Turkey: Selected Issues

This Selected Issues paper on Turkey reviews a decline in private saving and adoption of a fiscal rule. As economic confidence has improved, capital inflows have surged while private saving has fallen. Together, these developments have produced current account deficits and a strong lira, raising concerns that Turkey may be exposed to sudden reversals in investor sentiment. A fiscal rule with appropriate mechanisms to monitor its compliance should deliver predictable policies, thus allowing further reductions in debt and risk premiums, thereby contributing to macroeconomic stability.

Abstract

This Selected Issues paper on Turkey reviews a decline in private saving and adoption of a fiscal rule. As economic confidence has improved, capital inflows have surged while private saving has fallen. Together, these developments have produced current account deficits and a strong lira, raising concerns that Turkey may be exposed to sudden reversals in investor sentiment. A fiscal rule with appropriate mechanisms to monitor its compliance should deliver predictable policies, thus allowing further reductions in debt and risk premiums, thereby contributing to macroeconomic stability.

II. safe to save less? assessing the recent decline in Turkey’s Private Saving Rate1

A. Introduction

1. The recent large decline in the private saving rate is one of the key domestic trends underlying Turkey’s widening current account deficit. Over the last five years, the aggregate private saving rate, as conventionally measured,2 has fallen by more than 20 percentage points of GDP (Table 1). Together with the simultaneous rise in investment, this has driven a sharp deterioration in the external current account, despite a considerable increase in public saving over the same period. The sharp fall in private saving stands out against comparator countries (Figure 1), and at its current 11 percent, Turkey’s private saving rate is low by international standards. Turkey’s national saving rate also falls into the low range, while its investment rate hovers closer to the cross-country average (Figure 2). The resulting reliance on foreign saving exposes Turkey to adverse shifts in external financing conditions.

Table 1.

Turkey: Changes in Saving and Investment Between 2001 and 2006

(Percent of GDP)

article image
Source: IMF, World Economic Outlook.
Figure 1.
Figure 1.

Turkey: Evolution of Saving and Investment in International Comparison, 1995–2006

(Percent of GDP)

Citation: IMF Staff Country Reports 2007, 364; 10.5089/9781451838237.002.A002

Source: IMF, World Economic Outlook.1/ Average PPP-adjusted GDP per capita in 2000–2005 within US$ 1,500 of Turkey’s.
Figure 2.
Figure 2.

Turkey’s Saving and Investment Rates in International Comparison, 2006

(Percent of GDP)

Citation: IMF Staff Country Reports 2007, 364; 10.5089/9781451838237.002.A002

Source: IMF, World Economic Outlook.1/ Average PPP-adjusted GDP per capita in 2000–2005 within US$ 1,500 of Turkey’s.

B. Accounting for the Recent Fall in Private Saving

2. The sharp fall in private saving since the 2001 crisis is arguably linked with other important trends that have marked Turkey’s macroeconomic stabilization. The remarkable post-2001 recovery of the Turkish economy has featured strong GDP growth and disinflation. This has driven down real interest rates, boosting confidence at home and abroad—strong capital inflows have caused the lira to appreciate—and underpinning a combined asset and credit market boom. On the policy side, the sharp reduction in budget deficits has been another major achievement. Several of these coincident developments are candidate explanations for the observed fall in private saving. In particular, private saving might have declined because of improved income prospects arising from macroeconomic stabilization; “Ricardian” effects from higher public saving; relaxed liquidity constraints; wealth effects, including from an appreciated real effective exchange rate (REER); and lower interest rates.

3. Inflation-adjusted saving data confirm the basic trends seen in the unadjusted data. High inflation rates distort conventional measures of private and public saving. We therefore compute inflation-adjusted saving data (Box 1). Although these adjusted data show a more limited decline in private saving since 2001, they confirm the basic pattern (Figure 3). Thus, recent saving dynamics reflect more than a spurious effect of disinflation.

Figure 3.
Figure 3.

Turkey: Unadjusted vs. Inflation-Adjusted Measures of Private and Public Saving, 1980–2006

(In percent of GDP)

Citation: IMF Staff Country Reports 2007, 364; 10.5089/9781451838237.002.A002

Sources: World Bank Saving Database; Treasury of the Republic of Turkey; and IMF staff calculations.

4. Formal regression analysis is conducted to quantify the impact of the different determinants of private saving. Following De Serres and Pelgrin (2002), we use the dynamic error-correction model of Pesaran, Shin, and Smith 1999). The model specifies a long-run relationship between the private saving rate and its determinants but allows for gradual convergence and some short-run dynamics (Appendix I). Our choice of regressors is guided by theory and prior empirical work, although it must also cope with data limitations.3 Thus, we choose public saving, inflation, the real interest rate, per capita GDP growth, unemployment, terms of trade changes, REER deviations from trend, credit growth, and the old-age dependency ratio as regressors. As regressands, both unadjusted and adjusted private saving rates are considered (for data sources, see Appendix II). The sample comprises annual data from 1980–2005.

5. The analysis points to public saving, inflation, and growth as the key determinants of private saving. Essentially all of the coefficients have the expected sign (see Tables 2 and 3 for an overview of previous empirical studies), and several of them are significant (Table 4). The results are also remarkably similar across the two regressions for unadjusted and adjusted saving rates. In particular, a lower private saving rate tends to be associated with higher public saving, lower inflation, and lower growth.4 Negative terms of trade shocks, a high REER, low unemployment, low real interest rates, strong credit growth, and a high old-age dependency ratio would also contribute to a decline in private saving, although the corresponding coefficients are not statistically significant. Lastly, the private saving rate exhibits considerable inertia, adjusting only gradually to its equilibrium value. Table 5 shows what these estimates suggest about the origins of the recent fall in private saving: although Turkey’s strong post-2001 growth would have actually favored a higher private saving rate (consistent with some transitory income component), this factor was more than offset by the strong negative impact of disinflation and higher public saving. The appreciated REER also contributed in an economically (if not statistically) significant way.

Table 2.

Selected Empirical Panel Studies on Saving Rates 1/

article image

This table is an expanded version of Table 1 provided by De Serres and Pelgrin (2002).

Table 3.

Determinants of the Private Saving Rate in Previous Panel Studies

article image

The numbers refer to the different studies listed in Table 2

Table 4.

Turkey: Results for Baseline Regression (1980-2005) 1/

article image

White heteroskedasticity-consistent standard errors in parentheses. The significance level of coefficient estimates (two-sided t-distribution) is indicated by asterisks: *** 1 percent; ** 5 percent; * 10 percent.

Table 5.

Turkey: Contributions to the Change in the Private Saving Rate Between 2001 and 2005 1/

(Percentage points of GDP)

article image

Based on the regression results reported in Table 4

Indicates the impact of gradual error correction starting from the private saving rate “error” already present in 2001, i.e. the endogenous change of the private saving rate during 2001-05 absent any changes in explanatory variables from their respective 2001 values.

6. These results suggest that lower private saving rates chiefly reflect the improved outlook for economic stability, in general, and fiscal policy, in particular. Lower inflation arguably captures macroeconomic stability, whose positive effect on consumer confidence has been found in other studies to depress private saving rates. Likewise, the finding of a partial public-private saving offset matches previous results from the empirical literature.

C. The Case for Seeking a Higher National Saving Rate

7. Based on the regression results, Turkey’s current low saving rate may recover some ground as the income catch-up continues. High real GDP growth appears to have dampened the fall in the private saving rate, suggesting that continued strong growth may also help in reverting to a higher saving rate. This outcome would be consistent with the notion of a catch-up process during which improved macroeconomic conditions temporarily depress private saving rates while laying the groundwork for sustained higher income in the future. As the expected growth materializes, the saving rate would go up again, the current account balance improve, and the country begin to repay its accumulated external liabilities.

8. While this benign scenario suggests a hands-off approach, there may still be a case for policy action to raise national saving. In principle, the fluctuation of private saving rates over time should be a self-regulating process, based on the rational decision making of private agents. The main task for policymakers would thus consist in limiting distortions that bias private saving decisions and/or give rise to a socially suboptimal national saving rate. Such distortions arise, for example, from the taxation of saved portions of income and the returns thereon, from the provision of means-tested basic pensions, or from excessively loose fiscal policy. Overall, the extent of Turkey’s distortions does not stand out, except for the history of large public deficits. Nonetheless, Turkey’s national saving rate appears low, relative to other countries and to its long-run investment rate. A considerable academic literature suggests that this situation may be detrimental, whatever its precise causes.

9. In particular, the case for seeking higher national saving rates can be built on an apparent link with long-run economic development. Irrespective of the deep causes for a given rate of national saving, researchers have long documented a positive association with long-run growth. Although this linkage is open to different interpretations, several arguments would suggest that the causality may (also) run from saving to growth rates. First, in the presence of volatile international capital flows, national saving provides a reliable source of finance for investment and reduces the potential fallout from a “sudden stop.” From a different angle, Aghion, Comin, and Howitt (2006) argue that national saving serves as a catalyst for attracting critical FDI that spurs innovation and productivity growth. Lastly, Prasad, Rajan, and Subramanian (2006) hypothesize that strong reliance on external financing may erode competitiveness (through an overvalued currency) and thus worsen long-term growth prospects. This again underlines the value of domestic saving.

D. How Can National Saving Be Increased?

10. Fiscal consolidation remains a natural instrument to increase national saving, but greater weight should be given to expenditure measures. The findings of this chapter suggest that higher public saving is effective in raising national saving in Turkey, despite a considerable offset from the private sector. As noted above, this accords with the evidence for other countries, where authors have commonly estimated Ricardian coefficients below but sometimes close to 1. Interestingly, cross-country evidence also indicates that the private sector response may depend on the nature of fiscal consolidation. Specifically, both Edwards (1996) and Lopez, Schmidt-Hebbel, and Serven (2000) have found expenditure cuts to be more successful than tax hikes in containing a private saving offset. Although there is no conclusive evidence for Turkey on this point, it appears consistent with recent experience—largely revenue-based consolidation and a very high apparent private saving offset.

11. By contrast, tax incentives do not seem very effective in raising saving rates. Loayza, Schmidt-Hebbel, and Serven (2000) note the overall disappointing experience with special incentive schemes designed to promote private saving. The limited effectiveness is perhaps unsurprising, given the common finding—including in this chapter—of a small interest elasticity of saving.5 This is not to say, of course, that a reduction of antisaving distortions in the existing tax system should be disregarded. In fact, both Tanzi and Zee (1998) and Callen and Thimann (1997) point out the positive impact on saving of shifting taxation from income to consumption. In Turkey’s case, however, indirect taxes already play a very important role. Moreover, a positive impact on saving rates has to be weighed against other considerations, notably the distortion of labor-leisure choices, automatic stabilizer effects, and distributional objectives.

12. The envisaged pension reform has the potential to raise saving rates. The reform package currently under discussion would gradually reduce replacement rates for Turkey’s pay-go pension scheme. For the young generation, this change amounts to a negative wealth shock as entitlements shrink without a compensating fall in contribution rates. In response, individuals are likely to put aside additional private savings for retirement. Securing a positive impact on overall national saving requires, however, that the improved funding of the pension system not be offset by lower public saving elsewhere. This is, in fact, one of the key insights emerging from cross-country empirical evidence, including the studies cited in Bosworth and Burtless (2004).

13. Financial sector reform may also have a positive effect on saving rates under certain conditions. On the one hand, as Jappelli and Pagano (1994) have argued, the short-run effect is often negative because financial liberalization tends to facilitate access to credit and thereby reduce borrowing constraints. On the other hand, Li (2001) demonstrates for the particular case of mortgage lending that greater availability of credit resulting from mortgage liberalization may actually increase saving rates, because more households will start saving toward the purchase of a home (see Chapter V). Positive effects on saving could also arise from the further development of Turkey’s capital markets as households would gain access to a more attractive range of financial investment opportunities. Finally, there is promise in creating a more favorable institutional setup for private retirement saving plans, which currently play only a very limited role.

Adjusting Private and Public Saving Rates for Inflation

As first discussed by Jump (1980), national accounts data focus exclusively on revenues generated from current production flows and abstract from changes in net worth due to capital gains or losses. The data may, therefore, miss important reallocations of saving between the private and public sector arising from inflation. Specifically, for a given real return on financial assets, higher inflation raises nominal interest receipts but simultaneously erodes the real value of the underlying assets. However, only the rise in nominal interest payments is reflected in measured income. In the case of government debt, this asymmetry will cause private saving to be overstated at the expense of public saving in high-inflation periods. Likewise, it will lead to a spurious fall of the private saving rate in periods of disinflation, as witnessed in Turkey over recent years.

In light of this problem, we estimate inflation-adjusted private saving rates using time-series data on the stock of net domestic public debt. Although we are interested in adjusting the composition of national saving between private and public, we will abstract throughout from the possibility that national saving itself might also be mismeasured as a consequence of capital gains/losses vis-à-vis nonresidents. Thus, we are concerned only with the market value of net public debt held by residents. Following World Bank (1998), we make the further simplifying assumption that this is equal to the book value of net public debt denominated in domestic currency. The public sector consolidates the general government, including public enterprises, with the central bank, and hence their joint net domestic liabilities include the money base and exclude any net government debt held by the central bank.

The World Bank’s (1998) Private Saving Database provides time series for Turkey through 1994. For our study, we import the series for the domestic inflation adjustment (computed as outlined above) for the years 1980-94. For 2001-06, we calculate the adjustment accordingly. The intermediate period 1995-2000 is more problematic, because data for the required debt concept are not available. We thus estimate net domestic public debt by assuming that it corresponds to a certain share β of overall domestic public debt, for which a long time series of data is available. In the years preceding the end of the World Bank sample, β turns out to be fairly stable, close to 0.8. On this basis, we extrapolate for 1995-2000.

Despite the need for some approximation, the inflation-adjusted series likely provides a more accurate measure of recent movements in private versus public saving than the unadjusted data. In essence, the adjustment leads to a sizable downward revision of the private saving rate—mirrored, of course, by an equal upward revision of the public saving rate—and implies a more moderate decline since 2001 (Figure 3). Nonetheless, the basic feature of a marked fall in private saving remains.

Appendix

I The Regression Model

We start from the following unrestricted specification:

PSt=μ+λPSt1+βXt+γXt1(2)+ɛt,

where PS stands for the private saving rate, X is the vector of all explanatory variables, and X(2) is the subvector containing the variables that also enter with a lag, thereby generating additional short-run dynamics. The choice of variables to be included for the short-term dynamics is based on model selection criteria. Specifically, the Schwarz criterion suggests including only the lagged public saving rate. This relative parsimony also helps to preserve (scarce) degrees of freedom for the regression.

The above equation can be written in an error-correction form as follows:

ΔPSt=φ(PSt1ηθXt)+δΔXt(2)+ɛt,

where η=μφ;δ=γ;φ=(1λ) is the adjustment coefficient; and θi=βi/(1λ) is the vector of long-run coefficients for all regressors i except the public saving rate, whose coefficient is θJ=(βJ+γ)/(1λ).

II. Data Sources

The data used in the regression analysis are obtained from the following sources:

CEIC: Real effective exchange rate.

Data Insight: Nominal interest rate (on 12-month deposits).

IMF, International Financial Statistics: Claims on private sector.

IMF, World Economic Outlook: (Unadjusted) gross private and public saving; GDP deflator; terms of trade for goods and services; real GDP per capita; unemployment rate.

Turkish Treasury: Domestic debt (for computing the inflation adjustment, see Box 1).

World Bank, World Development Indicators: Population ages 15-64 and population ages 65 and above (for old-age dependency ratio).

World Bank, World Saving Database: Inflation adjustment for saving rates (see Box 1).

References

  • Aghion, Philippe, Diego Comin, and Peter Howitt, 2006, “When Does Domestic Saving Matter for Economic Growth?NBER Working Paper No. 12275.

    • Search Google Scholar
    • Export Citation
  • Beshears, John, James J. Choi, David Laibson, and Brigitte C. Madrian, 2006, “The Importance of Default Options for Retirement Saving Outcomes: Evidence from the United States,” NBER Working Paper No. 12009.

    • Search Google Scholar
    • Export Citation
  • Bosworth, Barry, and Gary Burtless, 2004, “Pension Reform and Saving,” Working Paper, The Brookings Institution.

  • Callen, Tim, and Christian Thimann, 1997, “Empirical Determinants of Household Saving: Evidence from OECD Countries,” IMF Working Paper 97/181.

    • Search Google Scholar
    • Export Citation
  • De Serres, Alain, and Florian Pelgrin, 2002, “The Decline in Private Saving Rates in the 1990s in OECD Countries: How Much Can Be Explained by Non-Wealth Determinants?OECD Economics Department Working Paper No. 344.

    • Search Google Scholar
    • Export Citation
  • Edwards, Sebastian, 1995, “Why are Saving Rates So Different Across Countries? An International Comparative Analysis,” NBER Working Paper No. 5047.

    • Search Google Scholar
    • Export Citation
  • Edwards, Sebastian, 1996, “Why Are Latin America’s Saving Rates So Low? An International Comparative Analysis,” Journal of Development Economics, Vol. 51 (October), pp. 544.

    • Search Google Scholar
    • Export Citation
  • Haque, Nadeem Ul, Hashem Pesaran, and Sunil Sharma, 1999, “Neglected Heterogeneity and Dynamics in Cross-Country Savings Regressions,” IMF Working Paper 99/128.

    • Search Google Scholar
    • Export Citation
  • Jappelli, Tullio, and Marco Pagano, 1994, “Saving, Growth, and Liquidity Constraints,” Quarterly Journal of Economics, Vol. 109 (No. 1), pp. 83109.

    • Search Google Scholar
    • Export Citation
  • Jump, Gregory, 1980, “Interest Rates, Inflation Expectations, and Spurious Elements in Measured Real Income and Saving,” American Economic Review, Vol. 70 (No. 5), pp. 9901004.

    • Search Google Scholar
    • Export Citation
  • Li, Xiaowei, 2001, “Mortgage Market Development, Savings, and Growth,” IMF Working Paper 01/36.

  • Loayza, Norman, Klaus Schmidt-Hebbel, and Luis Serven, 2000, “Saving in Developing Countries: An Overview,” World Bank Economic Review, Vol. 14 (No. 3), pp. 393414.

    • Search Google Scholar
    • Export Citation
  • Lopez, Humberto, K. Schmidt-Hebbel, and Luis Serven, 2000, “How Effective is Fiscal Policy in Raising National Saving?Review of Economics and Statistics, Vol. 82 (No. 2), pp. 22638.

    • Search Google Scholar
    • Export Citation
  • Masson, Paul R., Tamim Bayoumi, and Hossein Samiei, 1998, “International Evidence on the Determinants of Private Saving,” World Bank Economic Review, Vol. 12 (No. 3), pp. 483501.

    • Search Google Scholar
    • Export Citation
  • Ozcan, Kivilcim Metin, Asli Gunay, and Seda Ertac, 2003, “Determinants of Private Savings Behaviour in Turkey,” Applied Economics, Vol. 35 (No. 12), pp. 140516.

    • Search Google Scholar
    • Export Citation
  • Pesaran, M., Yongcheol Shin, and R. Smith, 1999, “Pooled Mean Group Estimation of Dynamic Heterogeneous Panels,” Journal of the American Statistical Association, Vol. 94 (June), pp. 62134.

    • Search Google Scholar
    • Export Citation
  • Prasad, Eswar, Raghuram Rajan, and Arvind Subramanian, 2006, “Foreign Capital and Economic Growth,” unpublished; Washington: IMF.

  • Tanzi, Vito, and Howell Zee, 1998, “Taxation and the Household Saving Rate: Evidence from OECD Countries,” IMF Working Paper 98/36.

  • Schrooten, Mechthild, and Sabine Stephan, 2005, “Private Savings and Transition: Dynamic Panel Data Evidence from Accession Countries,” Economics of Transition, Vol. 13 (No. 2), pp. 287309.

    • Search Google Scholar
    • Export Citation
  • World Bank, 1998, “Notes on Module 3 (Private and Public Saving) of the World Saving Database,” available via http://www.worldbank.org/research/projects/savings/data.htm.

    • Search Google Scholar
    • Export Citation
1

Prepared by André Meier.

2

Private saving is computed as a double residual, from balance of payments, investment, and public finance data. As a result, saving data inherit and probably compound the existing weaknesses of those data sources, such as the implausibly high level of inventory investment in Turkey’s national accounts. Moreover, a breakdown into household or corporate saving is not available.

3

In particular, we lack data on house prices (a possible source of wealth effects) and income expectations. The latter would be a valuable regressor, as modern consumption models imply an important role for expected future growth. Still, to the extent that recent high growth expectations derive from improved macroeconomic stability, it could be argued that their effect is at least indirectly captured in the regression. Another inevitable limitation of our exercise is the small number of annual observations available for the regression variables.

4

This statement is about statistical association and should not be construed as a causal attribution, given the joint endogeneity of the relevant variables. Instrumental variable (IV) techniques are not a promising solution, as valid instruments are elusive. Nonetheless, in trying to cope with one possible cause of endogeneity, i.e., measurement error in public saving data, we also ran IV regressions using lags of public saving as instruments. Although this increased standard errors, the qualitative evidence on the impact of public saving was unchange.

5

There is, however, some evidence that psychological aspects, such as the framing or packaging of retirement saving options, can have a sizable impact on individuals’ saving decisions. See Beshears et al. (2006).

Turkey: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Turkey: Evolution of Saving and Investment in International Comparison, 1995–2006

    (Percent of GDP)

  • View in gallery

    Turkey’s Saving and Investment Rates in International Comparison, 2006

    (Percent of GDP)

  • View in gallery

    Turkey: Unadjusted vs. Inflation-Adjusted Measures of Private and Public Saving, 1980–2006

    (In percent of GDP)