Will History Repeat Itself? An Assessment of Turkish Current Account Trends and Prospects Cevdet Akçay & Murat Üçer December 2, 2005.

Will History Repeat Itself?

An Assessment of Turkish Current Account Trends and ProspectsCevdet Akçay & Murat Üçer December 2, 2005

Outline

Why we worry at all, and a few facts and propositions on the current accountSpeculations on the future, taking history and convergence as referenceSome empirical observationsConclusions: Don’t worry be happy?

Current Account Balance and Capital Flows(as % of GNP)

-8

-6

-4

-2

0

2

4

6

8

10

1989Q4 1992Q2 1994Q4 1997Q2 1999Q4 2002Q2 2004Q4

Capital Account Balance (Inc. Net E&O)

Current Account Balance

Why worried? Because of history…

…and because of conventional wisdom

“What are sustainable rates of real appreciation or of current account deficits and what invites a crisis? ...it is safe to say that a rapid real appreciation – say over 2 or 3 years – amounting to 25 percent or more, and an increase in the current account deficit that exceeds 4 percent of GDP, without the prospect of a correction, takes a country into the red zone”

Excerpt from “A Primer on Emerging Market Crises”R. Dornbush, June 2001

Current account deficit will likely continue this way in the near term; financing composition and prospects are improving, but still mostly debt and unidentified inflows…

Current Account Balance(12-month rolling; in US$ billions)

-25

-20

-15

-10

-5

0

5

10

15

20

Jan-00 Sep-00 May-01 Jan-02 Sep-02 May-03 Jan-04 Sep-04 May-05

Current Account Balance (exc. energy bill)

Current Account Balance (inc. energy bill)

Capital Account (as % of GNP)

-15

-12

-9

-6

-3

0

3

6

9

12

1992Q4 1995Q2 1997Q4 2000Q2 2002Q4 2005Q2

Net Errors&Omissions

Direct Investment

Equity Portfolio

Debt Securities

Other Portfolio

Official Loans

Private Loans

But sustainable only under fairly benign assumptions while decline in “measured” debt reflects temporary factors, measurement issues, and real exchange rate appreciation

Basic sustainability arithmetic:

Non-Interest Current Account Balance

+ Non Debt Creating Flows >

(Real Foreign Interest Rate – Real

Appreciation – Real Growth Rate)

X Initial Net Debt Ratio

Assumptions:NICA = -5%NDCF = 2.5%Interest rate = 4%Real growth = 5%Real appreciation = 0%Initial net debt ratio = 30%

External Debt Dynamics(as % of GNP)

0

10

20

30

40

50

60

70

80

90

100

2000 2001 2002 2003 2004 2005Q2

Public

Private

Total

The widening in the deficit was driven by investment, and the corporate sector; good news, with caveats…

Saving and Investment Balances: Total(as % of GNP)

10

12

14

16

18

20

22

24

26

28

30

1999Q1 1999Q4 2000Q3 2001Q2 2002Q1 2002Q4 2003Q3 2004Q2 2005Q1

Investment

Saving

Saving&Investment Balances: Private vs. Public(as % of GNP)

-25

-20

-15

-10

-5

0

5

10

15

20

25

1999Q1 1999Q4 2000Q3 2001Q2 2002Q1 2002Q4 2003Q3 2004Q2 2005Q1

Public

Private

Overall Balance

And given the international backdrop, policy response could have been hardly different…

Public Sector Primary Balance (as % of GNP)

-5

-4

-3

-2

-1

0

1

2

3

4

5

6

7

8

1999 2000 2001 2002 2003 2004 2005e 2006f

Central Government

Rest of the Public Sector

Public Sector Primary Balance

Base Money and O/N Interest Rate

-15

-10

-5

0

5

10

15

20

25

30

35

40

Jan-02 Jun-02 Nov-02 Apr-03 Sep-03 Feb-04 Jul-04 Dec-04 May-05 Oct-05

0

10

20

30

40

50

60

70

Net Domestic Assets (left scale; billion NTL)

Net Foreign Assets (left scale; billion NTL)

O/N Interest Rate (right scale; %)

*Note: The sum of NFA&NDA make up the base money.

Speculations on the future, taking history and convergence as reference…How does the latest boom compare to history?How does Turkey compare to convergence economies when they started the chapter-by-chapter negotiations?

Total Investment

Total Saving

Total (S-I)

Private (S-I)

Public (S-I)

GDP Growth

Private Cons.

Private Inv.

Real Credit

1993-94

2 years* 3.85 1.10 -2.75 -1.33 -1.42 14.3 11.6 47.6 22.0

1 year* 3.26 0.41 -2.85 -0.45 -2.40 8.4 8.4 44.1 10.1

1997-98

2 years -1.67 0.11 1.78 3.87 -2.09 12.0 9.2 12.0 47.5

1 year 0.10 1.34 1.24 2.73 -1.49 3.3 -0.6 -1.7 4.0

2000-01

2 years 0.72 -4.84 -5.56 -2.24 -3.32 6.3 4.2 6.5 3.0

1 year 1.22 -2.90 -4.11 -5.44 1.33 8.6 5.6 16.4 15.9

2004-?

2 years 3.07 0.11 -2.96 -13.22 10.26 19.2 23.6 88.8 109.3

1 year 0.39 -0.88 -1.27 -4.63 3.36 4.2 4.4 15.8 31.4

Avg (2 years) ... ... ... ... ... 8.7 8.1 16.6 10.8

Avg (1 year) ... ... ... ... ... 4.2 4.0 8.0 5.3

* Years before the current account reversal.

Current Account Reversal Episodes: Past and Present

Investment-Saving Balances (%) Growth, Private Absorption, and Credit(as % of 4-quarter rolling GNP) (quarter-on-quarter; %)

Export Import US-10 RER ProductivityDebt Service

(as % of exports)

ST Debt * (as % of reserves)

External Debt* (as %

of GNP)

Terms of

Trade

Fed Funds

World Growth

1993-94

2 years* 23.8 41.7 -1.7 11.2 18.6 ... ... ... 4.4 -1.5 3.6

1 year* 13.2 36.6 -1.1 11.0 12.9 27.2 207.3 ... 2.7 0.0 1.1

1997-98

2 years 50.0 34.9 -1.1 12.5 5.0 ... ... ... 3.9 0.3 5.8

1 year 17.4 10.4 -1.1 7.4 -0.8 20.3 95.1 58.8 -2.5 0.0 2.7

2000-01

2 years 15.1 25.8 0.9 19.2 16.2 ... ... ... -13.4 1.7 6.4

1 year 13.7 19.6 -0.6 14.9 7.7 35.5 101.3 71.2 -4.2 1.1 2.3

2004-?

2 years 22.7 44.9 0.5 14.9 18.6 ... ... ... 0.6 1.8 5.4

1 year 4.7 9.2 -0.4 8.5 3.8 36.0 80.3 50.5 -3.3 2.0 2.4

Avg (2 years) 22.6 28.1 … 7.3 14.1 … … … 0.4 … 6.8

Avg (1 year) 10.7 13.2 … 3.6 6.8 … … … 0.2 … 3.4

* Years before the current account reversal.

(quarter-on-quarter; %) (quarter-on-quarter; %)

Current Account Reversal Episodes: Past and Present

Exports, Imports, RER, and Productivity External FactorsVulnerability Indicators

Before After Before After Before After Before After

Poland 15.1 9.5 6.8 4.4 9.0 0.7 19.4 14.7Hungary 18.3 12.1 4.6 4.5 4.3 0.8 16.9 13.2Czech Republic 8.5 6.4 -0.7 0.0 9.0 6.7 7.7 6.3Slovak Republic 6.1 8.6 1.5 2.9 6.6 8.1 14.4 7.5Romania 45.8 52.5 -1.2 3.9 -7.0 11.0 74.2 47.0Bulgaria 2.6 10.6 2.3 4.8 5.8 6.9 3.2 3.0

Turkey 7.4 ... 5.0 ... 16.6 ... 16.4 ...

Before After Before After Before After Before After

Poland -3.7 -5.8 -0.6 -1.8 -1.6 -0.8 10.5 11.7Hungary -4.5 -7.5 3.6 -1.1 -4.5 -4.9 8.7 7.4Czech Republic -6.4 -2.3 -4.2 5.9 -0.9 -1.5 4.8 7.6Slovak Republic -5.7 -1.7 -2.1 3.2 -3.2 -3.1 16.2 18.9Romania -3.6 -4.6 -0.8 -1.8 -16.9 -34.1 11.3 10.1Bulgaria -5.0 -6.4 1.2 0.4 1.5 1.3 16.0 18.3

Turkey -6.3 ... -4.5 ... -3.0 ... 9.5 ...

Source: IFS, Eurostat, World Development Indicators.

Note: Poland, Hungary Czech Rep. started negotiations on March 1998; Slovakia, Bulgaria, Romania in early 2000.For real exchange rate, "before" corresponds to cumulative appreciation from t through t-2; "after" correponds to cumulative appreciation from t+1 through t-1, where t is the year in which negotiations start. For all other series, "before" corresponds to t-1, "after" average of t and t+1.

Current Account Balance (as % of GDP)

Current Account Balance (exc. FDI; as % of GDP)

Fiscal Balance (as % of GDP)

Unemployment (%)

EU Accession Countries: Selected Indicators

Inflation (annual; %) GDP Growth (%) RER Appreciation (%) Interest Rate (%)

Turkey is more indebted and FDI outlook is less certain…

NIIP Excluding FDI (as % of GDP)

-40

-30

-20

-10

0

10

20

30

Poland Hungary Czech Slovak Turkey

We have conducted two empirical exercisesSearched for the Balassa-Samuelson effect in the dataLooked at the sort-run determinants of the current account in a vector autoregression setup

On the Balassa-Samuelson (B-S) Effect

Trade integration → higher productivity increases in the traded goods sector (TG S) → higher wages → (under the assumptions of full employment and perfect labor mobility across sectors) → higher wages in the non-traded goods sector (NGS) → No matching productivity increase in NGS → profitability concerns push up PNGS → CPI ↑ while (PTGS/PNG) ↓ → Real Exchange Rate (RER) appreciation given that the counter-party productivity differential is smaller.

ПNGS = ПTGS (common to all trading partners) + the rate of

depreciation (provided that the exchange rate is not constant) + a productivity differential measure reflecting the asymmetry of the productivity gains between the home country and the trading partners. Periods of prolonged appreciation due to periods of persistent productivity differentials OR periods of real exchange rate disequilibria?

Productivity Ratio(2002=100)

80

85

90

95

100

105

110

115

120

1989Q3 1991Q2 1993Q1 1994Q4 1996Q3 1998Q2 2000Q1 2001Q4 2003Q3 2005Q2

Real Effective Exchange Rate(as % of GNP)

60

80

100

120

140

160

180

1989Q1 1990Q3 1992Q1 1993Q3 1995Q1 1996Q3 1998Q1 1999Q3 2001Q1 2002Q3 2004Q1 2005Q3

REER

4 per. Mov. Avg. (REER)

Real Effective Exchange Rate vs. Productivity Ratio

Regression Output: Real Exchange Rate and Productivity Differential

Dependent Variable: RERLN_CPISample(adjusted): 1990:1 2005:2Included observations: 62 after adjusting endpoints

Variable Coefficient Std. Error t-Statistic Prob. C 5.1610 0.2033 25.3888 0.0000PROD_RATIO_LN 1.9627 0.4851 4.0460 0.0002PROD_RATIO_LN(-1) -1.2173 0.4820 -2.5257 0.0144DUM -0.1059 0.0353 -2.9994 0.0040AR(1) 0.8287 0.0821 10.0962 0.0000

R-squared 0.8093 Mean dependent var 4.7656Adjusted R-squared 0.7959 S.D. dependent var 0.1391S.E. of regression 0.0628 Akaike info criterion -2.6191Sum squared resid 0.2251 Schwarz criterion -2.4475Log likelihood 86.1906 F-statistic 60.4634Durbin-Watson stat 1.8582 Prob(F-statistic) 0.0000

Inverted AR Roots 0.83

RER Regression Inferences

The sum of the current and lagged productivity ratio coefficients is 0.746, thus over the long run a 1% Δ change in the productivity differential → a 0.746% Δ in the RER. For the whole sample period (1990Q1 through 2005Q2), the Δ in productivity differential is 30.7%, and the change in RER is 42.6%, implying that 53.8% of the Δ in the RER can be attributed to the Δ in productivity differential.The residual series from this regression is taken as an RER series that is cleared from effects of relative productivity changes, and thus encompasses all other drivers of the real exchange rate except for the productivity differential variable.The purpose is to gain insight into the impact of RER Δs on the current account deficit in the absence and in the presence of a Balassa-Samuelson effect.

On the Short Run Determinants of the Current Account

In the VAR framework, we included capital inflows, the real exchange rate(actual in one version and the residual in the other), output growth, and current account balance as endogenous variables.To the extent that a significant portion of the RER variation comes from factors other than BS, coefficients of the actual and the residual RER series should be roughly similar in the current account equations. Better use of VAR models are made through variance decompositions and impulse responses. We utilize both methods to gain insight into current account dynamics in general, and the role of real exchange rate in particular.

Variance Decomposition With Actual RER

Period S.E. INFLOWROLLNEW RERLN_CPI GNPGRWTH CAROLL

1 0.50470 11.75329 11.29438 9.27596 67.676362 1.08771 47.07753 16.17327 5.35560 31.393613 1.62187 53.43528 21.59098 3.42433 21.549414 2.00070 54.42295 25.84717 2.43786 17.292025 2.20218 53.20238 29.38726 2.01473 15.395646 2.27657 51.52776 31.90586 1.93086 14.635517 2.29745 50.60505 32.99190 2.03035 14.372708 2.31832 50.97212 32.70950 2.16042 14.157979 2.35162 52.14844 31.79162 2.22709 13.83285

10 2.38333 53.27852 30.98822 2.23170 13.50156

Cholesky Ordering: INFLOWROLLNEW RERLN_CPI GNPGRWTH CAROLL

Variance Decomposition of CAROLL:

Variance Decomposition With Residual RER

Period S.E. INFLOWROLLNEW RESID_DUM GNPGRWTH CAROLL

1 0.52450 15.95593 8.14417 8.61707 67.282832 1.12661 48.87406 9.31371 6.51905 35.293193 1.68273 53.40076 13.52149 5.51545 27.562294 2.02204 54.52912 14.39503 5.03067 26.045175 2.18133 54.36664 14.78839 4.96629 25.878686 2.22788 53.73622 15.26297 4.93185 26.068967 2.23412 53.46861 15.48195 4.90437 26.145078 2.25472 54.20191 15.21109 4.91704 25.669979 2.29934 55.62957 14.65318 4.94827 24.76898

10 2.34615 56.99822 14.12677 4.97654 23.89848

Cholesky Ordering: INFLOWROLLNEW RESID_DUM GNPGRWTH CAROLL

Variance Decomposition of CAROLL:

Impulse Responses With Actual RER

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to INFLOWROLLNEW

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to RERLN_CPI

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to GNPGRWTH

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to CAROLL

Response to Cholesky One S.D. Innovations ± 2 S.E.

Impulse Responses With Residual RER

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to INFLOWROLLNEW

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to RESID_DUM

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to GNPGRWTH

-1.5

-1.0

-0.5

0.0

0.5

1.0

1 2 3 4 5 6 7 8 9 10

Response of CAROLL to CAROLL

Response to Cholesky One S.D. Innovations ± 2 S.E.

On the Short Run Determinants of the Current Account

Coefficients of the actual and the residual RER series did come out to be similar, implying that a significant portion of the real exchange rate variation could be attributed to factors other than BS. For both set ups, variance decompositions clearly show that the current account is mostly a capital account driven phenomenon.To the extent that the real exchange rate has any impact on the current account balance, roughly half of that stems from that portion of the change in the real exchange rate that is not linked to productivity differentials. Impulse response functions also indicate that capital inflow shocks and shocks to both versions of the RER variable have an impact on the current account, the order reflecting the strength of impact very much in the spirit of variance decomposition inferences.

ConclusionsTurkey now is very similar to the convergence economies then; plus we detect some encouraging differences from earlier boom episodes, in addition to flexible exchange rate and a stronger banking sector.

Yet, we are still in transition. Our VAR exercise confirms a history of mainly capital

account-driven current account adjustments, which typically entailed a “correction” in the real exchange rate.

The B-S regressions suggest some overvaluation in this latest episode as well.

But an “old-style reversal” seems highly unlikely as long as EU momentum is maintained, international environment remains benign, and strong fiscal policy is supported by structural reforms.