DELEVERAGING IN A SMALL ECONOMY OF THE EURO

Transcription

1 DELEVERAGING IN A SMALL ECONOMY OF THE EURO Abel M Mateus EBRD and UCL Abstract: Deleveraging is essential for restoring financial sustainability and avoid recurring financial crisis. After one of the highest credit booms, Portugal as all Euro crisis countries is experiencing a substantial deleveraging of all economic agents, which limits GDP growth potential. The process only started in 2012 for enterprises and intensified for households, but is expected to be initiated in 2015 for the Public Sector. According to our estimates, in a base scenario, it will take up to 2030 for the economy to return to sustainable leverage ratios. In this scenario, enterprises will have to make most of the deleveraging until around 2020, while the public sector will take longer, with households already well in the way for stabilizing the debt ratios. The total debt ratio will have to be cut by 80 pp over 15 years, approaching the 2007 level. In a more stringent scenario, the overall debt over GDP will have to be cut by 130 pp, about the level of 2003, pre-crisis. Both these scenarios will impose a strong restriction on the expansion of the economy, which is expected to grow below its long-term historical path, and will require a major adjustment in corporate finance policies. We show that this challenge is of a heightened intensity than for the public sector. Finally, the paper summarizes policies can play a role in the deleveraging process. Tags: Financial crisis, Eurozone, Banking deleveraging London, 26 of March, 2015

2 1. Introduction Deleveraging is a process that reduces the debt levels over income or wealth indicators. At the macroeconomic level is of special relevance after a financial and banking crisis. We concentrate our analysis on the non-financial private sector. A study of the BIS 1 of 20 systemic banking crisis showed that the intensive phase of the deleveraging process lasted 6 to 8 years after the end of the crisis with a reduction of the debt to GDP ratios of about 38 pp, in most cases returning to the levels before the credit boom. But the more comprehensive work of Reinhart and Rogoff 2 show that overall deleveraging, including the Public Sector, may take up to two decades. The Portuguese banking and financial crisis was rather normal from an historical perspective: it lasted about 4 years ( ), GDP decreased about 8% in cumulative terms, and unemployment jumped about 6 pp from pre-crisis to its peak and public debt over GDP increased by 58 pp. Less studied has been the long deleveraging process after the critical phase of the crisis. The simple fact that deleveraging processes take long, any study is bound to be affected by different shocks and the role of idiosyncratic factors is always relevant. Comparing different deleveraging paths requires building a Dynamic General Equilibrium Model that is beyond our task here and has limitations of their own, since they can never capture the full richness of each country experience. There are usually three phases in the credit boom and bust cycle associated with a financial crisis (Figure 1). In phase I, during the credit boom, credit expands rapidly and at a much higher rate than GDP, which also increases with the boom. In the second phase debt stabilizes and then starts to decrease. With a recession due to the sudden-stop, the debt to GDP ratio continues to increase up to a peak and only then starts to fall as the reduction in nominal debt overtakes the fall in nominal GDP: we could distinguish phase IIa when the ratio of indebtedness is still rising and phase IIb in which the ratio starts to fall. The third phase starts with the recovery of GDP. The level of debt would continue to fall, but as nominal GDP gathers momentum the ratio of indebtedness starts accelerating the descent. Looking at Figure 3 it is clear that Ireland, Spain and Portugal entered in Phase IIb in 2012, while Greece has not yet passed that stage. Figure 1 BUST RECOVERY BOOM Phase I IIa IIb Phase III 1 Tang and Upper (2010). 2 See, e.g., Reinhart and Rogoff (2009).

3 Deleveraging affects output through different channels. First, households affected by the credit constraint have to cut demand in order to re-establish their sustainable debt levels, by reducing consumption and postponing further investments. This leads to a significant contraction in housing investment and consumer durables, which has a direct negative effect on output. Second, there is an additional effect due to the so called debt-deflation spiral: falling prices slow down the speed of deleveraging in terms of the real debt, which forces households to deleverage more aggressively. In consequence, households have to shed relatively more nominal debt, which pushes prices further down. Third, the deleveraging shock leads also to lower investment in capital as the real interest rate rises due to the perception of a higher credit risk of the enterprise. This is the credit channel studied by Bernanke and Blinder (1988) and Kiyotaki (1997) and Moore and Kiyotaki (1998). The problem is exacerbated in a monetary union because without an independent monetary policy the nominal interest rate may not fall sufficiently to offset the decrease in prices for the particular member country. Forth, real wages may also fall. Since labour supply tends to increase (due to the negative income shock for households) and the labour demand falls, the net impact on equilibrium employment is largely driven by the degree of wage flexibility: in most of the European economies with relatively high degree of real and nominal wage rigidities, the fall in wages is relatively contained and unemployment increases more significantly. The negative impact on the economy leads, via the automatic stabilisers, to a negative impact on the government's budget balance and to an increase in the public-debt-to-gdp ratio. At the same time, deleveraging also leads to a rebalancing of the economy as housing investment falls, thereby reducing the demand for non-traded goods. Demand shifts from housing investment towards consumption goods and production resources are redirected towards the traded sectors. The economy's net foreign asset position improves, not only due to the assumption described above, but also due to the decrease of the terms-of-trade (due to falling domestic prices) and net exports increase. A virtuous cycle may take hold once the impact of net exports is sufficiently high, leading to a recovery of production and incomes. The increase in tax revenues will ease the pressure on the government deficit and may create the conditions for an easing of the tax pressure on the economy. Consumption will recover and enterprise investment will start in the export sector and then will spread to the rest of the economy. As GDP increases and wage and price increases are restored to their normal levels deleverage ratios start to decline more rapidly. The most important variable to take into account in our analysis is the real interest rate in the Euro area. Figure 2 shows three phases in the monetary environment in the last ten years. The first phase, in the run-up to the global financial crisis, short-term real interest rates were close to zero, naturally feeding-up the credit boom. The second phase, corresponding to the flare-up of the crisis, both nominal and real short-term interest rates increased substantially. The real interest rate increased from zero to 2 percent. The third phase, during the Euro crisis that followed, from 2010 to end of 2012 the nominal rate stayed around 1% and with inflation catching-up, real short-term rates became negative. With the drop in inflation rates, and despite the setting of ECB rates at a level

4 01/01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/ /05/ /09/ /01/2015 close to zero, followed by the launch of Quantitative Easing in March 2015, real interest rates turned again positive. Figure 2 Euro-area nominal and real interest rates Real interest Euribor 3M Inflation Source: ECB and Eurostat Thus, the process of deleveraging took place, after 2012 in an environment of negative real interest rates. The ECB has announced a policy of supply of large liquidity to the markets in the next year or two, until inflation re-gains a level between 1 a 2% and real growth shows a robust increase. Thus, economic agents should expect low real interest rates for the next 2 or even 3 years. After this phase they should start to increase posing a new challenge for highly indebted agents. 2. How much deleveraging has taken place? Banks reduced substantially transformation ratios and re-capitalized, and on average they will reach EU average level in However, some large banks still need a substantial effort in the next 2 to 3years for recovering normal profitability. Households were already debt constrained at beginning of the 2000s, but they started to deleverage only in Enterprises started the process in 2012, but this is one of the most serious problems, giving the high levels of distress, mainly for medium and small enterprises. Large corporations have reduced the level of leverage more substantially, but they continue to be constrained to make new investments. The Public Sector, which had to take a substantial amount of debt from public enterprises and to a lesser extent form the banking sector, will start to

5 Perc. GDP reduce the debt over GDP ratio only after Debt sustainability tests for the Private Sector indicate that large segments of enterprises and to a less extent households, will be financially constrained in the next 5 to 7 years, but it is expected that banking sector will be in more solid basis in the next 2 years and after the problems of BES, BCP and some smaller banks in resolution are worked out. The problem of deleveraging will become certainly easier to solve once inflation regains its long term levels (2%) and GDP growth accelerates in the Euro region. However, we should not be too optimistic because interest rates are now at historical low levels. Credit market fragmentation has led to higher interest rates paid by Portuguese SMEs than core countries SMEs 3 which justifies the acceleration of SMEs deleveraging in the period. But at the same time, the low levels of non-performing loans of households is due to the fact that a dominant share of mortgage loans are priced at variable rates, linked to the 6 months Euribor. Once short interest rates increase it will also affect this segment. Box 1: Great Deleveraging Episodes There are very few episodes among advanced countries of steep overall deleveraging of the economy in recent history. We chose Finland ( ) and Ireland ( ). The first followed the shock of the fall of the Soviet Union which led to a decrease of about 10% of GDP in the period with both a strong increase in the private and public leverage (Figure B.1). After the stabilization, the overall debt ratio decreased by 55 pp of GDP over 5 years, due to both public and mainly the private sector. Figure B Finland: Debt Stocks Public Private Source: Ameco and Eurostat Decomposing by the factors of deleveraging: 23 pp was due to the GDP increase in real terms, 11 pp to inflation and 21 pp to a reduction in debt levels. It is remarkable that this 3 Several statistics show at least a 200 basis points spread.

6 Perc. GDP In the case of Ireland since 1978 there was only a short period of overall deleveraging (Figure B.2): from 1987 to 1990 with a reduction of the debt ratio by 21 pp of GDP. This was totally due to the increase of nominal GDP (23 pp in real GDP and 10 pp in inflation). What is also remarkable is the decrease in the public debt ratio, from 110% of GDP in 1987 to 24% in However, this was accomplished with an annual real GDP growth rate of 6.2% and inflation of 3.6%, a growth environment truly unique. Figure B Ireland: Debt Stocks Private Public Source: Ameco and Eurostat 2.1. Deleveraging by the banking sector After the peak levels reached in 2008, banks in the Eurozone started to reduce their transformation ratios, stocks of loans over deposits. From October 2008 to December 2014 that ratio was cut by 31 pp by Eurozone banks, according to data from the ECB (Figure 3). Portuguese banks were also forced to reduce that ratio one year later, and in December of 2014 the cut was even deeper: 46 pp.. The reduction was even higher in Ireland and Spain, countries that experienced a large credit boom, largely linked to real estate. In November 2014, the crisis countries (Spain, Ireland, Greece and Portugal) had an average transformation ratio of 131%, down from 180% in November 2008, a

7 Austria Belgium Bulgaria Cyprus Czech Republic Germany Estonia Spain Finland France United Kingdom Greece Croatia Hungary Ireland Italy Lithuania Luxembourg Latvia Malta Netherlands Poland Portugal Romania Sweden Slovenia Slovakia Euro area 2007Mar 2007Jun 2007Sep 2007Dec 2008Mar 2008Jun 2008Sep 2008Dec 2009Mar 2009Jun 2009Sep 2009Dec 2010Mar 2010Jun 2010Sep 2010Dec 2011Mar 2011Jun 2011Sep 2011Dec 2012Mar 2012Jun 2012Sep 2012Dec 2013Mar 2013Jun 2013Sep 2013Dec 2014Mar 2014Jun 2014Sep 2014Dec Figure 3: Convergence of Transformation Ratios for Crisis Countries and Germany Portugal Spain Greece Germany Ireland Euro Definition: Loans to Households and Corporations over Deposits same agents Source: ECB Data warehouse reduction of 49 pp, slightly higher than Portugal. What is also remarkable is that the standard deviation of this group was reduced from 66 to 4 in the same period. Among the Eurozone countries (Figure 4) subsist quite large disparities due to specific characteristics from low level of intermediation/credit to the economy (Croatia, Romania and Bulgaria) to countries still with high levels of leverage (Hungary and Czech Republic). Figure 4: Eurozone: Bank Transformation Rates (Nov 2014) Source: ECB Data Warehouse Total debt over GDP increased substantially from 2000 to (Figure 5): a major cause of the financial crisis that erupted in Greece, Ireland, Portugal and Spain. In the years up to 2013 there was

8 Belgium Bulgaria Czech Republic Denmark Germany Estonia Ireland Greece Spain France Croatia Italy Cyprus Latvia Lithuania Luxembourg Hungary Malta Netherlands Austria Poland Portugal Romania Slovenia Slovakia Finland Sweden not yet noticeable overall deleveraging in these economies, as most are still in a slow process of recovery. Figure 5: Total Debt over GDP for Crisis Countries and Germany Portugal Spain Greece Ireland Germany Source: Eurostat Figure 6: EU: Total Debt over GDP (2013) Source: Eurostat Except for economies with large off-shore centres (Luxembourg, Cyprus and Ireland) the countries of the EU with the highest overall debt over GDP in 2013 were Portugal and Greece (Figure 6) followed by Netherlands, Denmark and Spain.

9 Deleveraging by the five major banks in Portugal has proceeded at a high rate, with substantial cuts in credit and with increase in deposits (Figure 7). At the beginning of 2010 all banks, except the Figure 7 Portuguese banks: deleveraging ratios ST CGD BES-NB BCP BPI Source: Bank reports. Deleveraging ratios are defined by credit less provisions for bad credits over client s deposits. In most of the cases the ratio refers to consolidated accounts, but in several cases (BPI) refers to domestic aggregates. public bank, CGD, had leverage ratios above 180%, relying mostly on external wholesale short-term financing. At the end of 2014 all banks satisfied the condition that the deleveraging (transformation) ratio should be below 120%, as set by the troika Memorandum, except for Novo Banco, the good bank after the bankruptcy of BES, which suffered a high drainage of deposits. Let us look now at the different components of debt by economic agent, starting by the Private Sector (households, large and medium small enterprises (SMEs)) and then Public Sector Deleveraging by the Private Sector The Portuguese Private Sector has the second highest leverage ratios among crisis countries, only surpassed by Ireland (Figure 8). Spain and Ireland had already embarked in a strong deleveraging process that barely started in 2013 in Portugal. All these countries are already in phase IIb of the debt cycle, while Greece is still in phase IIa due to the large drop in GDP. Figure 8: Gross Private Debt over GDP

10 Portugal Greece Germany Ireland Spain Source: Eurostat The countries in the EU with the highest private sector debt ratio in 2013, excluding off-shore centres, were Netherlands (230%), Denmark (224%), Portugal (203%) and Sweden (200%) (Figure 9). Taking into account the differentials in income per capita of at least 50%, among this group of countries the Portuguese situation is even more serious. In fact, the ratio for Portugal should be much lower because there are much less households with access to credit as well as a larger share of SMEs without access to bank financing. E.g. the informal sector is much larger in Portugal (about 20% of the economy) than is any other Nordic country (less than 5%). The simple lack of collateral by those households and firms justifies that a larger share of the GDP (denominator) should be excluded in the case of Portugal. 4 Figure 9: Private Sector Debt Ratios (2013) 4 We have made this argument repeatedly in the late 1990s and early 2000s, e.g. in our book on the Portuguese Economy. It was rejected e.g. by the governor of the Central Bank at the time. The arguments supporting an increase in the debt ratios were that Portuguese banks had after the accession to the Euro a wider access to financing and there was a higher level of intermediation due to financial liberalization. We now know how these arguments led to the financial crisis not only in Portugal but also in Ireland and Spain.

11 Belgium Bulgaria Czech Republic Denmark Germany Estonia Ireland Greece Spain France Croatia Italy Cyprus Latvia Lithuania Luxembourg Hungary Malta Netherlands Austria Poland Portugal Romania Slovenia Slovakia Finland Sweden Source: Eurostat In the run-up to the global financial crisis the overall debt ratio of Portuguese agents increased by 89 pp of GDP at an annual rate of 13 pp (Table 1), with households and private corporations contributing each with about 37 pp and the non-financial public sector with 15 pp. From 2007 to the eruption of the Portuguese financial crisis there was a further increase of leverage of 33 pp at an annual rate of 11 pp, mostly due to the non-financial public sector with an increase of 20 pp. From the starting of the crisis until 2015 we estimate an increase of the overall debt ratio of 41 pp of GDP annual rate of 8 pp, exclusively due to Public Administrations to cover the government deficit and recapitalize the banks. The first peak of the household debt level was reached in September 2009, but the second maximum was reached in 2012 largely due to the fall in GDP. It was only by then that a steady decrease in the debt over GDP ratio started to hold. From 2012 to 2015 we estimate a decrease of almost 10 pp, returning to the level of late Table 1: Portugal debt ratios by agent over GDP

12 2000 Var 2007 Var 2010 Var 2015 Var 2020 Var 2030 Non-financial public sector Private corporations of which: Construction Real estate Enterprises: micro Small Medium Large Households Mortgages Consumption and others Total External financing Source: Banco de Portugal and Author s calculations Figure 10 gives the decomposition of the deleveraging between credit and nominal GDP, composed by real GDP and inflation. The reduction in the stock of loans started in the third quarter of 2011 and reached a rate of 4% in The fall in GDP as well as the decrease in the price index limited the decrease in the debt ratio, and only in 2013 with inflation and in the last quarter of that year with positive growth of GDP, combined with the cut in the stock of loans, the debt ratio was substantially reduced. Figure 10: Deleveraging by Households GDP Inflation Nominal debt Debt/GDP (RHS) Source: Banco de Portugal, INE In the case of corporations the peak of the debt ratio was reached in the last quarter of 2009, but it was only after 2011 that the debt ratio had an accelerated fall: from the peak to the end of 2014 it fell by about 20 pp. The largest rate of decrease in the stock of loans was at the end of 2013 reaching

13 8%, but at the end of 2014 was even higher, at a rate above 8%. The lowering of inflation rate presented a challenge for the reduction of the debt ratio. Figure 11: Deleveraging by Corporations GDP Inflation Nominal debt Debt/GDP (RHS) Source: Banco de Portugal e INE Large Corporations represent about 30% of total enterprise debt. Large means that the corporation is quoted in the Lisbon stock exchange, PSI-20. The stock of bank loans of the 10 largest PSI-20 corporations (EDP, PT, GALP, SONAE, JM, SEMAPA, REN, MOTA-ENGIL, NOS) increased from 14.8 to 16.5 Billion Euros from 2010 to 2014, an increase of 11%, while the stock of bonds increased from 20.8 to 23.9 Billion Euros, an increase of 15%. However, the deleveraging process only started in 2012 with a decrease of 13% in loans, partially compensated by an increase of 4% in bonds. Figure 12 shows the main aggregates for this group of firms. The increase in equity was mainly the result of recapitalization of GALP in The fluctuation in the net debt, that takes into account the liquid assets, was rather stable, but showed the same evolution as gross debt. The debt/equity ratio, in gross terms, shows a similar behaviour to gross debt, with a steeper increase due to the recapitalization of GALP. Total EBIDTA remained rather stable in the period , despite the deep recession in the economy. The ratio of EBIDTA to equity was also rather stable, at around 28%, if the operations in PT and the recapitalization are excluded. However, the situation and evolution differ substantially by company. PT increased substantially leverage, has a high debt/equity ratio, and registered a 950 Billion loss in its investments with BES. EDP and REN also have high levels of leverage, but on a stable path. SONAE has high leverage but was able to decrease the debt ratio throughout the period. GALP had low leverage ratios and in a

14 downward trend. SEMAPA and MOTA-ENGIL experienced a rapid increase and then small decrease. MOTA-ENGIL and JM are the companies with the highest profitability ratio. 5 Figure 12: Leverage ratios of top non-financial corporations 80, , , ,000 40,000 30, Gross Debt Net Debt Equity Debt/Equity (RHS) 20, , Credit and collateral The credit channel emphasizes the role of collateral values to explain the credit cycle. There two important asset prices that are relevant: house prices for households and real estate for commercial use and stock prices for enterprises. The cases of house booms in the 2000s were Ireland, which already experienced a fall of 48.5% since the peak of 2007 to 2013 and Spain with a drop of 35.3%. In Portugal house prices decreased by 13% in the period. 6 These reductions of asset values justify a significant deleveraging as collateral values are reduced. However, we suspect that in all three countries these reductions may be near the bottom. 5 Measured by the ratio of EBIDTA over Gross Equity. 6 The largest expansion in investment in housing in Portugal took place in the 1980s and 1990s. There is no reliable price data to judge if there was a significant price increase in the 1990s. In the first decade of the XXI century investment in housing was already stagnating due to excess supply.

15 Figure 13: House prices Ireland Portugal Spain Denmark Source: Eurostat For enterprises the value of their assets is closely related with the stock market. From the beginning of 2010 to July 2012 stock prices dropped to half of their value. There was a strong recovery until April 2014, as prices increased by 72%. After that date there was a substantial relapse, accelerated by the collapse of BES. By January 2015 stock prices had dropped already by 40%, approaching the minimum of Thus, the recent evolution of stock prices presents a challenge for recovery of enterprise production, leading to intensive deleveraging, as we see in Figure 11. In fact, the evolution of stock prices are highly connected with Tobin-q, since the prices of capital goods has remained almost constant. So, a sustainable recovery of investment is probably delayed by sometime into the future. Figure 14: PSI-20

16 04/01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ /03/ /04/ /05/ /06/ /07/ /08/ /09/ /10/ /11/ /12/ /01/ /02/ Source: Euronext Credit by firm size Portuguese enterprises are the second most leveraged in the EU (Figure 14) after Slovenia, with Italy and Spain in the third and fourth level. The leverage ratio reached a peak in 2008 due to the global crisis in Portugal and Spain, in 2010 in Slovenia (Figure 15). Italian firms have also experienced a substantial deterioration in the leverage ratios in the period. More recent data for large corporations the previous sample shows that the debt/income ratio for Portuguese enterprises has further deteriorated from 2012 to 2014 (from 418 to 437%) due to the decrease in enterprise income. Figure 15: Net debt over Enterprise Income (2012) Source: Eurostat

17 Figure 16: Net debt over Enterprise Income Portugal Spain Ireland France Italy Slovenia Source: Eurostat Figure 17: Credit growth by size Total Micro Small Medium Large Source: Banco de Portugal One of the statements made frequently is that small firms have their credit cut first than large firms. In fact, as Table 1 shows, the debt ratios over income of large and micro enterprises were similar in But Figure 17 shows the largest cuts in 2011 through the end of 2013 affecting small

18 enterprises. Large and micro enterprises were the least affected. After the middle of 2013 micro enterprises suffered the largest cuts. In terms of shares of the bank s portfolios large enterprises increased its share by 7 pp over the period, with the largest losses for small enterprises. Thus, the data confirms the statement Deleveraging by the Public Sector The increase in Public Debt in banking crisis is mainly due to recapitalization of banks, but this was not the case in Portugal. Table 1 refers that the increase in the debt of Public Sector, non-financial, was 56 pp of GDP. Let us study what were the factors that led to that extraordinary increase. Let us start with the recapitalization of the banking sector. The first major banking crisis originated in BPN. BPN was nationalized in November 2008 after the regulator found losses totalling 700 million Euros that have been recorded in an offshore account not integrated in the bank accounts, leading to a negative net worth. The bank had total resources of 4.78 Billion Euros at end of In order to cover the deposits the public bank, CGD, gave a loan to BPN, with a guarantee of the State of 4.2 Billion Euros. To dispose of the assets the State set up three funds: Parvalorem, Parpaticipadas and Parups. In 2012 Parvalorem auctioned four tranches of its assets amounting to 3.4 Billion Euros. At the end of 2013 Parvalorem and Parups had budgeted Billion Euros to be reimbursed to CGD for payment of the loan. And Participadas had a negative net worth of 176 million Euros. At the end of 2012 the expenditure registered in the Government Budget was Billion Euros. In the meantime, BPN was sold to BIC, an Angolan bank. According to a Parliamentary Commission the liabilities for the government from the BPN nationalization totalled 3.5 Billion Euros at the end of The estimates of the final cost to the State are still difficult to make but may amount to about 4.5 to 5 Billion Euros. The resolution of BES was made through the Resolution Fund (Fundo de Resolução), with a recapitalization of Novo Banco of 4.9 Billion Euros. The banks contribute with 1 Billion Euros and the State gives a loan of 3.9 Billion that will have to be reimbursed when Banco Novo is to be sold/privatized. If the proceeds of the sale are not enough to reimburse the State the banks will have to pay the remainder, through the Fundo in instalments. In the case of a smaller bank, BPP, with assets of about 500 Million Euros, the bank was liquidated. Part of the clients was reimbursed by the Deposit Guarantee Fund (Fundo Garantia Depósitos), and the rest lost their investments. In the MoU of May 2010 there was an amount of 12 Billion Euros targeted to bank s recapitalization. At the end of 2012 the banks had used about half of that amount in Cocos, Contingent convertible bonds (BPI: 1300; BCP: 2500; CGD: 700; Banif: 1000). At the end of 2014 all had been paid back except for Banif and 250 by BCP.

19 Thus, contrary to Ireland, the impact of the banking crisis in the Public Debt indicators was minimal. Table 2 gives the evolution of Public Debt in the narrower concept of Maastricht, when compared with Table 1. 7 Table 2: Public Debt (Maastricht concept) Total ( ) Stock Debt end-year Increase Effects: Primary deficit Interest and snow-ball effect Interest Nominal GDP Others d.q. Bank recapitalization Privatizations (-) Public Enterprises Source: Ministério das Finanças, Government Budget, annual reports The main conclusion is that the increase in Public Debt, after 2010, when the Program of Stabilization started, was the snow-ball effect (increase in interest payments and fall in real GDP) of about 20 pp, as well the reclassification of debts and inclusion of public enterprises in this statistic, with 13.4 pp. The primary deficit only contributed with 9 pp, mainly in 2010, at the start of the stabilization program. 2. Macroeconomic policies and deleveraging As a member of a monetary union Portugal has no independent monetary policy instruments for deleveraging, so labour and product market rigidities are particularly pernicious. There are two main channels via which rigidities make the impact of households' deleveraging relatively more painful. First, the optimal reaction of the economy to deleveraging is a downward adjustment in real wages. If working properly, this mechanism allows sustaining a stable level of employment leading to a smaller fall in output. Real and nominal wage rigidities shut this channel by making the downward wage adjustment slower (labour becomes relatively more costly and firms shed workers, which results in an additional decrease in output). The second channel works via a price effect on the interest rate. While in an economy characterized by flexible prices, prices fall relatively more on impact, they quickly start climbing back towards their original level. In effect, after an initial short period of deflation, moderate inflation sets in. In an economy without independent monetary policy these changes have a direct effect on the real interest rate. In fact, a few years after the start of the deleveraging process the real rate actually falls below its initial level and only then starts converging to the steady state. In contrast, in the rigid economy, the real interest rate remains above its steady state for an extended period of time. Since lower interest rates make corporate investment cheaper, investment falls less in the flexible economy, helping to sustain a relatively high level of output. 7 The main differences are debts of public enterprises.

20 A contagion from the housing sector to other economic sectors, as captured by a generalised increase in the risk premia in the deleveraging economy, aggravates the negative impact of the shock. This effect is primarily visible in a significantly larger fall in corporate investment. Consumption, employment and GDP also fall markedly more than in the benchmark case. This may thus give an idea of the magnitude of the effects when both households and non-financial corporations sectors face deleveraging pressures. A higher degree of openness can attenuate the negative impact of the shock by the stabilising effect of increasing net exports in the wake of falling domestic prices. Foreign trade dampens the impact of the negative demand shock for domestic production especially for the production of traded goods. Unemployment would then increase less and consumption fall less during the process of deleveraging. Note that a larger increase in current account as a share of GDP takes place: while exports and imports move (up and down, respectively) relatively less in the (more) open economy, due to their larger share in the total output, their aggregate adjustment is larger in terms of output. Public sector coupled with private sector deleveraging can generate substantial cumulative effects. Public sector deleveraging may aggravate the fall in GDP and all domestic demand components. Since private deleveraging deteriorates the government's budget balance through the effects in economic activity, the government needs to undertake substantial and credibly permanent restrictive fiscal policy measures to achieve a reduction in the public-debt-to-gdp ratio over the medium term. Moreover, as in the case of the debt-deflation spiral, also real public debt is affected by emerging deflation: falling prices increase the level of real debt which further increases, although temporarily, debt-to-gdp ratio. The best environment for overall deleveraging to take place is in an economy experiencing GDP growth and some inflation. Historically, about one third of deleveraging was due to real income growth and one third due to price increase. 8 Price increases will have the major impact if nominal interest rates are below inflation, i.e. with negative real interest rates. A deleveraging strategy should involve both public debt and private debt, and both strategies need to be coordinated, because of the macroeconomic consequences. A reduction in public debt, which is essential because of the sustainability of public finances and retain market access, should be carried out gradually and by further reducing public expenditures. It is not well recognized but the core instrument in the macroeconomic adjustment of the internal and external imbalances of the Portuguese economy in was the reduction of credit to the economy. This has been the key instrument in any IMF adjustment since the inception of that institution, and in theoretical terms is the most effective way to reduce aggregate domestic demand. Figure 18 shows the reduction in the domestic credit to the economy in four of the countries that experienced the strongest adjustment. When compared with these economies Portugal had the lowest rate of adjustment, which may be due to two factors: the weight of external assets in the 8 Of the average decline in credit to GDP of 40 percentage points inflation and lower credit contributed 14 percentage points each, and real economic growth a further 11 percentage points. Tang (2010).

21 2009Apr 2009Aug 2009Dec 2010Apr 2010Aug 2010Dec 2011Apr 2011Aug 2011Dec 2012Apr 2012Aug 2012Dec 2013Apr 2013Aug 2013Dec 2014Apr 2014Aug 2014Dec Jan 2009=100 national banking was substantially lower in Portugal than in Ireland, and domestic deposits increased substantially more than in economies like Greece that was subject to substantial withdrawal of deposits and outflows of capital. From January 2010 to December 2014 credit of the Portuguese banking system decreased by 18.8%. Over the four year period this represented a rate of 5% p.a.. Figure 18 Bank Credit to the Economy Greece Ireland Portugal Spain Source: Author s calculations based on ECB data Taking the total period from 2010 to 2014, including the beginning and end years, nominal GDP remained almost constant (Table 3), with a cumulative fall of only 4.2 and an increase in the deflator of 4.6. This is a remarkable change when compared with the cut of 18.8% in credit, revealing a strong increase in the velocity of money. Table GDP real -4.2 GDP deflator 4.6 Nominal GDP 0.2 Private Consumption -6.0 Public Consumption -9.3 Investment Imports 2.1 Total Domestic Demand -5.4 Net Foreign Trade 8.5 Credit Source: Author s calculations

22 Decomposing the change by domestic demand and net foreign trade, the reduction in total domestic demand of 5.4% was compensated by the increase in net foreign trade of 8.5% of GDP, due to an increase in exports of 31% and just a slight increase in imports of goods and services (2.1%). Private consumption retracted by 6% while public consumption had a larger drop (9.1%), but it was the demand for investment that had the largest reduction (31.7%). 3. Debt Sustainability Tests Despite the importance of the subject, sustainability tests are still in its infancy. There are three types of tests used by most economists: (a) Long-term transversality condition, equivalent to no Ponzi games, which excludes the possibility of debt explosions, so once the country reached a high level of debt it as to enter into a more or less long phase of decreasing debt ratios. The IMF is one of the institutions using this method. Basically the simulations start with some assumptions about interest rates and GDP nominal growth and it considers that the debt level is sustainable if it enters into a phase of steady declining ratio of debt over GDP. However, the time period or the end-point ratio are not precise and depend on the country circumstances. Research at the European Commission is based on the concept of stationarity (Cuerpo (2013) which is quite similar. The methodology suffers from several shortcomings. First, it does not predict a crisis. Second, it is extremely sensitive to the GDP level and its evolution, which is hard to predict on the long term. Third, the long-run projections are also very sensitive to assumptions about the interest rates. However, it is relatively simple to apply and has become an acceptable method worldwide. (b) Debt capacity simulations can be carried out for classes of households and firms to evaluate their solvency capacities. This method is very informative for the short and medium term, but requires a substantial amount of data. The BIS uses it in the 2014 Annual Report to get broad estimations. However it is doubtful if such shortcuts can provide any reliable information. (c) Financial distress indicators referring to both supply and demand of credit are also useful. These indicators, like the prices of insurance of debt instruments against bankruptcy, the amount of NPLs, indicators used by rating agencies, and so on, are usually short-term leading indicators of debt problems. The difficulty is that they cannot be used to make projections over the medium and long-term, so the other two methods above have to be used. Non-Ponzi sustainability We build two scenarios: a base scenario and a more stringent scenario that differ on the benchmark required for sustainability. The base scenario (Figure 19 and Table 4) assumes for benchmarks: (i) the base scenario of the IMF- CE for the Public Sector, (ii) for enterprises a benchmark defined by the average of EU countries without distress in their debt levels (Belgium, Germany, France, Italy, Austria, Finland and Sweden)

23 for 2013, to be reached by 2030, and (iii) for households a benchmark of the same EU countries with similar methodology. Figure 19: Portugal: Debt Ratios over GDP Micro Ent Small Ent Medium Ent Large Corp Public Sector Consumer credit Mortgages Dez 2008 Jun 2008 Dez 2009 Jun 2009 Dez 2010 Jun 2010 Dez 2011 Jun 2011 Dez 2012 Jun 2012 Dez 2013 Jun 2013 Dez 2014 Jun Dez Source: Banco de Portugal and Author s estimations In this scenario it will take up to 2030 for the economy to return to sustainable leverage ratios. In this scenario, households are in the turning-point, enterprises will have to make most of the deleveraging until around 2020, while the public sector will take longer. Overall, the debt ratio will have to be cut by 80 pp over 15 years, but the debt ratio will just be reduced to the 2007 level. Table 4: Base scenario for deleveraging ( ) 2015 Var 2020 Var 2030 Non-financial public sector Private corporations of which: Construction 16.6 Real estate 11.5 Enterprises: micro Small Medium Large Households Mortgages Consumption and others Total External financing

24 Source: Author s estimates In a more stringent scenario (Table 5), required by an aging population, the private sector will have to deleverage about 5 years sooner. The overall debt over GDP will have to be cut by 130 pp, about the level of 2003, pre-crisis. The benchmarks are thus defined by the precrisis levels, a methodology followed by a number of authors (e.g. BIS and Rogoff). The main differences in the deleveraging process are for corporations that reduce their leverage ratio by an additional 30 pp the double effort with the level for 2030 at 80%, which is also observed in some of the best European countries and BIS considers a benchmark. Second, the Public Sector reduces the debt ratio by an additional 15 pp to a ratio of 103% of GDP in 2030, substantially closer to the Maastricht level of 60%, since our projections include public enterprises. Debt service indicators Table 5: A more stringent Scenario for deleveraging ( ) 2015 Var 2020 Var 2030 Non-financial public sector Private corporations of which: Construction 16.6 Real estate 11.5 Enterprises: micro Small Medium Large Households Mortgages Consumption and others Total External financing Source: Author s estimates Debt service ratios have traditionally been used to gauge the sustainability of external debt in developing countries, like measuring debt service over exports of goods and services. They have rarely been used to gauge sustainability by economic agents. Figure 6 shows the estimation of overall debt service for Portuguese households. It shows a sharp increase in and then a substantial fall up to January The current levels are near the minimum observed in 2010, due to the sharp fall in the Euribor (see above Figure 1) coupled with the deleveraging that has been taken place.

25 2009 Jan 2009 Abr 2009 Jul 2009 Out 2010 Jan 2010 Abr 2010 Jul 2010 Out 2011 Jan 2011 Abr 2011 Jul 2011 Out 2012 Jan 2012 Abr 2012 Jul 2012 Out 2013 Jan 2013 Abr 2013 Jul 2013 Out 2014 Jan 2014 Abr 2014 Jul 2014 Out 2015 Jan Figure 20 Portugal: household debt service Source: Author s estimation based on Banco de Portugal data This evolution explains the low delinquency rates still observed in mortgage credit. What are the most vulnerable households? Costa and Farinha (2012) carried out a micro-analysis of debt levels of Portuguese households. Using her data we can say that only about 30% of households that contracted mortgage credit can be considered vulnerable: households with the lowest 20% of income had in 2012 a debt service ratio of 70%! and the next echelon had a ratio of about 30%. Comparing household indebtedness across the Euro area (Figure 21) shows that Netherlands and Denmark are the countries of our sample with the largest service ratios. Portuguese households benefit from the fact that most of their contracts are at variable rates indexed to the Euribor 6 months, while in some other countries (Netherlands, Germany, Belgium and France) are set at fixed rates (See ECB (2009), pg. 24). Debtors benefit when short term interest rates are low, as in the last 10 years. However, this is not a secular situation, so when short term real interest rates increase, the delinquency rate will certainly increase in Portugal.

26 Figure 21: Households debt service over Disposable Income Source: Author s estimations based on Eurostat data Obs.: We assumed a 20 year average maturity for all mortgages of all countries Average debt service ratios are not good predictors of financial crisis, as data for USA show. The ratio for American households show a relative stable ratio in the 2000 to 2013 period, and no sharp increase in the eve of the American financial crisis of Financial distress of banking debts The banking crisis has generated accumulated losses of about 18 Billion Euros, considering the recapitalizations and contributions by different institutions in the period (Table 6). Taking into account only the 5 major banks, the cumulated losses reached 11.6 Billion Euros, and despite recapitalizations, total equity in the balance sheets has decreased by 4.3 Billion Euros. Table 6: Banking losses and recapitalization ( ) Equity 2010 Reported Losses Equity 2014 Sant-Totta 3, ,809 CGD 7, ,493 BES 5,578-7,500 4,900 BCP 7,153-3,025 4,376 BPI 1, ,547 Total 25,474-11,614 21,125 Memo: Private recapitalization 6,039 Resolution Fund 4,900 Shareholders and bondholders (BES) 2,851 State Debt 2,203 Client s losses 350 Deposit Guarantee Fund 170 Total 16,513 Cocos outstanding 1,500 Total 18,013

27 2010-I II III IV 2011-I II III IV 2012-I II III IV 2013-I II III IV 2014-I II III IV Source: Author s estimates based on Bank Reports Have the signs of financial distress abated substantially? We start by looking at the ROE in (Figure 22). Profitability showed a decreasing trend due to the resolution of BES in 2014, but excluding that factor there are encouraging signs that the situation is turning around. Figure 22: Return on Equity Ratios ST CGD BES-NB BCP BPI Total Source: Banks Reports The most troublesome case is BCP which registered cumulated losses of 3 Billion Euros, but has closed the year of 2014 with break even in the last quarter. The loss in BPI was due to exceptional factors linked to regulation of its subsidiaries. Another indicator of financial distress is the evolution of impaired assets in the banks portfolios. Figure 23 shows that the highest increase in the ratio of non-performing loans (NPLs) to total credit is due to loans to enterprises (corporations). The overall ratio for the banking system jumped from 3.4% at the start of 2010 to 8.6% at the end of 2014, which is still a low ratio by international standards. A decomposition by agents shows that the largest increase is due to corporations: the ratio jumped from 4 to 14.3%. Although the ratio accelerated at the end of 2011 there were some signs of deceleration at the beginning of This deceleration was not more pronounced because of the sharp cut in the credit that occurred since June NPLs of household credit have been quite contained, with a much higher rate of consumer credit than for mortgage loans.