reiterate our existing trade recommendation of paying the 1s5s TRY cross-currency swap spread targeting -30bp with a stop-loss of -100bp.
For Turkey sovereign credit, we maintain our Market Weight rating. This balances our concerns about a likely medium-term deterioration of Turkey’s credit metrics in a presidential system on the one hand with relatively attractive valuations and likely reduced near-term political uncertainty in a “yes” vote on the other hand. In the near term, we see potential for further spread compression of Turkey against South Africa, especially in the 5y sector of the curve (Turkey ‘22s vs SOAF ‘22s), with South Africa remaining vulnerable to adverse developments.
In the corporate credit space, we also have a Market Weight rating on Turkish banks and corporates. In the case of a “yes” vote, we would expect bank seniors to benefit more than corporates given the more significant spread pick-up relative to the sovereign. Higher beta seniors trading at a discount of over 100bp to the sovereign as well as new-style Tier 2s yielding over 7% are likely best positioned to benefit, in our opinion, although this could be met with more Tier 2 supply. We would expect the opposite reaction to a “no” vote, with IG-rated corporates and more expensive bank seniors as well as old-style Tier 2s to be less vulnerable in any sell-off.
NO: Positioning for a “no” outcome looks most compelling via FX options
The less-expected “no” outcome will likely result in larger market movements as a higher risk of an early election would increase risk premia in the local bond curve, weigh on the lira and increase FX implied volatility, in our view.
