Bulgaria: Determinants Of Local Currency Bond Yields1
1. Bulgaria has a modest public debt level and, like many other emerging market economies, has enjoyed favorable financing conditions in recent years, both in terms of access and yields. Nevertheless, the stock of public debt in Bulgaria has increased considerably since 2014, with sizable gross financing needs expected in the coming years. This, coupled with ongoing policy actions by the US Federal Reserves and ECB, creates non-trivial uncertainties about the financing environment Bulgaria is likely to face going forward. Against this backdrop, this paper explores how Bulgaria’s local currency bond yields have reacted to domestic and global factors in the past as well as the likely outlook in the near future. Our analysis suggests that local currency bond yields in Bulgaria are most influenced by external factors. The model also allows for simulating bond yields under different assumptions on future external factors. Overall, the analysis in this paper highlights the importance of being prepared for sudden and potentially large changes in financing costs.
Public debt % GPD
Sources: Bulgarian authorities.
2. This paper is organized as follows. Section II presents some stylized facts; section III summarizes the model and estimation results; section IV discusses the outlook; and the last section concludes.
B. Stylized Facts
3. Bulgaria’s local currency bond yields are currently around the lowest levels since consistent data were published by the Reuters. Despite some generally mild ups and downs, Bulgaria’s yields have broadly trended downward from the peaks immediately after the global financial crisis. Consistent with literature findings, the yield declines in Bulgaria have been much larger than those in advanced economies, such as Germany. The yield for Bulgaria’s 10-year generic local currency bond currently stands at about 2¼ percent, only a small fraction of the peak level achieved right after the global financial crisis, and less than the half of the average for 2006–10. The yields for 5-year and 1-year bonds, at below 0.5 percent and close to 0 percent, respectively, are also at the historically low levels.
10 years Sovereign Yield - Selected Countries
5 years Sovereign Yield - Selected Countries
1 years Sovereign Yield - Selected Countries
4. Nevertheless, the inflation-adjusted real financing cost for Bulgaria is broadly comparable with those countries that have similar sovereign ratings by the S&P. As of March 2016, Bulgaria 10-year local currency bonds averaged the nominal yield of 2.3 percent, much lower than the average for the 11 economies that have a rating ranging from one notch above and one notch below Bulgaria’s. However, after adjusting for inflation (y-o-y), the real yield for Bulgaria’s ten-year bond was in the middle 50 percentile of this group of countries.
|Country||L-T local currency rating||10-year bond yield (Nominal)||10-year bond yield (Real)|
C. Determinants of Local Currency Bond Yields in Bulgaria
5. Following the case study of Poland by Ebeke and Lu (2015), we use a simple error-correction framework as the baseline model, relating the bond yields in Bulgaria to two types of determinants: global factors and domestic factors. Global factors can be further divided into two groups: liquidity indicators (ECB policy rates, US Federal funds rate, etc.) and risk indicators (e.g., VIX). Domestic factors can also be divided into two categories: politic risks and economic fundamentals. For political risks, like Comelli (2012) and Csonto and Ivaschenko (2013), we use the political risk indicator from the International Country Risk Guide (ICRG) database, which contains monthly data for 140 countries. For economic fundamentals, we tried a wide range of variables although difficulties often arise as such series often come at lower frequency than yields, with substantial volatility, and subject to multiple round revisions. The error-correction model will allow us to explore not only the long-run co-integration relationship, but also the short-run dynamics, which should be of particular interest to the policy makers as the US Federal Reserve has recently started to hike the short-term policy rate.
6. This section will focus on the 10-year local currency bond yields and the baseline error-correction specification is as follows:
BGR10Yt = a0 + a1PRt + a2FRPRt + a3log (VIXt) + a4PRRt + Xtθ + et
FRPRt = FR12X15t - PRt
ΔBGR10Yt = β0 + β1ΔPRt + (β2ΔLRPRt + β3Δlog (VIXt) + β5ΔPRRt + ΔXtδ + β6et-1 + εt
where BGR10Yt is the nominal yield of the 10-year local currency bond in Bulgaria, PRt is the 3-month euro-area interbank market rate,2 FR12X15 is the euro forward interest rate applied for the future period between twelve and fifteen months. The differences between the interbank market rate and forward rates are used to signal market’s expectation on the potential direction of the policy rate. Log(VIXt) represents the natural logarithm of VIX, which is used to approximate the perceived risks of investing in Bulgarian securities from the impacts of global financial stress. PRRt is political risk rating and Xt is a vector of economic fundamental variables.3et is the error term, and its one-period lag is used in the short-run dynamics as the error correction term.
The results are summarized in the text table below. Column (1) reports the results with 3-month interbank market rate and interest rate outlook, defined as the differences between the interbank market rate and forward rates, as the explanatory variables. The estimated coefficients are significant and these two variables could account for 20 percent of the variation of the 10-year local currency bond yield. The explanatory power of the long-term dynamics becomes much higher when VIX is added to regression. Variation in all of the variables in column (2) could account for 78 percent of the variation in the yield. The estimated coefficient of VIX is highly significant with expected positive signs–high VIX is associated with high bond yield. In column (3), we further added the indicators that approximate the domestic fundamentals. The estimated coefficients have the right signs but are significant only for political risks.5,6 Nevertheless, their overall contribution to explaining the variation of the dependent variable is rather limited. specification (3) is used as the starting point for derived the errors for estimating the short-term dynamics. The effective fed fund rate7 and VSTOXX, the euro zone equivalent of VIX, were also tried to represent the monetary policy stance and expected market volatility, respectively. They, however, did not perform as well as euro area interbank bank rate and VIX, which were used in specification (3).
|Dependent Variable: Local Currency 10Y Bond Yield|
|Interest rate outlook||0.72**|
|Political risk rating||−0.12***|
|Fiscal balance(% GDP)||−0.06|
|No. of Obs.||123||123||123|
|Dependent Variable: ΔLocal Currency 10Y Bond Yield|
|ΔInterest rate outlook||−0.04|
|ΔPolitical risk rating||0.02|
|ΔFiscal balance(% GDP)||−0.06|
|Error correction term (one lag)||−0.13***|
|No. of Obs.||122|
The most significant impact on short-term yield dynamics are from the changes of the VIX with the estimated coefficient having the expected positive sign and statistically significant. Same as in the long-run relation, higher VIX is positively associated with an increase in bond yield, which confirms the intuition that Bulgaria, being a small open economy with a rather open capital account, is sensitive to global financial development. The estimated coefficient for the error term derived from the long-run dynamics is significant with the negative sign as expected, suggesting a stable dynamic, but its magnitude is on the small side. The estimated coefficients for other variables are small and statistically insignificant.8
7. Building on the findings in the preceding discussion, this section will consider two simulations for Bulgaria’s sovereign yields with different scenarios for global factors in the next 12 months. For simplicity purpose, we will assume that domestic factors will remain stable; and in outlining short-term dynamics, we will only include the variables that have statistically significant coefficients (i.e., VIX and error correction term). For illustration purpose, we will consider two extreme scenarios. In the optimistic scenario, global financial condition will steadily improve, with the VIX gradually reaching the lowest values observed during 2006–15, and ECB will hold the policy rate stable. In the pessimistic scenario, global financial stress spikes with the VIX steadily rising to the highest readings observed during 2006-15, and the ECB will cut its policy rate by another 20 basis points in two steps. In both scenarios, it is assumed that the interbank market rates move in parallel with the ECB policy rate and the financial market perfectly forecast and price in the timing and magnitude of the ECB policy actions. In the pessimistic scenario, the 10-year local currency bond yield will likely steadily move up and reach close to 4 percent. The projected peak yield is expected to be significantly lower than that during the global financial crisis, mainly due to lower ECB policy rate and downward expectation about the policy rate. In optimistic scenario, Bulgaria’s yield is expected to slightly decline to around 2 percent, mainly due to the declining global financial stress as represented by VIX.
Bulgaria: Yield Outlook
Souarce: IMF staff estimates.
E. Conclusion and Implication
8. This paper finds that Bulgaria’s sovereign yields have been driven predominantly by external factors. In the long run, the variation of global financial condition, represented by VIX, counts for most of the variation of sovereign yields. In the short term, the change of global financial conditions, together with the distance from the long-term equilibrium, has a large and immediate impact on the sovereign yields. This is not surprising given that Bulgaria is a small open economy with close integration into the world market, although foreign participation in Bulgaria’s local bond market is relatively limited.9
Foreign Holdings of EM Government Debt Securities
Source: Sovereign investor base estimates by Arslanalp and Tsuda (2014).
9. The most direct policy implication from the finding seems to call for preparedness for potentially sudden spikes in borrowing costs due to exogenous changes in the external financing environment. This will require continued fiscal consolidation as envisaged in the government’s medium-term fiscal framework so as to reduce gross financing needs and debt service burden. In the meantime, with external financial conditions remaining tranquil, Bulgaria will likely continue to benefit from historically low financing costs.
ComelliF.2012. Emerging Market Sovereign Bond Spreads: Estimation and Back-testing. Emerging Markets Review13598–625.
CsontoB.IvaschenkoI.V.2013. Determinants of Sovereign Bond Spreads in Emerging Markets: Local Fundamentals and Global Factors vs. Ever-Changing Misalignments (IMF Working Paper No. 13/164).
EbekeChristian and YinqiuLu2015 “Emerging Market Local Currency Bond Yields and Foreign Holdings—a Fortune or Misfortune?” Journal of International Money and Finance59 (2015) 203–219.