Repurchase Agreements (REPOs)

Ivy Aryee

Centre for Policy Analysis

CEPA, Accra

June 2001

Repos are essentially an instrument to enable the BOG to fine-tune the level of reserves in the banking system between the weekly auctions or OMOs. Under a repurchase agreement the BOG buys T-bills from the DMBs and they, in turn, are contractually obligated to buy these bills back repurchase them ― within a short period of time ― typically one day but within three days. The interest rate on these is fixed at between 0.5 to 1.75 percentage points above the 91-day T-bill rate established at the most recent auction.

The purchase of T-bills from the DMBs by the BOG injects reserves into the banking system thus increasing the stock of Reserve Money. When the DMBs close out the repo by repurchasing the T-bills, their deposit holdings at the BOG are debited, reducing DMB reserves and hence the stock of Reserve Money.

In a "reserve repo" the BOG sells T-bills from its own account ― at the regular weekly auction T-bills are sold on behalf of the Government Treasury. The accounts of the transacting DMBs at the BOG are debited as appropriate. Reserves of DMBs are reduced and in the process the stock of Reserve Money falls. When the BOG buys back these T-bills in the specified period ― typically one to three days ― reserves of DMBs and stock of Reserve Money correspondingly rise.

Thus repos and reserve repos alter the stock of Reserve Money on a temporary basis. This contrasts with the outright sales and purchases of T-bills at the OMO which effect more permanent changes in the reserves of DMB and the stock of Reserve Money correspondingly rise.

Thus repos and reserve repos alter the stock of Reserve Money on a temporary basis. This contrasts with the outright sales and purchases of T-bills at the OMO which effect more permanent changes in the reserves of DMB and the stock of Reserve Money. Changes in Reserve Money are used by the central bank to induce changes in money supply either to festivals like X’mas or to keep money supply in conformity with policy objectives.

Operationally, the BOG calls DMBs each morning to determine their reserves positions. This provides the Repo Committee with information on the overall stock level or reserves of, and its distribution among, the DMBs. By noon, the BOG would have determined the amount of repos it will accept ― zero in the event of overall adequacy and only imbalances among the DMBs, and otherwise a value equal to the system-wide deficiency. In addition, the BOG set a ceiling (not revealed to the DMBs) for each DMB. Approval must be obtained from the Governor before the BOG would accept repos from a DMB in excess of its ceiling.

If such approval is not granted, the DMB would have to use the secured lending facility, which carries a penalty ― interest charge above the repo rate ― to obtain the additional reserves needed. If any DMBs shows a marked tendency to have frequent recourse to this lending facility, that in itself signals to the BOG that the DMB concerned may be acting imprudently. The BOG expects and encourages DMBs to manage their reserves needs by dealing with each other.

There is evidence of use of repos by the BOG as substitute for OMO. In July 2000, for example, BOG officials reportedly confirmed that for the better part of June the BOG had relied on reverse repos to drain reserves from the system. From mid-June to mid-July the five auctions were under subscribed by an average of 32.6%. Clearly, the BOG was unable to conduct OMO during the period under review. It therefore turned to reverse repos to meet its Reserve Money targets agreed with the IMF. Interest rates at the auction had reached about 45 percent per annum, apparently judged by the authorities as too high (in an election year and with the private sector faced with high financing needs on account of high foreign exchange exposure losses). Recourse to reverse repo could hopefully arrest the trend of continued (increase) in market interest rates.

This instrument, however, was totally inappropriate to the task at hand whatever was thought to be the consequences for interest rates. Indeed as may be expected, the additional sales of T-bills at the "given" rate of interest would be inadequate to the task in the face of the financing needs of government. Not surprisingly on July 6 2000 the BOG took the more drastic step of raising the mandatory cash reserve requirement by 12.5%.

Foreign Exchange Operations

In the past, faced with the monetary overhang in QIII 2000, the BOG would have sold foreign exchange as means to mop up the excess liquidity in the economy. Given the uncertainty about the success of OMO or the BOG unwillingness to let market forces determine interest rates, the essentially temporary nature of repos and reverse repos and the difficulties with frequent changes in the reserve ratios, foreign exchange operations (FEMO) have been one of the BOG’s main instruments of monetary control.

A sale of foreign exchange has the same effect on Reserve Money as the sale of T-bills or any other securities that the central bank classifies as a reserve asset that the DMBs are required to hold. Individuals and companies pay for foreign exchange by drawing down their cedi deposits at banks or by reducing their cedi cash holdings. In the former case bank reserves decline; in the latter currency with the NBPS falls. A fall in any components of Reserve Money is ceteris paribus a fall in Reserve Money. BOG could, of course, sell foreign exchange directly to the DMBs in which case their deposit accounts at the BOG are debited which reduces that component of Reserve Money.

Unless sterilized (sale of forex offset by a matching OMO purchase of government securities so that the sale of forex is in effect an exchange of one asset ― foreign exchange/securities ― without any change in the cash reserve holdings of DMBs and therefore Reserve Money), such intervention in the forex market by BOG not only reduces the stock of Reserve Money, but also moderates the rate of depreciation and even result in a nominal appreciation of the exchange rate against foreign currencies. Now the maintenance of a stable value of the exchange rate is one of the objectives of the BOG set by the Fourth Republican Constitution. Furthermore, the exchange rate is highly visible and often regarded as an indicator of the stability or lack thereof of the economy (Friedman’s conjecture. See however, Flood and Rose, Nov. 1999, "Understanding Exchange Rate Volatility without the Contrivance of Macroeconomics" Econ. Journal Vol. 109 No. 459: F660 – F672). Following the debacle with the VAT in 1995 the BOG tended to use the exchange rate as a nominal for the economy: a stable nominal exchange rate presumably regarded as a tangible sign that the authorities have inflation under control.

Thus in 1996 Budget Statement, government undertook to moderate the deterioration in nominal exchange rate "through appropriate intervention". About $100 million from the sale of Government shares in AGC were earmarked for this purpose. Consequently, in the face of an annual (year on year) inflation rate of 33 percent, the year on year rate of depreciation of the cedi (against the dollar) was limited to 22 percent. Sales of forex by the BOG (including of foreign currency loans to finance cocoa crop purchases) and the ADB at below market process help to dampen the depreciation of the cedi.

The BOG also has regulations on the books that banks cannot maintain long or short positions in foreign exchange that exceed a specified percentage of their capital. Thus, banks with long forex positions ― forex assets in excess of forex liabilities ― are supposed to sell via the interbank market to banks with short positions. These regulations reinforce the BOG’s practice of direct intervention in the forex market by keeping available supplies moving through the market.

The success of forex sales as a monetary instrument depends on the supply available to the BOG. During periods when the TOT improve (as in 1998), or when donor net inflows are particularly large, or when windfalls like the divestiture of AGC or privatization receipts or the stock of international reserves is sufficiently large, the BOG can successfully resort to FEMO. However, attempts to use the exchange rate as a nominal anchor for inflationary expectations when government fiscal deficits are large and persistent (as in the 1996 - 1999 period) have an especially pernicious effect on international competitiveness as the real exchange rate appreciates.

Furthermore, when the TOT turns adverse (as from QII 1999 to date) and the forex available to BOG dwindles, the stage is set for speculate attacks on the currency in forex markets. The tripling of crude oil prices in international markets and the inability of government to raise petroleum products prices no doubt contributed to the sharp increase in the demand for foreign exchange (and the record oil bill in 2000). In the event, given the low end and dwindling stock of forex reserves, the BOG could not intervene to moderate the ensuing depreciation of the cedi. From QIV 1999 through most of 2000 the cedi depreciated by about 100 percent on a year-on-year basis.

The ineffectiveness of OMO together with the viability to use FEMO significantly limited the options available to the BOG to control money supply growth supply. In an attempt to manage the growing shortage of forex in November 1999, the BOG resorted to forex swaps with the DMBs. A bank needing forex receive dollars now paying BOG in cedis. With the expectation that the bank will be receiving forex (from customer or aid inflows) at some future date, the bank at that time will swap the dollars for cedis with the

BOG. The exchange rates at which the current and future swaps occur determine whether the BOG is engaging in subsidized sales of forex.

In the summer of 2000, the IMF deemed the foreign currency swap arrangement to amount to multiple currency practice, which is prohibited under the PRGF Arrangement with Ghana, and directed the GOG to cease these practices, though it granted a waiver on the grounds that the Ghanaian authorities were unaware of this implication signaled the failure of FEMO ― the traditional methods of forex intervention, and was a clear example of the riskiness of using the nominal exchange rate as a nominal anchor in the context of severe macroeconomic imbalances.

Reserve Requirements

There are two reserve requirements – primary/cash and secondary. These are set at "fixed" percentages of total DMB deposit liabilities including foreign currency deposits. Primary or cash reserves are currently fixed at 9% of total deposits ― changed from 8% in July 6 2000. Primary reserves consist only of deposits of DMBs at the BOG ― vault cash does not count towards meeting this reserve requirement.

Secondary reserves must be held in the form of approved government paper primarily T-bills and are currently set at 35 percent of deposits liabilities.

In modern banking systems reserve requirements are used more as a prudential measure than as an instrument of monetary control. Indeed changing the reserve requirement as an instrument of monetary control is considered outmoded. As a prudential measure, however, reserve requirements mean that banks must set aside a fraction of their deposits in order to meet ordinary and reasonable withdrawal demands of customers. The fact that the cash is there when customers demand it maintains the NBPS confidence in the soundness of the banking system. Frequent changes in reserve requirements for monetary control purposes impose accounting costs on banks and thus raise the cost of financial intermediation. For this reason required reserve ratios tend to remain constant for long periods of time in industrial countries.

Historically, most of the changes in broad money supply in Ghana, occur from changes in Reserve Money ― the money multiplier has been relatively stable. Changes in reserve requirements cause corresponding changes in the money multiplier. The money multiplier, m, can be written as

m = ( 1 + c ) / ( r + c )

where m is the money multiplier, r is the primary/cash ratio (as would normally be the case, r ≥ 9 percent), and c > 0 is the cash/deposit ratio of the NBPS. Consider an increase in the required primary ratio which pushes the actual to r’ (r’ ≥ r) with corresponding change in m to m’. Then it can be shown that

So that increasing the mandatory primary ratio causes the multiplier to decrease.

Furthermore,

by assumption r’ > r and so

Hence, │∆m / m│ < │∆r / r│

And therefore the percentage decrease in the multiplier is smaller than the percentage increase in the mandatory primary ratio.

The DMBs typically have held primary reserves in excess of the requirement ― averaging about 3.5 percentage points over the nineties. As implied in the above analysis, r is the observed or actual primary ratio normally higher than the required (on account of excess reserves of the DMBs). Now if the banks respond to an increase in the mandatory reserve requirement by reducing their excess reserves holdings, then the overall reduction in the multiplier would be smaller. In the case of a total off-set ― excess reserves reduced by an amount equivalent to the mandatory reserve increase, the observed/actual reserve ratio remains unchanged i.e. r’ = r as that ∆r = 0 and consequently ∆m/m = 0 i.e. the multiplier is unaffected by the change in the mandatory reserve ratio.

Holdings of excess primary reserves are not unused ― banks want to be able to meet the withdrawal requests of their customers and the amounts of these requests cannot be predicted with pinpoint accuracy, particularly when macroeconomic conditions are unstable. Again, if penalties for failing to meet reserve requirements are high, banks have a further incentive to hold excess reserves.

High total ― primary and secondary ― reserve requirements could severely constrain the supply of credit to the private sector. For example, over the decade of the nineties mandatory total reserve requirements averaged about 50 percent of DMB-deposit liabilities. Thus the government (and the BOG) claimed about half the money deposited with the banking system, leaving only 50 percent as potentially available for lending to the public at large.

On the other hand, the evidence is that the banks have consistently failed to lend to the full potential. Excess total reserves over the period of the nineties averaged about 16 percent of deposit liabilities of banks. And leaving aside 1996 when the DMBs failed (as a collective) to meet the secondary reserve requirement of 50.8%, the excess secondary reserves average about 15 % over the nineties (falling to 12.7% if 1996 is included). Banks have willingly held far more government securities than required. These securities offer very high margins over the banks’ cost of funds ― interest paid on deposits ― and at little risk if any.

The banking skills required to participate in the T-bills market and to manage their portfolios are clearly available to the banks. The alternative to investing in government securities is to sift through a large pool of high-risk potential borrowers and to collect those that are most likely to repay their loans. This would require very different banking skills ― a set of skills likely to be rather scarce given the persistently high demand for bank finance by the government. The risks of default on private sector debt are moreover increased by unstable macroeconomic conditions often manifested in high and variable real interest rates.

The DMBs go to the discount houses (DHs) if they are unable to get reserves through the interbank market. Banks with excess reserves lend to the DHs which, in turn, buy T-bills with the money. DHs are specialists in dealing with the mismatch of types of funds available in the market. They thus serve a function of maturity transformation: if they get a steady flow of funds from the banks, they could treat a portion as ‘permanent’ and buy 91-day T-bills with that portion. If the flow of funds from the banks becomes highly volatile, the permanent portion, in effect, shrinks to zero. DHs, in such a situation, would tend to turn away these funds since there are no appropriate investment outlets for such funds.

When the DHs were first established in the early 1990s the intention was that banks would deposit their excess funds on call at the DHs. Deficit banks would borrow from the DHs, typically by issuing unsecured, negotiable certificate of deposit (CD) of short maturity ― one week or less. If the banking system as a whole forced to approach the BOG for accommodation. The BOG lending to the DHs would have to be secured by T-bills.

In practice, borrowings by the DHs from the BOG were at punitive rates. Apparently this was to forestall round tripping ― DHs borrowing at a preferential rate, investing these funds in T-bills, borrow more, invest more in T-bills and so on indefinitely. The DHs, in turn, passes on these punitive rates to its borrowers. But this led to the situation in which banks found it cheaper to deal directly with each other ― by passing the DHs. In the process the interbank money market was created and has evolved into the primary market in which banks borrow to meet their reserves needs or to lend out their excess reserves.

This is an informal market operating on the honour system. Banks deal directly with each other as against a more formal market arrangement relying on (an electronic network) and market makers. Interbank transactions are unsecured. Nonetheless the interest rate on overnight loans has been usually below the T-bill rate and at least for the most credit-worthy banks, is usually the cheapest source of funds.

Banks may also enter into repo with a DH. The bank then sells T-bills to the DH is a fixed margin over the current interest equivalent on T-bills. This margin is typically 0.5 percentage points higher that the BOG margin in its repo transactions. There is some counterpart risk in repos depending on the quality of the bank. The DHs, therefore, adjust the interest margin charged to the bank to reflect this risk. ‘The risk’ stems from the possibility that the bank may not have the funds to close the repo ― buy back the T-bills it sold to the DH. Again banks could issue CDs to DHs. This being unsecured lending by the DH, it charges a higher interest rate than it would on repos.

Conclusion

Casual empiricism suggests that the financial sector in Ghana is more volatile than the non-financial sector. This would make the interest rate preferable to the stock of some monetary aggregate as the instrument of monetary policy. Indeed the evidence is that GOG and BOG officials consider the interest rate as the monetary policy tool. Thus for year 2000, the 1999 Budget Statement asserted: The BOG will use interest rate policy to reinforce its monetary policy objectives. Indeed BOG publications have often asserted that the (toughness of) monetary stance can be judged by the discount rate.