Managing Terms of Trade Shocks in Compatible Control Regimes

J.L.S. Abbey

(Executive Director)

Centre for Policy Analysis

CEPA, Accra

March 2001

Introduction

For developing countries in general, the problems posed in applying modern macroeconomics are severe because economic structures are so different. Financial markets are often virtually absent, many economies are small, open, and periodically hit by temporary trade shocks, and most of them are heavily regulated by government controls.

Furthermore, the lack of attention to institutional characteristics in the neo-classical approach and the absence of viable micro-foundations in the structuralists’ theories have tended to make the exchanges between the two schools polemical. Meanwhile, many developing countries have experienced dramatic macroeconomic events, and have embarked upon large policy experiments, in an alarming vacuum of comprehension. This vacuum has arisen because both schools of thought – the neoclassical and structuralists – are right: theory must be tailored to structure to be applicable, but an atheoretic approach is inadequate. Policy research is a must.

Ghana is classifiable as a controlled open economy – a small, open economy with a weak financial market, subject to a variety of government controls and liable to ‘temporary’ shocks in terms of trade. In 1998, Ghana experienced a positive terms of trade (TOT) shock equivalent to a 26.5 percent improvement in the barter terms of trade. The question of interest is how did the government ‘manage’ the windfall? Did it tax it? Did it increase public expenditure, and if so, what? Faced with a shock of comparable magnitudes (actually somewhat bigger in 1977) on account of the Brazilian frost of 1975 which caused a large but temporary increase in the price of coffee, "the governments of Kenya and Tanzania arrived at radically different answers" to this question:

In search of an appropriate lesson for Ghana and guide to policy as to ‘best practice’ for managing shocks, the temptation is to examine the experiences of these two countries with a view to assessing the efficacy of their respective policy responses. The two responses, however, are not directly comparable, owing to key differences in policies in place before the windfall. The differences constitute the less obvious but probably more important, type of policy issue, namely the way in which private behavior is constrained by government regulations which predate the windfall. For even if the private sector is permitted to receive the windfall, its responses will be shaped by the control regime.

Like the Kenyan experience, we will demonstrate that the policy response in Ghana was inconsistent and grossly sub-optimal. Not only were resources misallocated, but the regime was "particularly ill-suited to the efficient utilization of temporary windfalls, which are a recurring feature of the economy". The World Bank has in the context of the subsequent sharp reversal in the terms of trade noted the incapacity of the government to manage shocks, leaving the economy vulnerable to periodic shocks.

The Tanzanian policies have been described as "an example of an incompatible control regime". The central aspect of such policy configurations is that they are not sustainable and so must be "eventually changed". Indeed the particular policy combination adopted by the Tanzanian government "gives rise to a disastrous cumulative contraction in the economy" which may be referred to as an "implosion". The Ghanaian episode shares some similarities especially in the fiscal policy aspect with the Tanzanian, but differs critically in that there was a regime of universal price controls in Tanzania which is what makes that scenario an ‘incompatible control regime’.

The way in which controls shape private responses to windfalls which private agents receive is central to policy analysis. Controls restrict how private agents make use of windfalls; foreign exchange controls inhibit the acquisition of foreign financial assets; import quotas may restrict the volume of imports, or alter their composition; and interest rate ceilings may alter the volume and composition of investment.

Fiscal Policy Responses – Stabilizing Taxation

A relatively passive or laissez-faire public response is inappropriate for three reasons. First, permanent national income rises with the windfall, and so does tax revenue at existing tax rates. In the absence of a mechanism assuring an automatic rise in public expenditure, an arbitrary fall in the public sector deficit will result. On the other hand, a reasonable assumption is that there is a positive elasticity of desired public expenditure with respect to permanent national income. The Ghanaian experience at the time of the windfall was that the expenditure elasticity was, if anything, higher than the elasticity of tax revenues at existing rates – also positive. Some change in the level and/or structure of tax rates may be necessary. The size and shape of the public sector would no doubt change. The critical point, however, is that there can be no guarantee that these automatic changes will correspond to what is wanted. This is why the issue of the appropriate long-run level of expenditure and taxation design arises.

Second, the marginal cost of public funds may temporarily drop during the boom. The windfall can be likened to an economic rent, so that it can be taxed away with minimal disincentive effects. Thus to the extent that the windfall was both temporary and once for all, there were powerful arguments that a high proportion of the incremental income be taxed at source (see also Economic Reforms in Ghana).

Finally, appropriate fiscal response may be influenced by the nature and composition of public expenditure, particularly by the allocation between tradables and non-tradables on the one hand, and consumption and investment on the other. The sharp appreciation of the real exchange rate will imply some shift in the optimal allocation of public expenditure between tradables and non-tradables, more markedly in the short-run than in the long run. It may also have implications for the level of government spending, as opposed to its allocation.

The CEPA position implicitly assumed that cocoa farmers were likely to be over-optimistic in their expectations were the windfall allowed to accrue to them. This would imply that cocoa farmers might attempt to consume too high a proportion of the windfall, and the government must attempt to prevent this. Ideally, the government would raise a loan, returning the money when the windfall was clearly over. This was exactly the proposal in the CEPA call for a Cocoa Buffer Fund. In the event, through the mismanagement of the exchange rate, the cedi depreciated nominally by only 4% -- implying high real appreciation – and with producer prices unrevised, the cocoa windfall was shared between users of foreign exchange and the Government of Ghana (GOG). Indeed, the cocoa export tax reached a high point – the highest in recent times.

Inter-temporal considerations indicate that a stable deficit is unlikely to be optimal during a windfall. Moreover, even the direction of change in the deficit is ambiguous. What needs stressing, however, is that even if the government has more accurate perceptions than the private sector concerning the underlying instability, the obvious response is to disseminate the superior information rather than to act on the assumed ignorance of private agents as actually happened with the implicit rejection of the cocoa buffer fund.

But the main source of sub-optimality of the official policy stance is the implicit assumption that the monetary authority, the BOG was capable of exercising the required custodial function over windfall resources. This is a critical empirical question requiring substantiation. Indeed, it could be argued that it was the absence of this capacity that led to the choice of an inappropriate exchange rate regime. The resultant misalignment caused an import binge that damaged the import-competing sector. It also spawned an "over-borrowing syndrome" that caused considerable havoc when the exchange rate regime collapsed.

Foreign Exchange Controls

Under the Exchange Control Act, private citizens (unless otherwise authorized) cannot hold either foreign currency or foreign assets. In a context of strict application, private agents are compelled to accumulate and decumulate domestic assets. And in the absence of a bond market the only financial asset is money. The opportunity cost of using funds for a proposed investment project is therefore whichever is the greater of the return on the best other project available and the return on money – which will be assumed to be zero in nominal terms.

Figure 1: Savings/Investment Behavior in Periods of Commodity Booms and Bursts

savings/investments

NPV0 schedule : the pre-boom net present value (NPV) of project discounted at the opportunity cost of funds over the life of the project.

A transient boom or windfall would leave permanent income unrevised and hence increase savings. In the presence of exchange controls, the boom may so increase desired savings that there are not enough domestic projects with a positive net present value (NPV) to absorb them, the balance being an asset demand for money. Since, by assumption, the nominal return to this is zero, this indicates that the discount rate used to generate the NPV is zero during the boom phase. Exchange controls – prohibiting the legal acquisition of foreign financial assets – cause this fall in the domestic interest rate during the boom, shifting the schedule from NPV0 to NPV1 (see Figure 1).

The effect of this control is partly a dead-weight loss in the form of a misallocation of investible resources. NPV2 is the schedule corresponding to absence of such foreign exchange controls – citizens are free to hold interest-bearing foreign securities. The discount rate being higher, the NPV schedule is shifted downwards relative to controlled case NPV1.

The exchange controls cause a lowering the discount rate during the boom and thus induce a quantity ad of investment, which yields NPV equivalent to the shaded triangle (bcd) less than obtainable from foreign securities. This is the dead-weight loss existing from being constrained to invest in domestic assets only.

Additionally, there is a transfer – the unshaded area (adhg) from domestic lenders to domestic borrowers, and a further transfer – shaded rectangle (dcef) – from holders of money as an asset to the monetary authority (the BOG). This latter occurs because the BOG accumulates the foreign exchange equivalent of df, which it holds in a custodial role on behalf of private citizens and which is therefore a component of the change in the net foreign assets of the central bank (D NFABOG), with a corresponding increase in Reserve Money (D H) as cash held by the non-bank public sector (NBPS). This is demonstrated below.

Let W denote windfall income and h the propensity to accumulate financial assets. Then, under foreign exchange controls and in the absence of bonds, the demand for financial assets (hW) translates into an equivalent asset demand for currency by the NBPS. Consequently from the balance sheet of the BOG we obtain the following:

If further, domestic money banks (DMBs) are successful in meeting their desired lending targets – exclusively by consumption loans since, by assumption, there are no further investments with positive NPVs – then through the money multiplier (m) – assumed to be stable – the change in money supply (D M) is given by:

Thus, the impact effect of the boom, given foreign exchange controls, is:

• temporarily to raise the demand for money by hW, i.e., in addition to the increased demand for transactions balances;
• temporarily to raise money supply by mhW; and
• for the NFABOG to correspondingly increase by the foreign exchange equivalent of hW.

An excess supply of money (m-1)hW results which, assuming m is sufficiently large to ensure overall excess money supply, raises nominal prices of tradables (Tr) and of non-tradables (N). Under the floating exchange rate regime, the exchange rate depreciates by the extent of the increase in the nominal price of tradable goods. Both of these effects are temporary, being reversed as monetary assets are converted into real assets with the relative price (PTr/PN) unchanged. (Note that trade policy is assumed unchanged).

The management of this trajectory, however, poses two difficult judgments for the Bank of Ghana.

• first, it must be borne in mind that a portion of the increased demand for money, hW, is in respect of a temporary demand for financial assets and will be reversed; and
• second, with enough consumption lending by the DMBs the money supply would increase by mhW. Consequently, if m is large, then in spite of increased transactions demand for money, there might be an excess money supply at the initial price level.

It is important to note that in the floating exchange rate regime, if this eventuality of excess money supply occurs, the rate depreciates in spite of the temporary accumulation of international reserves by the BOG (D NFABOG = hW > 0)

Consequently, in the predetermined exchange rate regime – pegged or crawl – a possible ‘second best response’ may be for the BOG to depreciate, not necessarily appreciate, the nominal exchange rate although its holdings of international reserves are increasing. This response, which could be contrary to the exchange rate determining rule implicitly used by the BOG, is precisely what CEPA proposed in 1998. This could have not only eliminated the tax on cocoa farmers (on account of real cedi appreciation) but together with the proposed fiscal policy enabled the proposed Cocoa Buffer Fund to be created as cushion for the projected reversal in export earnings in 1999. Significantly, and above all, it would have halted the process of continued real appreciation of the cedi that was causing havoc in the tradables sector.

Over the period 1996 to 1998, the BOG intervened with exchange rate determination, selecting a path, which it attempted to maintain by means of reserve accumulation and decumulation. Given this policy stance, a temporary windfall could generate a divergence between the actual exchange rate and the equilibrium exchange rate. The divergence reflects differences between the policy rule of the BOG and the determination of the floating rate. Moreover, this divergence or exchange rate misalignment has consequences for the price level and the real economy. Since the BOG appeared to have been totally unaware of the custodial role that the Exchange Control law and other restrictions confer on it, it has found it difficult to accept responsibility for the exchange rate collapse. It claimed that the quantity of forex sold in 1998 was not much different than in the previous years 1996 and 1997.

Consider the exchange rate E = E bar

It may be noted that in the floating exchange rate regime, the two changes in the demand for money induced by the windfall are accommodated by a fall in the price of tradables as the nominal exchange rate appreciates.

In the fixed exchange rate case, however, accommodating any increased demand for money must now be through money supply increases. With a stable money multiplier, m, this, in turn requires an increase in reserve money (H) such that

The increase, D H, is achieved as private agents in aggregate exchange foreign exchange for cedis, the counterpart being reserve accumulation by the BOG (D H - D NFABOG). Thus, in contrast to the floating regime where all the increase in foreign financial assets is held by private agents, under the foreign exchange control assumption, they are held by DMBs. In the fixed rate regime the BOG acquires some of these holdings – specifically D M/mE bar US dollars (where E bar is the fixed exchange rate per unit of US dollar and m is the money multiplier).

To recap, the asset demand component of the increased demand for money is transient. The transactions demand component generated by higher permanent income, however, is permanent. Consequently, the transient component of the corresponding increase in Reserve Money is best considered as a temporary loan from private agents. The permanent component, of course, represents resources permanently available to the authorities. (Increases in foreign exchange equalization account are relatively transferred to GOG as non-tax revenues). Because of the fixed exchange rate assumption this resource transfer occurs in the early stages of the windfall, as the BOG becomes a custodian of part of the foreign financial assets that in the absence of exchange controls would have been temporarily accumulated by private agents.

Conceptually, therefore, the following components of official foreign exchange reserves can be distinguished:

• foreign exchange reserves acquired by the BOG directly or indirectly through taxations;
• foreign exchange acquired in exchange for cedis which will be held permanently by private agents for domestic transactions; and
• foreign exchange acquired in exchange for cedis, which will only be held temporarily by private agents.

Corresponding to the second, the BOG has only notional liability but in respect of the third, it has a liability, which it will only know about, and be able to date if it fully understands the intentions behind the current money demand of private agents. The information problem of the BOG is therefore to interpret changes in reserves by attributing them to temporary and permanent in the private sector, and to temporary and permanent changes in the budget deficit and exchange and trade controls. This problem is formidable. Private sector responses are complex, involving overshoots in the real exchange rate (RER) and in money demand. Furthermore, the BOG could at best discover changes in trade controls and the budget deficit after a lag. If it misinterprets the accumulation of reserves, it may transfer larger resources – from the exchange equalization account – to the GOG than warranted; or more likely under the managed float regime, intervene by more than appropriate, causing relative appreciation of the rate and a divergence of the actual path from the free float one.

Another consequence of the fixed exchange rate is that the price level rises during the boom. The price of tradables, being fixed, any rise in that of non-tradables raises the overall price level. The mechanisms to raise the money supply are via a payments surplus, and accumulation of a payments surplus is a flow adjustment. Since the rise in the transactions demand for money is immediate, however, the impact effect of the windfall is to create excess demand for domestic currency. This will in turn have a moderating effect on the rise in the price of N (and increased quantity of imports). This effect is gradually eroded as the payments surplus leads to monetary accumulation. Equilibrium is restored once the surplus has accumulated in excess of the desired trajectory of temporary foreign financial assets by the extent of the increase in the transactions demand for money. It is therefore possible that prices of N, and hence the price level, continue to rise for some time after their initial jump. In Figure 2 below, this price is represented by movement from E to C and thence to A.

A third consequence is that, because the exchange rate is fixed the DMBs can pay an interest on domestic bank deposits equal to the world interest rate without making any allowance for a depreciating exchange rate.

In sum, the fixed exchange rate implies that

• the BOG automatically acquires a custodian role for temporary holdings of windfall foreign exchange;
• there is a higher price level during the windfall, and a smaller initial increase in the relative prices of N; and
• there is a lower nominal domestic interest rate.

Under a managed float regime, the accumulation of reserves might trigger a revaluation of the exchange rate. The implicit policy rule of the BOG, which determined the rate, appeared related to reserve levels. Since an appreciation of the rate will occur automatically in a floating-rate regime, such a revaluation would in itself reduce the divergence implied by a fixed exchange rate regime.

The CEPA policy proposal thus involved possible asymmetry in response to reserve accumulation and decumulation. The political costs – as seen in 2000 – of devaluation is so high especially in an election year, that in April the Minister of Finance produced a policy package, which centered upon import and other controls. The apparent misinterpretation by the BOG of the international reserves buildups during the boom of 1998 created the illusion that a permanently higher exchange rate was sustainable. In CEPA’s view, the legacy of the windfall was an overvalued exchange rate. Which was clearly unsustainable when the TOT deteriorated the following year and international reserves at critically low levels. The associated sharp fall in the cocoa export tax compounded the budgetary imbalances and the continued accumulation of payment arrears.

In Figure 2, WW’ shows how perceived welfare – comprising expected unemployment effects, both short and long term, future rates of inflation and associated income distribution effects – rises with the short-run rate of inflation.

With the floating rate regime, the optimal point is A, with inflation OS. This is the ‘desired’ inflation rate. It also reflects a policy of ‘living with inflation’.

O R Q S Rate of Inflation

Under a fixed rate regime (assuming no structural changes, no autonomous capital movements, and no shifts in demand patterns or differences in productivity growth compared to the rest of the world that might make an inflation differential compatible with an unchanged balance of payments), the balance of payments can be in equilibrium only if the country’s inflation rate equals that of the Rest of the World or OR.

With inflation rates higher than OR, a balance of payments deficit would result. On the assumption that this yields a net welfare loss (interest payments, etc., exceed the value of extra reserves received) the welfare curve becomes BT. The optimal point on BT is C, with inflation OQ>OR but less the desired rate OS. Over time foreign exchange reserves decline or accumulate, accommodating debts build up; the curve BT swings down tending to be vertical; C moves towards B. In the process the ‘optimal’ and actual coincide at OR.