5
Australia under credit watch
International lenders continue to monitor Australia’s overseas borrowing position. In normal times, they calculate that the higher interest they can earn in Australia compensates them for the risk that Australian borrowers will fail to repay in full and on time. Should this confidence be disrupted, they will refuse to roll over their loans and demand repayment. If their loans are not repaid in foreign exchange and on time, a banking crisis will ensue. To assess the likelihood that confidence will be disrupted, one needs to look at the indicators that are monitored by the lenders. Though the macroeconomic indicators of vulnerability to crisis that are watched by the markets are closely related to the indicators we have already discussed, there is another group that we have not yet described — indicators of the adequacy of Australia’s foreign-exchange reserves.
The Australian government’s first line of defence in the event of a loss of confidence in the Australian economy by overseas lenders is its foreign-exchange reserves. An indicator of the adequacy of these reserves (akin in some ways to the capital adequacy of a bank) is the year-ahead foreign-exchange-cover ratio. The higher this ratio, the less likely is an over-borrowing crisis.
Australia’s annual gross international borrowing requirement
High short-term foreign debt — that part of foreign debt which matures within twelve months of any given date — played a key role in the 1997 Asian crisis. Short-term international debt is of critical importance. As it falls due, it has to be repaid, rolled over, or replaced by loans from other investors. Total foreign debt is also important, because lenders know when its components are going to mature, and so require repayment or refinancing. If such medium-term debt is judged too high, they may not be willing to roll over or replace short-term debt, even when it would otherwise be manageable.
When short-term debt and potential short-term debt are high, and especially when they are high relative to foreign-exchange reserves, there comes a point where foreign lenders lose confidence and are unwilling to roll over the debt. The lenders demand repayment, triggering an economic crisis. The rule of thumb for developing and emerging economies that wish to avoid a debt-rollover crisis is that they should match their short-term foreign debt with the same level of foreign-exchange reserves.
Since the Second World War, Australia has not found it necessary to observe this rule of thumb, having consistently incurred greater short-term debt than its foreign-exchange reserves, and (so far) has done so without disaster. However, this does not mean that there is no upside limit to Australia’s capacity for short-term overseas borrowing. To assess Australia’s position, we have offset its foreign-exchange reserves against its short-term debt. Its annual international borrowing requirement at the end of each quarter is accordingly defined as its short-term foreign debt (being that part of its foreign debt issued by both the public and private sectors that matures over the next four quarters), plus its expected current-account deficit over the next twelve months (for the sake of simplicity, taken to be equal to the quarter’s current-account deficit multiplied by four), less its (public-sector) holdings of foreign-exchange reserves.
This annual borrowing requirement can be assessed in relation to various other economic magnitudes — for example, the level of export earnings — but in the first instance we compare it to GDP.
Figure 2 profiles the annual gross Australian international borrowing requirement since the mid-1990s. Over the second half of the 1990s, the ratio ranged from 25 to 30 per cent of GDP; over the first half of the 2000s, the ratio was reasonably stable in the vicinity of 35 per cent. In the run-up to the Global Financial Crisis, the ratio rose rapidly to hit 51 per cent in the September quarter 2008. During 2009, in response to the GFC of the previous year, Australia undertook large-scale fiscal expansion, which resulted in further increases in imports and borrowing. This raises the question, why didn’t Australia experience an economic crisis in 2009? We have already provided the answer: China undertook a large-scale monetary and fiscal expansion that raised the price of Australian mineral exports. This put a floor under the currency, such that the Australian dollar rose from 66 cents per US dollar in December 2008 to 91 cents by December 2009. This rising exchange rate pushed the annual gross international borrowing requirement down to 40 per cent of GDP by the December quarter of 2009. Had China not expanded, and had Australia not been a major supplier of Chinese imports, it is strongly arguable that the expansion of 2009 would have raised the balance-of-payments deficit (through increased imports unmatched by exports) and so brought the GFC to Australia.
By watching the annual borrowing requirement, we correctly predicted (to within a year or two) the Asian economic crisis of 1997 from the viewpoint of 1994, and the Australian recession of 1991 from the viewpoint of 1985. Alarmingly, the borrowing requirements that foreshadowed these events were lower than those that prevailed for Australia in 2008 and indeed those that prevailed at the end of 2015, when the ratio was 45 per cent of GDP. This raises the question as to whether Australia is already in highly dangerous territory. However, over the past couple of decades, capital outflows and the potential hedging of short-term foreign debt have changed the significance of the international-borrowing-requirement indicator.
Figure 2: Australia: Year ahead gross and net borrowing requirements and foreign-exchange-cover ratio
Private-sector short-term overseas assets
If the current level of Australia’s international borrowing requirement as a percentage of GDP had the same significance as it had in the 1990s, the Australian economy would now be in extreme danger of crisis. However, allowance should be made for the increase in Australian short-term foreign lending. The question then is: how much allowance?
In the post-war period, foreign borrowing and foreign lending could be offset against each other because most foreign debt and lending was held by governments or their Reserve Banks. In the build-up to a crisis, governments could sell their foreign assets to augment foreign-exchange reserves and so reduce the risk of default on foreign debt.
This is not the case today. Holdings of foreign short-term financial assets are dominated by private-sector institutions, and may be divided into two classes. The first class comprises foreign short-term assets held by financial institutions (mostly banks) as partial offsets to their short-term foreign borrowings. These assets are held as part of the banks’ strategy for managing their overseas obligations. It can be assumed that in the event of a crisis these assets will be at least partially available for offset against overseas liabilities.
The second class comprises foreign short-term assets held by institutions (mainly investment funds), and indeed by individuals, without any corresponding overseas borrowing. The motives for doing this vary from insurance against exchange-rate movements through to outright speculation against the devaluation of the Australian dollar. Expectations of devaluation will be particularly strong in the period immediately before a crisis when the exchange rate is already falling. In this situation, domestic holders of foreign assets that are not offsets to foreign liabilities will have every incentive to keep these assets offshore and therefore prevent their being offset against overseas liabilities.
Given this distinction, Australia’s international-borrowing requirement should be adjusted to allow for bank holdings of short-term overseas financial assets, defined as those that will mature within a year. In this respect, financial deregulation has returned Australia to a situation akin to that before the 1930s Deptession, when Australia’s foreign-exchange reserves consisted largely of the ‘London balances’ of the commercial banks, with one important difference: the exchange rate is now market-determined. This means that, in addition to their function as reserves, bank holdings of overseas liquid assets have a ‘hedging’ function: they are held to compensate banks that have borrowed overseas for adverse changes in the exchange rate.
We allow for the partial availability of bank overseas holdings of short-term assets to meet short-term demands for overseas funds by subtracting them from the borrowing requirement after allowing for their hedging function.1 On this definition, the net annual borrowing requirement is equal to the gross annual borrowing requirement, less bank short-term foreign lending, other than that held for hedging purposes.
The resulting net borrowing requirement series is profiled in Figure 1. The adjustment for bank holdings of foreign short-term assets reduces the annual gross international borrowing requirement from $737 billion to a net requirement of $505 billion in the December quarter of 2015 — a reduction from 45 per cent of GDP to 33 per cent. The gap between the unadjusted measure and the net measure has widened considerably over recent quarters, because the ratio of foreign lending to GDP increased from 51 per cent in June quarter 2014 to 70 per cent in December quarter 2015, with the share of short-term foreign lending increasing from 42 per cent to 47 per cent.
Purchasing overseas short-term assets is not the only way in which the Australian banks protect themselves against exchange-rate risks. They also hedge by making what are effectively insurance payments over and above the value of foreign assets and liabilities. However, this provides only short-term protection, particularly into the future. As the risk profile of the Australian economy increases, especially in regard to exchange-rate risk, the insurance premiums will increase rapidly to the point of unaffordability, which will force institutions and enterprises to reduce their hedging defences. In a severe crisis, there is an additional risk that the costs to the counterparty to the hedging arrangement will be so large that the counterparty defaults — as happened during the GFC. Counterparty default is particularly likely if the exchange rate falls sharply, and especially if this occurs in the context of world-wide currency instability, as would be the case in a second version of the GFC where the counterparty could experience losses across many segments of its portfolio.
As the vulnerability of the Australian economy increases, we expect that the hedging ratio, both measured and unmeasured, will decline back to its historical norm. By 2021, we assume that it will fall back to 25 per cent, even if the ratio of short-term foreign lending to total foreign lending remains close to its current level.
The year-ahead foreign-exchange-cover ratio
These data on the borrowing requirement can, alternatively, be presented as the year-ahead foreign-exchange-cover ratio, defined as the ratio of effective reserves to short-term obligations. The numerator of this ratio, effective reserves, is defined as total official foreign-exchange reserves, plus short-term foreign lending multiplied by the national hedging ratio. The denominator, short-term obligations, comprises gross short-term foreign debt plus the year-ahead current account deficit (taken to be four times the current quarter current-account deficit). This ratio is also profiled in Figure 2. In 2015, it averaged 0.33, implying that if foreign investors refused to roll over any short-term debt and lend any new debt to finance the current account deficit, Australia would last four months before being forced to default. This represents a considerable improvement from the value of 0.18 (two months) experienced in the immediate aftermath of the 2008 GFC. Indeed, the values of 0.17 to 0.19 experienced in 2000–01 and again in 2008 identify periods when the Australian exchange rate came under considerable downward pressure.
Australia has been able to improve its foreign-exchange-cover ratio over recent years because relatively low current-account deficits have been accompanied by a reduction in capital outflow relative to GDP. An influx of foreign funds to purchase Australian dwelling and farm assets may also have played a role. However, the end of the mining boom and the recent adverse movement in the terms of trade are likely to bring a return to high current-account deficits and capital outflows, which will push the foreign-exchange-cover ratio back towards 0.20 by 2018. If high current-account deficits and capital outflows continue after this (as they will, if policies remain unchanged), the ratio will keep on declining and so increasingly warn of economic crisis.
We now turn to the other side of crisis prediction, that reflecting the capacity of an economy to generate macroeconomic indicators that maintain confidence in its ability to repay its creditors. Though defensive reserves are important, the fundamental causes of financial crises go deeper. We have already summarised the recent work by the European Commission on the antecedents of national financial catastrophes. Similar studies have been published by the IMF and others, generating a literature on the indicators of impending catastrophe that is of compelling interest to international lenders.
Because each crisis is a tumultuous, unique event, there is no set path to calamity. There is no combination of indicator values that guarantees a catastrophe — just as no combination short of the complete avoidance of overseas borrowing guarantees immunity. However, based on recent experience, it is possible to estimate the probability that a crisis will occur. In our work, we have developed a summary indicator of Australia’s vulnerability to a crisis arising from excess debt. This crisis-risk indicator extends the EC approach; it is an index of macroeconomic indicators with a proven relationship to the risk of catastrophe. It provides a second indicator of vulnerability, alongside the year-ahead foreign-exchange-cover ratio.
Crisis risk assessed from macroeconomic indicators
We have already considered the EC list of macroeconomic indicators of vulnerability to crisis and concluded that, for Australia, the risk arises chiefly because of its excessive overseas borrowing. We accordingly concentrate on this aspect of macroeconomic risk, building on a model developed by Luis Catao and Gian Milesi-Ferretti of the IMF.2 We have applied this reasoning to Australia’s position, developing the model documented in the report Carbon crisis: systemic risk of carbon emission liabilities, prepared by National Economics for Beyond Zero Emissions, and published in December 2014. This report includes the impact of an externally imposed carbon price shock along with other negative shocks on the Australian economy, including the shocks considered below. The model was developed to summarise the experience of medium-sized economies with significant overseas debt, including all such countries that have experienced economic catastrophe over the past four decades. Our analysis excludes small economies since they are not likely to be relevant to the Australian case.
Harkening back to the European Commission scorecard, the IMF identified seven indicators that influence the probability of an economic conflagration. They were:
We have incorporated these indicators into an index of the risk of an Australian economic crisis using weights estimated from data for all relevant countries over the period from 1970 to 2011. The higher the indicator value, the greater the level of risk. As shown in Figure 3, the average risk-indicator value for the period was 0.52, with the years of highest risk being 1986 at 0.74 and 2000 to 2001 at around 0.72. In 2011, the risk of crisis abated to a low level. However, after 2013, with the winding down of the mining boom, the fall in commodity prices, and a downward adjustment in the exchange rate, the risk indicator increased rapidly and is now back to its average value for the period since 1970.
Figure 3: Australian systemic crisis risk indicator from 1970 to 2015
The critical question is: what is the threshold value of this risk indicator at which a crisis is likely?
The warning codes
Our two indicators — the year-ahead foreign-exchange-cover ratio and the macroeconomic crisis-risk indicator — are not independent, if only because reserves of foreign exchange fluctuate with macroeconomic flows across the balance of payments. However, the two indicators are sufficiently different in derivation and implication to justify separate treatment. Both should be treated as warnings. It is therefore appropriate that they should be interpreted in a way similar to bushfire warnings.
Bushfires are similar to financial catastrophes, not only because they are highly destructive. For both, it is possible on the basis of past experience to estimate the imminence of catastrophe, whether or not it actually occurs in any particular case. Like the indicators of the danger of bushfire (temperature, wind speed, and the like), the signals of impending financial collapse can be summarised into numerical values that can in turn be converted into warning codes such as those issued by the Victorian Country Fire Authority (CFA). The major difference is that fire warnings refer to the need to take action to escape from wildfire any time within the next few hours or days, whereas our alerts refer to the need to take drastic action to escape from financial conflagration within the next few months or years.
The CFA codes range from low through moderate to high, very high, severe, and extreme, to Code Red. These same codes can be applied to describe the risk of financial collapse as measured by our two indicators. Like bushfire warnings, the ratings are strongly based in historical experience — in this case, the historical record of the events leading up to economic conflagrations overseas, coupled with the historical record of Australia in avoiding such conflagrations since the 1930s. However, again like bushfire warnings, they are subject to uncertainty. It is inevitable that an element of personal judgement will enter into the assessments of probability, though we have tried to be conservative.
The codes have been calibrated so that Code Red status means that, unless immediate evasive action is undertaken, the probability of an economic catastrophe within the next five years is 50 per cent or more. As with bushfires that, due to carelessness or unusual circumstances, can still occur at times of low or moderate danger, so also with economic catastrophes: a low or moderate warning code means that catastrophe is still possible within the next five years, but its likelihood is rated at 15 per cent or less.
Calibrating the year-ahead foreign-exchange-cover indicator
Armed with this rating system, we now provide a more detailed description of the two summary crisis-predictor indicators, starting with the capacity to ward off a crisis.
The year-ahead foreign-exchange-cover ratio incorporates the same data as the net annual borrowing-requirement ratio, but is expressed in a way that allows easier interpretation. In this form, it provides the basis for bushfire-code warnings. The codes allocated are based on evidence from the historical record that the exchange rate comes under particular pressure (relative to its purchasing-power-parity value, which in 2016 was 69 cents to the US dollar) when the year-ahead foreign-exchange-cover ratio is in the 0.16 to 0.20 range. Code Red status, when a crisis should be expected with a 50 per cent probability or more, is set at a cover ratio of less than 0.12, or less than six weeks of annual cover in the event of a total refusal by overseas investors to finance Australia’s short-term debt rollover requirements, plus the new debt required to finance the current-account deficit. Those familiar with the bushfire warning system will know that Code Red means ‘Get away, and fast.’ In the present context, it means that economic conflagration is not far short of unavoidable, so hold on tight for a crisis that will destroy the financial structure of the economy and much else besides. After the conflagration, the country may be given a period of grace to reconstruct from whatever foundations it has left, or may be effectively taken over by administrators acting in the interests of financial institutions overseas.
Short of Code Red, the bushfire warnings include various options to ‘stay and defend’. The analogy in the present context is that at high or even severe levels of warning, there is room for disaster-avoidance policies. The worse the assessment, the greater is the need for such policies.
At the 2016 value of the ratio of 0.33, we categorise the risk as moderate; that is, we judge that Australia is not seriously at risk of an economic crisis within the next five years. However, the historical record shows that such a risk can increase sharply. During 2000, the risk was classified as severe, as it was again in 2008. It remained very high during 2009 and to the end of 2011. The present respite from risk may prove to be merely temporary — a matter to be considered below.
Calibrating the macroeconomic-crisis indicator
We have calibrated the macroeconomic-crisis indicator by analysing the impact of different threshold levels, offsetting the cases where the indicator correctly predicted a crisis against the cases where a crisis was predicted and did not eventuate, taking the experience of all indebted medium-sized countries into account. Historical experience thus analysed indicated that Code Red status is reached when the crisis-risk indicator attains a value of 0.84. The historic record provides instances of countries where the crisis-risk indicator has exceeded this value for a year or so without disaster ensuing, and similarly there have been instances where disaster has occurred at lower levels of the indicator, but 0.84 represents the value where economic conflagration becomes more likely than not.
More technically, at a threshold of 0.84 the number of crises correctly called is 26 out of the 39 crises that occurred in the estimation database, so that the estimated model correctly called two-thirds of the total. However, at this probability threshold the model incorrectly called 101 crises that did not occur, in many cases because the government concerned took evasive action. Higher threshold levels would increase not only the correct call rate but also the incorrect call rate, and vice versa. The 0.84 threshold is accordingly a compromise that produces similar performance standards to those found in other crisis-prediction models. Code Red status occurs when the crisis-risk indicator exceeds this threshold level for more than four consecutive quarters.
It should be noted that the data in Figure 3 is in annual terms and, therefore, the 2008 estimate of 0.6 is affected by quarterly averaging. There were days during the December quarter of 2008 when the exchange rate was very low compared to the rest of the year and the crisis-risk indicator would have been approaching its threshold value of 0.84. The high-risk years during the period for which data is available were 1986–1987 and 2000–2001, both occasions when the exchange rate was low.
Unlike our other indicator, the year-ahead foreign-exchange-cover ratio, a fall in the exchange rate directly increases the value of the cash-flow crisis-risk indicator, because it reduces the US dollar value of Australian GDP. A fall in the exchange rate also indirectly increases risk via its impact on the value of foreign debt and any negative so-called J-curve effects on the balance-of-payments current-account deficit.
From this perspective, the two indicators complement each other, with each emphasising a different set of factors critical in determining whether or not a crisis will occur. The crisis-risk indicator captures the risk of a crisis from an uncontrollable fall in the exchange rate such as can result from a sudden loss of confidence. The cover-ratio indicator simply measures the obvious: if a country does not have sufficient resources to meet its obligations, a crisis will occur unless its resources are somehow augmented.
According to the crisis-risk indicator, from 2010 to 2014, when the Australian exchange rate was significantly over-valued compared to its PPP value, the risk of a crisis was low. Given the value of the indicator at the end of 2015 of 0.51, the Australian crisis-risk ratio had risen to high, having been low in the December quarter of 2014. These ratings concur with the year-ahead foreign-exchange-cover ratio in concluding that in 2016 the chance of economic crisis was high but not yet severe, but the trend was adverse. As the bushfire warning system indicates, Australia has the opportunity to take evasive action, and the breathing-time to do so.
But how much breathing-time? This raises the question of the values of the indicators over the next five years.