Selected Issue


Selected Issue

Potential Output1

After a decade of strong economic growth following the 2001-crisis, Turkey has experienced a slowdown in growth since 2012. A central question is to what extent this is a cyclical phenomenon or whether potential output growth has slowed. This Selected Issues Paper finds that after several years above 4 percent, potential output growth slowed during the global financial crisis and has remained lower since then. In the early 2000s, the main driver of potential output growth was total factor productivity growth. During the mid-2000s this was replaced by strong capital growth rates, and from 2008 also higher potential employment growth—mainly as result of growth in the female labor force participation rate. Going forward, potential employment growth and capital growth are expected to contribute less than during 2008–14, with capital growth likely constrained as access to financing from abroad could become more challenging. We therefore expect that potential output growth from 2015–20 will be lower at 3–3½ percent. Importantly, with labor and capital contributing less than from 2008–14, maintaining such growth rates, requires a return to positive total factor productivity growth and continued access to external financing to finance investment unless domestic saving increases. Reforms that can increase future potential growth include policies to increase the domestic savings rate and the low female labor force participation rate as well as reforms that allow the convergence of total factor productivity to the level of advance economies to speed up.

A. What is Potential Growth?

1. Following IMF (2015), we define potential output as the “level of output consistent with stable inflation (no inflationary or deflationary pressure)”. We use an economic definition of potential output as in IMF (2015), where potential output growth is estimated using an explicit theoretical economic model. This differs from methods that rely solely on filtering techniques of output data to determine trend output. We apply a simple model (Blagrave and others, 2015) that takes into account the relationships between inflation, unemployment and potential output, which thus gives a theoretical foundation for the potential output estimates instead of using a concept of trend output as a proxy for potential output.

2. Potential output growth is not necessarily sustainable output growth. Easy access to financing from abroad allows investment to temporarily exceed the domestic saving rate substantially (i.e., the current account deficit is large). This allows a higher capital stock growth, than what can be supported by domestic saving, and thus allows for higher potential output growth. However, this might not be sustainable if it comes with an unsustainable increase in external debt. In the absence of policies to actively increase domestic saving, correcting the external imbalance will likely imply an increase in the cost of external financing. As IMF (2014) shows, such a correction reduces the current account deficit mainly through lower investment. This matters for potential output growth as capital stock growth falls when the level of investment is reduced to a lower (more sustainable) level. It also implies that estimates of future potential growth, narrowly defined as above, will depend on an assessment of the access to external financing (the ability to run large current account deficits) and the level of domestic savings (which depends on policies). In the same way that temporarily easy access to external financing can allow potential output growth to exceed the sustainable output growth a domestic credit boom can also lead to periods of unsustainably high potential growth rates.

B. Estimating Potential Output

3. A multivariate filtering (MVF) approach is used to estimate past and future potential output growth. The model incorporates the relationship between the output gap (the deviation of output from its potential level) and inflation and (the Phillips curve) as well as the relationship between unemployment and the output gap (Okun’s law). For Turkey these relationships are estimated to be:


π is the inflation rate, ŷ is the output gap, U is the unemployment rate, and Ū is the structural unemployment (the non-accelerating inflation rate of unemployment, NAIRU). A higher level of inflation or unemployment below NAIRU will thus be associated with a positive output gap. In addition to the two equations above, the model includes stochastic processes for (the level and growth rate of) output, the output gap, and (the level and growth rate of) unemployment (see IMF, 2015 and Blagrave and others, 2015 for details). Using data for 1998–2014 and near and medium term staff projections through 2020 for output, inflation, and unemployment, the model gives estimates of past and future potential output and NAIRU.

4. The estimates show a slowdown in potential output growth since before the global financial crisis. For 2007–20 the estimate of potential GDP growth using the multivariate filtering technique gives similar results as the trend GDP growth rates obtained from a simple Hodrick-Prescott (HP) filter using only data on real GDP. Both approaches show that potential output growth slowed to 3 percent in 2009, then rebounded to around 4 percent from 2010–13 (see text chart), and slows to 3–3½ percent from 2014. However, estimates for future potential growth rates are highly sensitive to staff’s projections for actual real GDP growth, which show medium term growth of about 3½ percent. For the period 2003–06, the HP filter shows about one percentage point higher potential output growth rates compared to the MVF approach.


Estimates of Potential Output Growth


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Sources: Staff estimates

C. Main Potential Growth Drivers Since 2002

5. To determine the main drivers of potential output growth, a production function approach is applied. Potential output, as estimated in section B, is assumed to be a (Cobb-Douglas) function of total labor and capital input in production as well as their total factor productivity (TFP):


Y¯ is potential output and is obtained from the filtering technique discussed above. K is the capital stock, which is obtained from the OECD data.L¯ is potential employment. Ā is potential TFP, which is obtained as the residual from the production function. α is the share of capital in the production, which we set at 0.5. This is in line with the findings of e.g., Chen (2010) for developing economies and Ungor (2014), Altuğ and others (2008), Ismihan and Metin-Özcan (2009), and Tiryaki (2011) for Turkey. Potential employment is a function of working age population, labor force participation, and the structural unemployment rate (NAIRU):


where Ū is the NAIRU, W is working age population, and LFPR¯ is the trend labor force participation rate. NAIRU is estimated as part of the estimation of potential output using the MVF approach. Data for working age population and labor force participation are from the UN and OECD respectively.

6. In the early 2000s potential output growth was strong and driven by high TFP growth. From 2001 to 2004, potential output growth increased from 3 percent to 5 percent. Capital growth was low as result of a low level of investment in productive capital at around 9 percent of GDP2. And with the contribution from potential employment also limited in those years, the main driver of the higher potential growth was TFP growth, which contributed 3–4 percentage points. However, part of the reason for the strong TFP growth following the 2001 crisis was also that capacity utilization—which is not taken into account explicitly and therefore included in TFP—increased strongly from 70 percent in 2001 to 80 percent in 2004.

7. During the mid-2000s, potential output growth remained high but its composition shifted from TFP growth to capital growth. Investment in productive capital almost doubled as a share of GDP and reached 15 percent of GDP in 2005–06; capital growth went from contributing ½ percent in 2003 to over 3 percent in 2007–08. As the domestic savings rate did not increase during these years, the higher level of investment in real productive capital was financed solely from abroad—the current account thus went from balance in 2002 to a 6 percent of GDP deficit in 2006. With potential employment growth roughly stable from 2002–07, the sharp increase in the contribution from capital growth implies that the contribution of potential TFP growth slowed in the run-up to the global financial crisis.


Capacity Utilization Rate


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Source: Haver Analytics.

Potential Output Growth Decomposition


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Source: IMF staff estimates.

8. Potential output growth started to decline in 2006, declined further going into the global financial crisis, and has remained lower since—mainly as result of lower potential TFP growth. Investment, and thereby capital stock growth, quickly rebounded following temporary slowdown in 2008–09. Potential growth was further supported by strong growth in potential employment with labor force participation increasing from 46 percent in 2007 to 50 percent in 2013—this was the result of a sharp 7 percentage point increase in female labor force participation (and the female employment rate). With both capital and labor growing strongly on average from 2006–14, the slowdown in potential output growth can be exclusively attributed to declining TFP growth—as shown in IMF (2015) slower TFP growth since the global financial crisis is a general pattern for EMs (excluding China).

Potential Output Growth and Its Drivers, 2002-20

article image
Note: From the production function, In(Yt) - In(Yt-1) = [In(At) - In(At-1)] + α[In(Kt) - In(Kt-1)] + (1-α)[In(Lt) - In(Lt-1)], which implies that (approximately) %ΔYt = %ΔAt + α%ΔKt + (1-α)%ΔLt.

D. Future Potential Growth

9. Potential output growth is expected to average 3–3½ percent from 2015–20 under current policies. Potential employment growth is expected to contribute ½–1 percent to potential output growth, which is lower than the high growth rates in 2008–14 but higher than before 2008. Capital stock growth is expected to contribute about 2 percent to potential output from 2015–20, which is lower than during the years leading up to, during, and after the global financial crisis. With about 2½–3 percent contribution from capital and labor, potential TFP growth of ½–1 percent is necessary for Turkey to achieve a potential output growth of 3–3½ percent going forward. This is a significant increase from the negative growth rates after the global financial crisis but less than during the first half of the 2000s.

10. The slowdown in future potential employment is mainly the result of a slowdown in the labor force participation growth. As mentioned earlier, growth in potential employment contributed strongly to potential output growth after 2007 as result of increased labor force participation. To estimate future potential employment, we use the United Nations population projections to determine the growth rates of the future working age population. For future trend labor force participation, a cohort based model is used (Aaronson and other 2014 and Balleer, Gomez-Salvador, and Turunen, 2014, see also IMF, 2015). Labor force participation for age group, a, gender g, at time t is determined as:


where αa,g is the (age and gender specific) constant, Ia,t is a dummy that takes the value 1 if the individual at age a at time t was born in the year b (cohorts for 5 year intervals are used). βb,g thus captures the marginal effect on labor force participation of being born in year b (the cohort effect). Finally, a trend (linear and squared) is included (the effect of which are age and gender specific). The model is estimated separately for each gender with data from 1990 to 2014 for labor force participation for the age groups 15–24, 25–34, 35–54, 55–64, and 65+ years. For projections for future years, the trend is kept constant at its 2014 level. The labor force participation rate of different age group remains roughly constant after 2015, as it changes only as result of cohorts moving into different age groups. The strong upward trend in female labor force participation observed from 2007–12 is thus not assumed to continue after 2015.

Dependent Variable: Female Labor Force Participation

article image

Dependent Variable: Male Labor Force Participation

article image

11. The overall labor force participation rate declines slightly after 2017 as result of overall population ageing. The share of the 55+ year olds in the total population—who have a low participation rate—increases, whereas the share of the 15–34 year olds declines. Together with United Nations estimates of the future working age (15+ years old) population—which show a 1–1½ percent annual growth on average from 2015–20—and the NAIRU estimates from the MVF, this gives a more modest potential employment growth from 2015–20 than what was observed from 2008–14.


Trend and Actual Labor Force Participation Rates


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Sources: OECD, United Nations, and IMF staff estimates.

Potential Employment Growth Decomposition


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Source: IMF staff estimates.

Population Share of Difference Age Groups

(Percent of total population over age 15)

Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Sources: UN, World Population Prospects: The 2012 Revision.

12. Capital growth is expected to contribute less than during the last decade. After the increase of the investment rate in the early and mid 2000s—mainly financed through higher current account deficits—capital growth remained high until 2011 only interrupted by a brief slowdown in 2009. However, after investment peaked at 16 percent in 2011, it has stagnated and as a share of GDP slowed to less than 14 percent in 2014. Staff expects the investment rate to decline slightly as the 1 percent of GDP narrowing of the current account between 2014 and 2020 is not expected to be driven by increased domestic savings. Only if the external financing environment allows for an increased level of investment or if structural reforms or fiscal consolidation succeed in raising the domestic savings rate will the capital growth be able to contribute further to potential output growth.


Capital Growth and its Components


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Sources: OECD, TurkStat, and IMF staff estimates

Investment in Productive Capital

(Percent of potential GDP)

Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Sources: TurkStat; and IMF staff estimates.

13. TFP growth is assumed to increase compared to recent years but remain lower than during the early 2000s. As has been the case in EMs in general (excluding China) TFP growth has been on a declining trend since the beginning of the global financial crisis. We assume that TFP growth will increase from its negative contribution since the beginning of the global financial crisis but remain lower than during the rapid convergence experienced during 200–205. Turkey’s per capita income leading up to this episode was less than 10 percent of that of the United States and increased strongly to 20 percent by 2008 and has remained at that level since. While this gives some justification why the convergence speed is likely to be lower going forward than in the early 2000s, any estimate of future potential TFP growth is highly uncertain.

E. How Can Turkey Achieve a Higher Potential Output? And What Could Cause it to Slow Down?

14. Increasing female labor force participation could significantly increase potential output for several years. With a participation rate of 34 percent in 2014 (up from 25 percent in 2005), Turkey has by far the lowest female labor force participation rate among all OECD countries. If Turkey manages to increase the female labor force participation rate to the level of the second lowest OECD member (Mexico at 47 percent) over a 10-year period, this could boost potential GDP growth by about 0.6 percent in each of the next 10 years, which increases the potential growth rate to above 4 percent in the medium term.


Female Labor Force Participation Rate for OECD Countries


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Source: OECD.

15. Increasing investment could add to potential growth—however, this would require increased savings or a higher current account deficit. In spite of the large current account deficit, Turkey’s investment rate is not high compared to other emerging markets economies. Staff expects a roughly constant investment rate from 2015 to 2020. If the investment rate were to increase by 2 percent of GDP compared to the baseline projections, this could boost annual potential growth by about 0.3 percentage points on average from 2016–20 assuming that about 60 percent of the increased investment will be in productive capital.3 However, with the current account deficit projected to remain high in future years, any increase in investment is likely to require at least the same increase in domestic savings (see also SIP on increasing the private savings) as increasing borrowing from abroad would be challenging. Moreover, even if domestic saving increased it is more likely to lead to a reduction of the current account deficit than an increase in the domestic investment rate.


Potential Output Growth Under Different Scenarios


Citation: IMF Staff Country Reports 2016, 105; 10.5089/9781484338209.002.A004

Source: IMF staff estimates.

16. Reduced access to external financing (due to Fed tapering and/or re-assessment of country specific risks), could lead to lower future potential output growth. A current account deficit of [2.9 percent] stabilizes Turkey’s negative net foreign asset position according the Fund’s external balance assessment (EBA). However, this is based on a medium term growth rate of 3.5 percent. As shown by IMF (2014), if the current account adjusts through restricted access to foreign funding (a higher external financing premium), lower investment would be the main source of adjustment. Lower investment leads to lower potential growth, which implies that a lower current account deficit can be sustained. The necessary reduction of the current account deficit to stabilize net foreign assets, if the adjustment happens solely through reduced investment, is about [three] percent of GDP. Such an adjustment would lead to an annual potential growth rate that is 0.5 percentage points lower on average from 2016–20.

17. A return to positive total factor productivity is necessary to maintain potential growth rates above 3 percent. Since the beginning of the global financial crisis, TFP growth has contributed negatively to potential growth. If TFP growth from 2016–20 does not return to ½–1 percent growth as assumed, but instead stays at zero growth, and at the same time access to financing disappoints, the potential growth rate could fall to just over 2 percent on average from 2016–20. With per capita income of around $10,000 Turkey still has potential to maintain growth rates well above those of advanced economies. Reforms to speed up this convergence process could help increase potential output growth. The 10th development plan lays out some broad goals for this such as increased spending on R&D, infrastructure, education, and innovation, etc. However, these policies are yet to be implemented and the extent to which it is possible to significantly boost total factor productivity growth remains uncertain and will likely take time.

F. Policy Implications

18. The slowdown in economic growth in recent years is partly a result of lower potential output growth. Estimates of potential output growth show a slowdown since the mid-2000s and lower realized economic growth is thus driven mainly by a slowdown in potential growth rates. Output for 2014 is (in staff’s view) close to its potential level and the persistent high inflation in an environment with declining commodity prices also indicates that the output has not fallen significantly below its potential. In addition, with the current account still elevated, higher growth rates will be unsustainable as they would lead to further build up of the external imbalance.

19. The structural slowdown can only be addressed through structural reforms to increase total factor productivity growth, national savings, and the female labor force participation rate. Cyclical stimulus to boost output from fiscal and monetary policy should be avoided as this would lead to unsustainable growth. Higher domestic saving—through policies to increase private or public saving—is necessary to help avoid that a correction of the external imbalance happens through a drop in investment. With inflation well above target, monetary policy should focus on bringing inflation back to target. Expansionary fiscal and/or monetary policies should only be considered if the economy falls into outright recession, with output falling to a level significantly below potential. Addressing the recent slowdown therefore calls for policies to increase potential growth. While this is challenging, policies to boost female labor force participation, increase domestic saving to allow for a higher (or at least maintain the current) investment rate, and structural reforms that help faster convergence of total factor productivity provide opportunities to increase Turkey’s growth potential.


Prepared by Uffe Mikkelsen.


As a measure of productive capital we use private and public investment in machinery and equipment from the national accounts data (i.e., construction investment is excluded).


Investment in productive capital was 60 percent of total investment on average from 2012-14 (and also around 60 percent in most years for the last 10 years).

Turkey: Selected Issue
Author: International Monetary Fund. European Dept.