Turkey: ending the stop-and-go growth?

As the year 2021 starts, it is once again time for Turkey to make some choices. For nearly the past eight years – ever since Chairman Bernanke hinted in May 2013 that the Fed would scale back its QE programme – Turkey has been plagued by major macroeconomic imbalances (inflation, external deficits and/or public deficits). Although they did not emerge all at once, these imbalances have worsened over the years, increasing the vulnerability of the Turkish economy.

This paradigm shift is most clearly seen in the rapid depreciation of the Turkish lira, with periods of overreactions around a general downward trend (see chart 1). Growth has been driven by a largely accommodative monetary policy, but this came at a price: a lasting surge in inflation in recent years, far above the levels seen in the first decade of this century.

Short-term adjustments had to be made on several occasions – in January 2014, August 2018 and November 2020 – through an abrupt tightening of monetary policy. The accelerated depreciation of the Turkish lira was fuelling both inflation expectations and gold imports, which strained the external deficit. Yet these bouts of adjustments did not result in another paradigm shift, and after a few months, monetary policy returned to its accommodative bias.

Over the same period, Turkey’s external deficit, which is structural, rarely vanished. Only the 2018 currency crisis forced Turkey to grow at a slower pace than its trading partners in 2019, bringing import growth more in line with exports growth. Even so, the current account deficit returned to high levels in 2020 (3.7% of GDP) due to the drop-off in tourism revenue.

With the outbreak of Covid-19 in March 2020, the Turkish authorities provided massive economic support, via fiscal and monetary policies as well as credit. But this short window was soon over. In August, the TRY began depreciating rapidly again and the policy mix had to be tightened. The central bank raised its key rates in November, and President Erdogan announced a new economic policy framework that was designed to make Turkey more attractive for foreign investors. He also said Turkey would return to a more rules-based policy mix (although without being more specific).

But for how long? Do these announcements foreshadow a radically different model? In this paper, we will try to describe what Turkey could be like in the next decade, depending on whether or not these pledges generate lasting structural change.

In the absence of change, the past situation can be the best predictor for the future. As we will see, the last decade has seen a shift from a rule-based philosophy towards a more complex strategy offering greater flexibility in terms of economic policy, but at the price of greater opacity and less regard for policy rules.

Monetary policy and inflation: a dangerous link

The monetary policy framework is very representative of the discretionary nature of economic policy. For the past several decades, Turkey has been hit by chronically high inflation, which has only fallen below 5% for four months over the past forty years (5% was the inflation target during the inflation targeting period, which guided monetary policy in the early 2000s). Chronically high inflation has taken its toll on the functioning of the Turkish economy.

Monetary policy has become increasingly heterodox over the past decade

Two crises pushed the Turkish authorities to opt for a more discretionary economic policy. The first was the combination of the great global recession of 2008 and the ensuing Eurozone crisis, which led to a change in the monetary policy framework. The second was the 2016 coup, which was followed by an easing of fiscal constraints (see below).

In the first decade of the century, the monetary policy framework was reformed to have an inflation targeting scheme, a strategy that proved its worth as inflation fell very sharply during the period, to a low of 6.3% in 2009 (down from an average of 79% in the 1990s, see chart 2). Yet with the increased volatility of capital flows, Turkey was regularly exposed to liquidity restrictions, which led it to switch to a new monetary policy framework as of 2011.

The new framework, which changed on 19 November 2020, was based on a broad interest rate corridor and discretionary liquidity management (which could change directions overnight, without the need for a monetary policy committee meeting) that drew to varying degrees on a variety of policy instruments (interest rates corridor). This led to an effective monetary policy rate (weighted average of the different instruments used) which could diverge sharply from the main policy rate (i.e. the central corridor rate).

The central bank also introduced a Reserve Options Mechanism (ROM). This mechanism can be used to require banks to hold a certain ratio of their reserve requirements for Turkish lira deposits in gold or foreign currency. The value of the ratio was able to change according to the central bank’s discretionary decisions. The central bank used the ROM to boost its foreign reserves (since these foreign currency holdings are included in it). This instrument is also countercyclical for banks, since required reserves in foreign currency are set aside during favourable periods and can be drawn on when foreign currency liquidity comes under pressure. In cases of highly volatile capital flows, the ROM covers the banks’ foreign currency liquidity needs to a certain extent, without requiring central bank support (it does not draw on its net foreign reserves, i.e. excluding ROM). This explains why Turkish banks accommodated the periods of highly volatile capital flows that Turkey experienced as of 2013, without too negative consequences. The ROM also helped gradually to decouple domestic credit fluctuations from capital flows, which clearly shows the new flexibility offered by this countercyclical mechanism (see chart 3), as in 2015, 2018 and especially 2020.

This new system played a key role in maintaining Turkey’s economic growth at a time when other emerging countries dependent on capital inflows, like Russia, Brazil and South Africa, were hit by an abrupt downturn in growth as of 2015-16.

Yet this strategy involved some risks. Less dependent on capital flows fickleness, credit growth was maintained at a very fast pace for a long time. As a result, non-financial private sector debt did not decline like it did in other countries. Inter-company lending also increased via longer days of sales outstanding (which were extended from 65 business days in 2007 to 76 days in 2019, according to Euler Hermes).

Although volatile capital flows became less critical for banks, capital inflows have not covered the current account deficit since 2014, which explains why the central bank’s foreign reserves are so low (see chart 4) and why the Turkish lira is under pressure. Clearly, capital inflows are still necessary, and apparently the government has finally decided that Turkey needs a new influx of them.

The link between TRY depreciation and inflation

A monetary regime characterised by severe currency depreciation tends to favour the de facto dollarization of the economy. This means that economic agents tend to set their prices in foreign currencies, even for transactions in the local currency. In the extreme case of full, de facto dollarization, currency depreciation is carried over 1 for 1 to inflation. This makes it very hard to control inflation.

We have formulated a model that characterises the link between foreign prices expressed in the local currency (foreign prices in USD multiplied by the exchange rate, labelled f in the formula below) and domestic prices (p). We will use the following simplified formula (log-linearized, with the constant ln(µ)):

p = ln(µ) + ß f

A statistical analysis shows that there is co-integration, which implies the use of an error-correction model to show the dynamics of the pass-through effect of foreign price increases on Turkish inflation over time. As Garcés-Diaz (2001) and Colliac (2006)^[1], we used the following formula:

dp_t = F(p_t-1 – ßf_t-1 – a) + dp_t-1 + ?df_t + e_t

This formula can be used to measure both the intensity and rapidity of the pass-through effect by calculating the month-on-month impact of an initial 10% increase in international prices expressed in TRY. The coefficient F, attached to this initial shock dft gives the instantaneous value of the pass-through effect. The long-term part of the equation pt-1 – ßft-1 – a and past inflation dpt-1 add additional inflation as of month T+1, and will continue doing so in the following months until cumulative inflation reaches the value of coefficient ß of the long-term relationship. The negative value of coefficient ? associated with the long-term relationship guarantees convergence, which works for all of the time periods for which the extent of the price pass-through effect is estimated.

We define four periods (1980-89; 1990-2002; 2003-2012; and 2013-2020). The first two ones can be characterised as decades of high inflation and repeated currency crises. The third period is marked by sharp disinflation and corresponds to the period of inflation targeting. The fourth period starts with the Fed’s announcement of “tapering”, which was marked by less USD liquidity and by investors’ heightened sensitivity to the macroeconomic imbalances in the emerging countries.

With the exception of the period 2003-2012, when the Turkish lira became more stable (except during global crises), in all of these periods, TRY movements were predominant over those of international prices in USD in the formation of international pricing trends expressed in TRY. Currency depreciation began playing a wider role again in the last period.

The extent of the pass-through to inflation, which was unitary in the long-term, was very rapid in the first two decades under consideration. The average impact of the shock corresponds to the month in which at least half of the impact on inflation has materialised, i.e. 3 months in both decades (see table 1). Consequently, it was impossible for Turkey to have a monetary policy, since Turkish prices were responding fully and rapidly to any increase in foreign prices.

In the 1990s, numerous emerging countries – but not Turkey – entered into a disinflation period, which was partially driven by smaller degrees of pass-through from depreciations. This trend did not reach Turkey until 2003. In the decade that followed, a 10% depreciation generated only 5.9% inflation in the end, with a longer transmission period: the average duration of the shock increased to 14 months.

Starting in 2013, the situation began to deteriorate, not so much due to the total impact on inflation (6.2% for a depreciation of 10%) or the average duration of the shock (still 14 months), but to the higher initial impact. Chart 5 shows the cumulative impact of a 10% depreciation on these two periods: whereas in the 2000s, the impact is 1.1% after two months, it rises to 1.9% over the same period of time after 2013. This illustrates the greater rapidity in the adjustment of domestic prices expressed in lira. This more rapid increase has the potential to trigger a new depreciation movement, which means that monetary policy loses part of its autonomy.

The growth/risk trade-off has deteriorated

There is not just one way to obtain growth. Yet the search for higher growth rates is often accompanied by bigger imbalances.

Less and less growth

Against the background of economic support policies and the risks they engender, Turkey has been able to foster significant growth. Yet the repeated shocks of 2016, 2018 and 2020 have taken their toll. The determinants of growth can be analysed to identify the shortfalls that led to a more boom-bust growth profile (see chart 6).

An accounting analysis of Turkish growth shows that it was based mainly on accumulation factors in the most recent decade^[2] (see chart?7). The investment ratio was maintained at a high level, enabling significant growth of the capital stock, which is one of the fundamental factors of long-term growth. Employment also rose strongly, reflecting an increase in the working-age population and a higher participation rate in the formal economy. Unsurprisingly, the accumulation of production factors slowed sharply towards the end of the period, first in 2019 (in the wake of the 2018 crisis) and then due to the Covid-19 pandemic.

Yet since 2013, growth has been obtained through “perspiration rather than inspiration”, to use Paul Krugman’s famous expression on Asian growth in the years that preceded the 1997-1998 crisis. In Turkey (as in Asia at the time) total factor productivity declined (see chart 8), despite growing investment in both physical and human capital.

One could assume that there is a causal link between the decline in total factor productivity and inflation. Although in the short term, inflation tends to follow nominal exchange rate changes, in the medium to long-term, the real exchange rate is the key determinant of competitiveness. In periods of high growth, this exchange rate tends to appreciate, but this is rather triggered by an accumulation of factors that creates scarcities (so inflation) than productivity gains. As a result, the competitiveness deteriorates and the likelihood of a depreciation increases, favouring additional inflation pressures. The patterns of the exchange rate and inflation trigger price formation in foreign currency.

When the investment rate is high but with low returns, it can also have the following implications:

From a strictly accounting perspective, the marginal efficiency of capital deteriorates: the incremental capital output ratio (ICOR) – i.e. the number of investment points necessary to reach 1 point of GDP growth – is high. In recent years, ICOR has increased in Turkey (see chart 9). Yet the generated investment must be funded and some of which is debt-financed. At some point, higher debt may have an impact on the level of interest rates, which may imply less private investment in the future.

In the short term, it seems very unlikely that today’s investments will generate sufficient exports in the future to reabsorb the external deficit. Granted, it is normal for a country to import when it cannot produce in sufficient quantity locally all of the goods it needs, particularly those with high value-added such as capital goods. When productivity increases, however, the skills of local companies also improve, and the country does not need to import as much, and can even become a net exporter (as in the Asian model, of which Vietnam is one of the most recent examples to date). If a virtuous circle is not created, it is still possible to restore the external balance, albeit through a decline in imports, which hurts domestic growth by restricting domestic demand. This is what happened in Turkey after the exchange rate crisis of August 2018.

Yet, capital was invested, and a lot of it. Between 2010 and 2018, investment in construction grew at an annual rate of 10% in volume, vs 5.7% for machines and equipment. This predominance, inherent in labour-intensive growth models (with such components as residential and commercial real estate), might explain the decline observed in total factor productivity.

At the same time, foreign direct investment diminished constantly over the past decade. It averaged only 25% of the current account deficit (and 15% in 2020), compared to 56% in the previous decade. Although Turkey continues to have a sizeable production base, including for foreign companies, the majority of these investments are relatively old, and recent investments are not high enough to increase visibly the capital stock.

Public finances: the new weak link?

Until recently, fiscal policy was considered as a mitigation factor in terms of risk, with a low public debt ratio, notably due to its high nominal growth rate (even though the share attributable to inflation has increased over time). Clearly, a moderate public debt ratio was a favourable factor in early 2020, which made it easier to support the impact of the Covid-19 crisis on public debt.

Yet fiscal policy had changed course as of 2016. The downturn in growth contributed to this policy change, but the deterioration in the fiscal balance was still striking, even after adjusting for cyclical factors (see chart 11). As a result, the fiscal deficit could only be kept rather small for a few more years with the support of nominal GDP growth. When growth contracted, the fiscal overruns became fully apparent.

A swelling public debt increased the interest expenditure (which is expected to hit 3% of GDP in 2021), which is a handicap for future fiscal consolidation efforts. The debt profile has also deteriorated:

The average maturity of public debt has shortened from 7 years in August 2018 to 5 years at year-end 2020, under the impact of the decline in the maturity of domestic debt from 4.2 years to 2.9 years (see chart 12).

The share of government debt in foreign currency has increased over the past decade (to nearly half), unlike most of the other big emerging countries, for which this share has diminished in recent years. In Turkey, the share of public debt in foreign currency has been bolstered in part by the depreciation of the Turkish lira (which increases the countervalue expressed in TRY).

The interest rate paid on public debt has increased, driven up by the increase in risk premiums, as illustrated in chart 13.

With the growing volatility of the Turkish lira, currency risk became more significant and had an impact on all these three factors. In Turkey’s case, it played at the same time on the maturity of public debt and its share in foreign currency: the country had only limited capacity to take on long-term debt in its own currency, but it was able to choose between short term debt in TRY and long-term debt in foreign currency. Consequently, these risks went hand in hand. As a result, the TRY yield curve is incomplete and not very liquid beyond the average maturity of domestic debt (about 3 years). This is a handicap for monetary policy, which needs to be able to play on the term structure of interest rates to be fully effective.

The increase in currency risk seems to be having a major impact on the interest expenditure, as shown by the breakdown of the premium on a 5-year government bond in TRY, compared to a USD Treasury bond with the same maturity. Currency risk has grown (risk premium on USD Turkish bonds compared to the equivalent in TRY), unlike the more limited role of counterparty risk (risk premium on USD Turkish bonds compared to the equivalent US government bond).

Moreover, based on recent trends in 5-year CDS premiums, part of the sovereign risk seems has been driven by the existence of a significant currency risk^[3]. Particularly, when the percentage of debt in foreign currency increases default risk becomes more sensitive to swings in the exchange rate. The removal of the central bank governor on November 7^th foreshadowed the monetary policy decision of November 19th, which was favourable for the stabilisation of the lira. After the removal, the default risk premium declined by 142 basis points (bp), even more than the currency risk premium (down 100 basis points on November 23^rd).

Turkey: Quo Vadis?

In economic policy, the policymaking process gives a meaningful idea of the objectives. There is abundant economic literature on the relative advantages between rules and discretion, notably following the emerging market crises in the 1990s, which favoured the adoption of monetary and/or fiscal rules in the 2000s. But the paradigm changed with the relative success of the big emerging countries during the 2008 crisis: although activity declined, they were not hit by a financial crisis, nor did they need IMF financing. Greater resilience was then used to justify foregoing certain pre-existing rules.

Turkey began to tolerate greater fluctuations in terms of inflation and interest rates. Without returning to the extreme monetary fluctuations of the 1990s, economic cycles became more boom-bust: a ‘cold turkey’ approach^[4], – based on abrupt (and reversible) monetary policy adjustments – was preferred over gradualism. In practice, this resulted in several last minute, brutal key rate hikes of several hundred basis points, each of which followed periods of low interest rates despite high inflation. Using a more gradual approach coupled with a more proactive monetary policy, interest rates would not have had to be so high to obtain a given level of inflation.

Stop-go approach or gradualism, which will it be?

When we compare two recent periods in Turkey, the decade 2003-2010 provides a good example of gradualism while the decade 2011-2020 reintroduced a dose of discretion, as policy consolidation decisions were less automatic. Using the same typology as we used for the determinants of economic growth, we see two possible growth trajectories for the next decade^[5].

The first approach uses the same patterns observed during the last years. It is characterised by a series of stop-and-go, with accumulation factors’ (labour and capital) growth increasing each time the economic situation stabilises, and declining when the next shock occurs. The investment rate and the labour market participation rate are both correlated to the cycle. In this environment, productivity trends are closely related to the dynamics of the capacity utilisation rate: when capacity is underused, it is less productive, and vice versa. Given the boom-bust nature of growth, the composition of investment cannot be changed. And since Turkey is not up the value chain, the imports elasticity of growth remains high. As Aguiar and Gopinath (2007) indicate in their analysis of emerging country growth in the 1990s, rather than a trend, we see a succession of short cycles that tend to drain away any real tendencies (chart 14)^[6], which is the pattern followed by Turkey in this first approach.

Gradualism, in contrast, focuses on restoring an economic equilibrium using a policy mix that tries to correct any imbalances. It results in a decline in growth in the years following the post-Covid rebound. In other words, it tries to bring down the deficits and inflation, and even if it fails to do so in the very short term, the policy mix becomes more restrictive to make monetary conditions somewhat more neutral. The tightening of monetary policy implemented by central bank from November 2020 (with +675 bps rate hikes) may correspond to this strategy if it indicates a structural change to its reaction function to inflation changes. It means that cyclical policy should provide less support, and it takes longer for the reforms accompanying the shift towards an economic equilibrium to have a positive impact on growth. Even so, the reforms encourage greater foreign direct investment, which triggers an upturn in the growth of capital stock. A different type of financing to match a somewhat different type of investment, which sparks an upturn in productivity.

Low differences can be seen between the two growth regimes over the next 5 years: for different reasons, they both engender an average annual growth rate similar to that of the previous five years (around 3% per year). Since the stop-go profile sparks changes that are not durable, the average annual growth rate is basically the same in the period 2026-30, due to a series of growth cycles that are rapidly interrupted. In contrast, gradualism fuels a virtuous circle in which growth is more stable and accelerates. Although this approach is accompanied by the return to a rules-based framework, for both monetary and fiscal policy, it also means the Turkish lira is much more likely to stabilise. In the early 2000s, gradualism resulted in a real appreciation of the lira, although admittedly, part must be attributed to productivity gains generated at the time. Moreover, this trend was accentuated by widespread capital flows that were favourable for the emerging countries. This would have to happen again in the future, as the low share of foreign investors (3% of public debt in November 2020) makes it easier for them to become interested again provided policy is the right one. The resulting appreciation of the real exchange rate would generate a disinflationary gain that would strengthen this better balanced growth regime.

Impact on market potential of the Turkish economy

In recent speeches, President Erdogan has often repeated that one of his top priorities is to boost the country’s attractiveness for foreign investors. In 2020, foreign direct investment probably fell to an all-time low (USD 3 bn). At the same time, the share of imports in the Turkish economy has fallen rather sharply over the past five years (even after excluding oil and precious metals), as industrial output continued to grow (chart 15). The lira’s depreciation provided a strong incentive to substitute local inputs for imports. Yet this movement also strained the profitability of supply chains in industries that were heavily integrated internationally, introducing currency friction that did not exist for competing economies with more stable currencies. The clearest example is the automotive industry, where central Europe is now a serious competitor. On the whole, the degree of openness of the Turkish economy has declined.

The decline in imports also reflects the decline in the investment rate in the most capital-intensive sectors, such as capital goods. Of course, the boom-bust cyclical profile since the mid-2010s, and inflation’s erosion of household purchasing power, have also strained the cyclical sectors, including automobiles, metallurgy and plastics.

Even so, regardless of the period (the 2000s, 2010-2014 or 2015-19), the current account balance has continued to show a deficit, even though it has narrowed over the past five years compared to the previous 5-year period. The value of imports is still higher than that of exports: there has not been an upmarket shift. And yet this deficit has to be financed through capital inflows. This was the case through the mid-2010s, when demand continued to be financed through capital inflows. Thereafter, the lack of financing ended up draining the growth momentum.

Whichever growth regime prevails in the years ahead (stop-go or gradualism) will make a difference in terms of the level and sustainability of Turkey’s market outlets, as well as on the sector breakdown. Our two policy patterns for the next decade illustrate two relatively divergent outlooks, although reality is of course more complex and will fall somewhere in between the two.

By perpetuating a boom-bust growth profile, and given the lack of external financing in recent years, the stop-go scenario promises relatively mild and volatile import growth, with more limited development of heavy investments. At the same time, household demand would keep pace with the erratic nature of the economic cycle. Under this framework, Turkey would remain an industrial powerhouse, protected from factory loss in part by the size of its domestic market. Even so, its market positioning would remain roughly the same, and the growth of market outlets would mainly be concentrated in raw materials and basic goods (mainly metals) used in industry.

Gradualism offers a different model, one that promises a return to the stable growth of the 2000s built on a stable macroeconomic environment. This would trigger a new wave of foreign investment attracted not only by stronger local growth, but also by a better performance of Turkey’s exports, facilitated by this model. Consequently, almost all sectors would benefit from this momentum, but with this difference: it would be easier to finance import growth, and the most capital intensive sectors, like capital goods, would become the biggest market outlet again.

A different set of countries would see their market outlets grow in Turkey depending on the trajectory that is followed. Under the stop-go scenario, there would be few big winners, with the exception of commodity exporting countries. Over the past two years, Russia has become the leading exporter to Turkey, in front of the former one (China). During the early 2000s and in 2010-2014, it was the world’s biggest manufactured goods exporters that posted the strongest growth (China, Germany, the US, Italy and France). A priori, these countries would be the main beneficiaries if the economy were to return to stable growth, with financing that is less stop-and-go. This would allow Turkey to resume investment in machinery and equipment, and all of these countries have corporates capable to provide these kinds of goods.

****

The Turkish experiment and other examples show that the method plays a vital role in the capacity to change. The first scenario is based on shock treatment, with the risk that efforts will not be sustained. A stop-and-go approach to economic policy risks carrying over to growth, which could adopt the same profile. Reforms would not be sustained, and there would not be enough time to appreciate all their advantages. Using gradualism – a model that worked so well in the 2000s – would allow to get back on the road to sustained faster growth. Through institutional changes that are made to last, gradualism fosters growth and limit inflation, which is necessary to generate more stable growth in the medium term.

^[1] Garcés-Diaz D. (2001), Determinacion del nivel de precios y la dinamica inflacionaria en México, Cemla, Monetaria, Vol. 24 n° 3, pp. 241-270

Colliac S. (2006), Les conditions préalables à la dollarisation totale: application à l’Amérique Latine, thèse pour le doctorat ès sciences économiques, Université Montesquieu Bordeaux IV.

^[2] We use a Cobb-Douglas production function, a text book representation that links GDP with capital stock, employment and total factor productivity.

^[3] For more information, see Berg and Borensztein (2000), The Pros and Cons of Full Dollarization, IMF Working Paper n° 00/50; Garcia and Lowenkron (2005), Cousin risks: the extent and the causes of positive correlation between country and currency risks, Textos para discussão 507, Department of Economics PUC-Rio (Brazil).

^[4] The expression “cold turkey” refers to an abrupt break with or withdrawal from a previous policy commitment, or in this case, a kind of shock treatment. It contrasts with gradualism, a strategy based on a gradual adaptation. See Giamattei (2015), Cold Turkey vs. Gradualism - Evidence on Disinflation Strategies from a Laboratory Experiment» Discussion Paper V-67-15, University of Passau.

^[5] Growth is estimated over the next decade based on changes in capital stock, employment and total factor productivity using the Cobb Douglas function (estimated relationship between 1980 and 2017).

^[6] Aguiar, Mark, and Gita Gopinath (2007), Emerging market business cycles: The cycle is the trend, Journal of Political Economy 115(1):69-102