General Equilibrium Perspective on Twin Deficits Hypothesis: An Empirical Study with US Results

Abstract

 This study proposes an alternative analytical framework for testing the so-called twin deficits hypothesis from the general equilibrium perspective. The income–expenditure equilibrium takes both the behavioural relationships of the saving and investment into consideration. The empirical results of cointegration tests show that the US fiscal balance, current account balance, real income and interest rates (short- and long-run) are co-movement for the observed periods between 1970Q2 and 2011Q4. Also, the causality tests suggest that budget deficit does indirectly Granger-cause current account deficit via short-run interest rate and real income. The real income and interest rates variables determining the behavioural relationships are important in understanding the US twin deficit slogan. In short, the empirical results validate the twin deficits hypothesis in the USA. Some policy implications have been drawn, especially on the implementation of the “fiscal cliff” policy. This study also recommends portfolio balance approach for future twin deficits research.

Key Words:

• Budget deficit
• current account balance
• general equilibrium perspective
• income–expenditure equilibrium
• the USA

Jel Classification:

• F32

Acknowledgements

I would like to acknowledge two anonymous reviewers for their comments and suggestions. I also extend my gratitude to Evan Lau from the Faculty of Economics and Business, Universiti Malaysia Sarawak for the comments made on the crude draft of this study. However, all errors and omissions rest with me.

Notes

1. However, Miller and Russek (1989)'s study finds that a $1 change in the fiscal deficit eventually leads to roughly a $1 change in the trade deficit.

2. Assumed an economy is relatively open that will probably have domestic developments dictated by the foreign balance to a certain extent. Hence, the budget balance of a country will be affected due to the large inflows or through debt accumulations – a country will run into budget deficits (Reisen, 1998).

3. For example, Japan's current account surplus is stable during the 1990s, despite the country's sharply declining fiscal condition as the changes in private saving can offset changes in fiscal policy, leaving a country's current account balance largely unaffected (Bartolini & Labiri, 2006, p. 2).

4. They investigated the effect of fiscal consolidation on the current account. They examined contemporaneous policy documents, including Budget Speeches, Budgets, and IMF and OECD reports, to identify changes in fiscal policy motivated primarily by the desire to reduce the budget deficit, and not by a response to the short-term economic outlook or the current account. The results based on this measure of fiscal policy changes suggest that a 1% of GDP fiscal consolidation raises the current account balance-to-GDP ratio by about 0.6 percentage point, supporting the twin deficits hypothesis. This effect is substantially larger than that obtained using standard measures of the fiscal policy stance, such as the change in the cyclically adjusted primary balance.

5. For simplicity, Baharumshah et al. (2009) have assumed S = I or S – I = 0 for their reduced form twin deficits equation, CAD = b₁ + b₂BD + b₂INV (where INV is investment). Conversely, the twin deficits hypothesis is hold if the estimated coefficient of BD is statistically difference from zero at conventional levels, i.e. 10%, 5% or 1%.

6. Also, Baharumshah et al. (2009)'s study documents that investment has a noticeable impact on current account deficit from the causality results of five ASEAN countries.

7. The current account balance is usually referred to as the trade balance (net exports) that is, the balance between the value of exports and imports for a given period. However, the current account balance includes net services and net transfers – these particular sub-accounts is usually minimized and not considered due to the fact they normally represent a small fraction of the total (Ventosa-santaulària et al., 2013, p. 1319).

8. It is partially hold as Bartolini and Labiri (2006, p. 6)'s study that twin deficits hypothesis that a larger fiscal deficit leads to an expanded current account deficit by its effect on national saving and consumption.

9. In this context, three cointegrating relations are potentially highlighted by economic theories. The first is from the income–expenditure equilibrium about the current account determination, CAD = f(y, r, BD). The second relation is from the Keynesian theory on the interest rate determination, r = f(BD), and lastly, BD = f(y) from the Wagner's law. In view of the objective of the present study, this is interesting to work on the first cointegrating relation than of others which is offered by income–expenditure equilibrium.

10. Darrat (1989) finds that budget deficits have not caused significant changes in the long-term interest rate, but a reverse causality. Again, the empirical results from Darrat (1990) show that structural federal deficits and the corporate bond rate are not cointegrated. The present study does not examine the causal relationship between budget deficit and interest rates, but to illustrate the role of interest rates on current account balance from the both saving and investment relations.

11. The long-term interest rate is the long-term government bond yields (10-year), while the short-term interest rate is the 3-month or 90-day rate and yield (certificates of deposit).

12. We also computed the similar analysis for Equation (8) and find no cointegration between lnCAD, lny and rl (or rs). The dummy variable Dummy_BD is exogenously considered in the VAR system. The null hypothesis of none cointegrating relation is not rejected at 10% level of significance (trace test) with a p value of 0.80 (0.89) for rl (rs) as interest rate.

13. We consider only the trace test since “… it [the trace test] shows more robustness to both skewness and excess kurtosis in innovations that the maximal eigenvalue test” (Cheung & Lai, 1993, p. 326).

14. The dummy variable is statistically insignificant but it does not interprect no twin deficits. As illustrated by the income–expenditure approach (Section 3), either income (y) or interest rate (r) play a crucial role in validating the twin deficits hypothesis.

15. Their method allows non-causality test without pretesting cointegration either the underlying variables are cointegrated or non-cointegrated of an arbitrary order. Table 1 shows all of the underlying variables are non-stationary in levels, i.e. I(1). The details of this testing method are available from Toda and Yamamoto (1995). Methodologically speaking, their method is straight-forward by involving two steps. The first step is to identify the true lag length of k and the maximum order of integration (d_max) of the variables in the VAR system. An augmented VAR(k + d_max) is then estimated (by OLS estimator). The second step is about the computation of the Wald statistics for the restrictions on the parameters of the VAR(k) model. It is to reject the null hypothesis of “x does not Granger-cause y”. The VAR system includes the endogenous variables of lnCAD, lnBD, lny and rs (and rl). Given the quarterly data, k = 4 and d_max = 1 are specified for the augmented VAR.

16. The full computed results are not reported here, but they are available from the author upon request.

17. The impulse response functions are computed by Eviews statistical software with a VAR(4) due to the general rule of thumb that quarterly data are employed for the variables in Equation (7).