Investment Strategy and Greek Shipping Earnings: Exploring the Pre & Post 'Ordering-Frenzy' Period

The influx of shipping receipts from the world's leading fleet has been critical for the development of the Greek economy. Following the fateful September of 2008, the range and speed of the shift in direction of the freight rates, combined with the general economic and credit climate, have had a significant impact on Greek receipts from shipping. The paper investigates both the long and short term responses of shipping flows in the Greek balance of payments on the basis of monthly data. It also evaluates the role of the change in Greek ship-owners’ investment strategies from 2006 onwards. The paper presents estimations of the impact of freight market determinants on the shipping flows through a Greek shipping freight rate index, the loans outstanding and the second-hand vessel’s price index. The findings provide evidence in favour of a change in the investment behaviour of the Greek shipping companies after 2006. Inflows and outflows tend to be dictated partly by the cash-flow position of the companies involved in the S&P market, as indicated by the positive relationship with freight rates.


Introduction
Developments in international freight markets, following the culmination of the Following the collapse of the financial markets in September 2008, one of the longer periods of shipping prosperity and successive records for freight rates, which had started around 2003 ended abruptly. The speed of decline in freight rates, culminating in some of the main shipping markets reaching record lows, was similar to the previous rises. Moreover, the market reversal came at a time when the world fleet was projected to grow at record rates, having increased since the start of the new century by about 50% (UNCTAD 2009: 38). In shipping, the adaptation of supply, which entails either resorting to scrapping, dictated mostly by cash-flow pressures (Stopford, 2009), or to cancellations of existing orders, is slow; this makes adjustment both long and painful in times of crises.
Despite a first spate of cancellations after the autumn of 2008, the orderbook at the start of 2009 was still over 40% for tankers and containerships and at over 70% for bulk carriers of the respective current fleet (Clarkson, 2009a) with no immediate prospect for returning to pre-boom levels.
The combination of the steep decline in main markets since late 2008 and the levels of accrued debt obligations, which had been necessary to support the rise in new orders, led to significant cash-flow pressures on shipping companies, second-hand acquisitions and related debt obligations notwithstanding.
At the level of national economies, cash-flow pressures on market players were combined with weaker revenues due to falling freight rates and a significant lay-up rate.
This impacted on shipping earnings which in most instances had shown significant increases alongside freight market levels (Bank of Greece, 2008), especially when earnings are measured in dollars. This increase was also noted for Greece. For a number of shipping nations, reductions in shipping receipts could not have come at a worse time as pressures on public finances were multiplying as a consequence of the crisis. The case of Greece has stood out in this respect within the eurozone in late 2009 due to pressure on spreads on public debt and credit default swaps (CDS).
Apart from their overall contribution to the economy, Greek shipping earnings have been most significant in filling the deficit gap of the current account balance created itself mainly by chronic trade balance deficits. Shipping has served for redressing a variety of imbalances in the Greek economy since the creation of the modern Greek state (Harlaftis et al, 2009) aided, in more recent decades, by tourism and, in previous periods, by significant remittances from Greek emigrants (Thanopoulou 1994; Bank of Greece, 2008). However, both tourism and international shipping are by their very nature widely exposed to the vagaries of the world economic climate. Turning points in the state of the world economy and/or shipping markets are thus of direct relevance to any researcher aiming at tracing trends in the respective net contribution of either activity.
On top of the state of the markets which is largely reflected in the long run by the level of receipts from shipping services (Bragoudakis and Panagiotou,2007) shipping's net contribution to the current account depends also on the goods balance resulting from transactions in ships. The Sale & Purchase (S&P) market, however, can be assumed mainly to redistribute liquidity within a shipowning community, unless patterns of investment shift homogenously towards net sales or net purchases. Thanopoulou (1996) indicates that although there was a clear anticyclical investment pattern among a group of Greek shipowners, homogeneity of the direction of investment was not the case in the past. The combination of the markets' rapid decline and of a peak standing orderbook by Greek shipowners could prove critical for the volatility of the net contribution of shipping, even if second-hand acquisitions and related loan obligations are not to be taken into account.
Greek orders in the upward phase of the present cycle shared until 2006 a number of traits with the traditional anticyclical patterns of the past (Thanopoulou, 2007). Until the early 1990s, anticyclical moves in mainly the S&P market of second-hand ships led -along with a number of other factors related mainly to fleet specialization and the life of vessels -to an increase in the competitiveness of the Greek-owned fleet (Thanopoulou, 1998). However, any similarities with anticyclical investment patterns of past decades (Thanopoulou, 1996) practically disappeared after 2006. The ratio of existing orders to tonnage in service in Greece rapidly converged to world levels in a number of shipping segments (Harlaftis et al, 2009). The combination of weakened revenues and the high orderbook was hardly beneficial for either the cash-flow position of companies or for net receipts from shipping related activities at the level of the Greek economy.
This paper aims at modelling the impact of the crisis on the Greek net shipping balance. Explanatory variables include monthly freight rates and monthly bank loan data, the latter serving as a proxy for cash-flow pressures on shipping companies and the availability of finance. The freight rate index series published by Clarkson Research Services is included in the model after being weighted to reflect the direction of specialisation of the Greek owned fleet. Thus, a single Greek Shipping Freight index is created as well. The overall approach of the authors is novel in that the model both inflows and, separately, outflows of both shipping goods and services. This approach allows to distinguishing and estimating the power of the determinants of inflows and outflows separately. Thus, the approach is opener to one which models than the overall net contribution of shipping to the national economy.
The paper is organized as follows: Section 2 presents the main features of freight markets and Greek shipping investment. Section 3 discusses some theoretical considerations about the impact on the net balance of shipping and presents the data. The econometric analysis is pursued in Section 4 where methodology and techniques are discussed first and the results obtained about the impact of the main drivers on shipping inflows and outflows are also presented. In Section 5 some concluding remarks are summarized.

Freight markets and Greek shipping investment: cycles and reversals
Post-war Greek economic development remained sustainable mainly through the influx of foreign earnings from shipping and tourism, of direct EU transfers and through public borrowing from abroad. Reliance on shipping in this regard can be traced back to the creation of the Greek state (Harlaftis et al, 2009). At the same time, the overall contribution of shipping to the economy has been more than significant (University of At around the same time, the world's total orderbook was standing at historic highs. Indeed, even one year after the financial crisis broke, world orderbook-to-existing-fleet ratio was still near 50% (see : www.Clarksons.net data). Meanwhile, fleet to order ratios have yet to be reduced to anywhere near their levels before the start of the market boom ( Figure 2).
[ Figure 2] In terms of the net contribution of Greek shipping to the balance of current account, outstanding orders and loan balances acquired a pivotal importance. In the very short term, as the state of the markets deteriorated rapidly, the negative impact of the decline in freight earnings was exacerbated by increased obligations resulting from direct payments to shipyards or for servicing early loan instalments. The burden of outstanding debt obligations on a per ship basis tracked the pattern of orders created in the years prior to the start of the financial crisis ( Figure 3) it was highest or high for orders which had yet to be delivered or were delivered late and thus missed or shared little of the highs in the markets.
[ Figure 3] Orders placed -and loan obligations created -before that date would pose no loanservicing problems on a ship by ship basis being either fully repaid or being left with a low ratio of outstanding debt to both projected income and asset values. Despite disparities of positions among companies due to different investment timing and exposure, the overall ratio of orders to fleet in service cannot but impair the degree of liquidity available at the level of the whole national industry, and is reflected in the development of the net contribution of shipping to the Greek current account.
Greek shipping had for many years combined successfully a protracted period of exploitation of vessels (Thanopoulou, 1998)  Due to the fact that TSI are dominated by the receipts of shipping revenues received for the provision of cross-trade shipping services, it is assumed that -in the long-run -TSI depends predominately on the freight rates that the Greek-owned fleet earns. It also depends on the inflows to service loans provided by domestic bank branches through the use of Greek-based operating accounts, depicted both by the outstanding loans to shipping companies and the second hand vessel price index 2 . Grammenos, Nomicos & Papastolou (2008) on the occasion of the discussion of shipping bond defaults. 2 One should also expect that the exchange rate would have a short-term effect through Euro denominated operating expenses of the Greek domiciled shipping companies.
By contrast, as the percentage of the Greek goods export trade is less than 0.1% of the global trade, the structure of the Greek TSO are dominated by the capital outflows for the purchase of second-hand vessels or the construction of new ones. Therefore, it is assumed that the prices of vessels (second-hand or new) as well as outstanding loans should play an important role in determining TSO.
From the above discussion it becomes apparent that the power of the driving forces behind TSI and TSO could be different and any attempt to estimate a one single equation for a hybrid net balance 3 would not be appropriate from econometrical point of view.
The other element of interest is to pinpoint the turning point for Greek investment behaviour. The way to account for the observed change of behaviour is through splitting the period under discussion -in the short-term analysis -into two distinctive sub-periods.

Data
The data are monthly and covers the period from January 2002 to June 2012.The monthly balance of payments data are from the Bank of Greece. The data includes the exports (sales) and imports (purchases) of ships as well as the receipts from and payments for sea transportation services. The data were subsequently expressed in USD terms by the use of the monthly average exchange rate as published by the European Central Bank. be also cointegrated. The terms of µ and ν are the constant parameters, while the parameters of α 1 , α 2 , α 3, α 4 and β 1, β 2 denote the long-run elasticities of (1) and (2). The terms of d1t and d2 t are respectively the long-run residuals.
The specifications of the ECM for the TSI and TSO can be written in a general form as: where ∆ denotes first differences in logarithms and ε 1t, ε 2t ~ N(0, σ2). Dummies have been also added in the models (3) and (4) to capture any impulse disturbance of the endogenous series. It should be noted that the Hendry -type 'general -to -specific' procedure is used to reduce the ECM to its parsimonious form.

Empirical results
The hypothesis of a unit root in the logarithmic levels of the series cannot be rejected. By contrast, the hypothesis of a unit root in the first differences is rejected in all cases in favour of the alternative of stationarity. This finding suggests that all employed series are I(1). In Table 1, we present the results of the unit root tests.
[ In Table 3, we present a suite of alternative ways to test for the existence of cointegration for the long run relationships (1) and (2). The standard ADF (1979) and DF-GLS (1996) unit-root tests provide evidence in favour of stationarity of the long-run residuals of TSI and TSI which means that d1 t~I (0) and d2 t~I (0), indicating that the corresponding series are cointegrated. These findings are also strongly supported by the results of the complementary residual based Engle-Granger (1987) and Phillips-Ouliaris (1990) cointegration tests.
[ Table 3] As it has been noted earlier the estimated parameters of (1) and (2) are much closed across the three alternative cointegration methods indicating the robustness of the models.
Nevertheless, DOLS method was selected as the most robust method and the results of this method are presented in Table 2. The cointegration tests are summarised in Table 3 based on DOLS method of estimation.
Finally, the parsimonious ECM estimation for the short-run dynamics of shipping inflows (TSI) and outflows (TSO) which allows both of capturing any short-run fluctuations and obtaining information on the speed of adjustment to equilibrium are presented analytically in Table 4 and Table 5.

Shipping inflows
The long-run elasticities of the shipping inflows (TSI) based on DOLS (see Table   2) are presented by the following equation: Regarding the short-run shipping inflows TSI dynamics from the results of Table 4 and looking the total sample (Jan.2002-June 2012), it becomes apparent that the shipping freight index with one month time lag affects the amount of the shipping inflows by 0.14% in terms of growth rate.
[ Table 4] The short-run elasticity of α 2 =0.14 which represents the short-dynamic effect of the It should be noted that the estimated short-run shipping inflows dynamic equation (see Table 4) is characterized by relative good fitness and passes all diagnostics χ 2 tests for the hypotheses that there is no serial correlation and autoregressive conditional heteroscedasticity; that the residuals follow the normal distribution; and finally that the equation is well specified. One remaining problem concerns the stability of the estimated parameters of model (3)  [ Figure 5] As Figure 5 shows, the null hypothesis of the parameter stability over the subperiods cannot be rejected at the 5% significance level.

Shipping outflows
On the other side, the long-run run elasticities of the shipping outflows equation (TSO) based on DOLS (see Table 2) are the following: As Greek export trade is relatively small, the outflows for the purchase of secondhand vessels or new ones dominate the shipping outflows. Thus, the latter depends on the index of second-hand vessels price and the provision of loan funds. The pass through rate of the outstanding loans to the shipping outflows accounts for 0.44, meaning that a 10% increase in loan stock leads to 4.44% increase in the shipping outflows. Finally, the impact of the second-hand vessels' price index on the shipping inflows is much higher from 1, which means that a 10% increase in the second-hand vessels' price leads to a substantial increase (14.2%) in the outflows.
In the short-run, we estimate correspondingly an error correction model (ECM) using the estimates of disequilibrium with one lag (d2 t-1 ) to obtain information on the speed of adjustment to equilibrium of TSO. Regarding the short-run shipping inflows (TSO) dynamics at the whole period (Jan.2002-June 2012), the results from Table 5 indicate the significance of the contemporaneous impact of the outstanding loans and the second-hand vessels' index as well as, the important role of the Greek Shipping Freight index.
[ Table 5] Specifically, the outstanding loans have a contemporaneous effect on the outflows β 3 =0.35, in accordance with the fact than any additional loan funds are destined for the purchase of new vessels. The short-run elasticity of β 2 =0.23 which represents the shortdynamic impact of the ClarkSea index on shipping outflows, measures the effect of the one-month lagged value of the freight index on the shipping outflows growth rate. The short-run elasticity of the price index of the second hand vessels is estimated at β 4 =0.80 but it is statistically significant only at 10% level. The exchange rate of the current month affects the short-term decision of the company for transferring funds out from Greece.
The positive sign of this coefficient β 5 =1.09 indicates that when the USD depreciates against the euro then the outflows in USD terms increase. The error correction term coefficient is γ 2 =-0.27 and reveals that the speed of adjustment to the equilibrium steady state is relatively slow which means that the system need almost four months to reach the equilibrium level.
The estimated short-run shipping outflows dynamic equation (see Table 5 [ Figure 6] As Figure 6 shows, the null hypothesis of the parameter stability over the subperiods cannot be rejected at the 5% significance level.

Conclusions
As

REFERENCES:
Alpha Bank (2009 II  III  IV  I  II  III  IV   2004  2005   CUSUM of Squares  5% Significance   II  III  IV  I  II  III  IV  I  II   2010 II  III  IV  I  II  III  IV  I  II   2010  2011  2012 CUSUM of Squares 5% Significance

APPENDIX: GREEK FREIGHT RATE INDEX METHODOLOGY
In previous work (Bragoudakis & Panagiotou, 2007), the ClarkSea index was used as a proxy of the freight rate that the Greek shipping companies were receiving. However, it was apparent that a freight rate that captured the structure of the Greek fleet was needed in the context of the models proposed here. To this end, a new index was constructed based on fleet data and representative TCE earnings for the various fleet segments. The Greek fleet exhibits a high concentration in the dry-bulk and the tanker sector, while its presence in container-shipping is relatively limited (Greek Shipping Cooperation Committee 2010). For the construction of the Greek freight rate index, we firstly needed representative earnings for the three abovementioned sectors, which represent more than 92% of the total Greek-owned fleet capacity and secondly the respective weights these being their shares in the total fleet capacity. The data and the methodology for the construction of the index are presented below:

Data:
The earnings data were retrieved from the Clarkson Research Services-Shipping Intelligence network database (http://www.clarksons.net/). According to the Sources and the Methods for the Shipping Intelligence Weekly (May 2009) "Earnings are estimated as daily time charter equivalents (TCEs) of voyage freight rates, and expressed in $/day on the voyage." Clarkson Research Services publishes weighted average earnings for Dry bulk and Tanker vessels, which were used for the construction of the index 5 . In the case of containers -as a weighted average is not calculated by Clarkson -the earnings of the Sub-Panamax (2,750 TEU) were used as this represent the average size of the Greekowned container vessels. The advantage of using TCE compared to any other freight indices (i.e. BDI or World scale) is the fact that we focus on the actual income of the Greek Shipping companies. This fact is even more important in the case of World scale, which is updated at an annual basis and may create distortions between the actual earnings and the World scale index (Tsolakis et.al, 2003) Weights: The weights to be assigned in each sector were calculated as the share of each of the three sectors (in dwt-terms) on their total capacity, averaged over the estimated period. We have decided to use fixed weights rather than annual updated weights because the structure of the Greek-owned fleet has not experienced significant changes over the last decade.