Understanding the Economic Contribution of Tourism in Malta: A Literature Review

The paper presents a critical assessment of the key studies which present empirical estimates for the contribution of the tourism sector to the Maltese economy. The observed discrepancies in the estimates derived from these studies has in part led to a situation in which there is a widespread divergence relating to the speci c economic contribution of tourism in Malta. The paper evaluates the estimates derived by these key studies in the context of both the strengths and weaknesses of their respective modelling frameworks, which range from the construction of tourism satellite accounts, to input-output models and computable general equilibrium modelling, as well as an evaluation of the overall quality of the data utilized. The paper therefore, attempts to identify and clarify the main causes behind the observed variations in the resulting estimates and through a systematic comparative assessment also aid in the providing a further understanding of the potential economic contribution of tourism to the Maltese economy. On the basis of this assessment it is suggested that the contribution of tourism to the Maltese economy should account for approximately 5.7 per cent of total Gross Value Added when taking account solely direct e ects, 12 per cent once indirect e ects are included and roughly 17 per cent accounting also for the induced e ects. Furthermore, on the basis of this assessment, the paper highlights the need for further research in this area given the signi cance of the tourism sector to the small island economy of Malta.


Introduction
In a number of empirical studies, which will be discussed in the paper, indicate that this estimate may, to a large extent, be overstated.
In an attempt to construct the rst Tourism Satellite Account (TSA) for the Maltese economy, Sacco (2016) estimates a more conservative contribution of Malta's tourism industry at 5.7 per cent of the total Gross Value Added (GVA) (or 6.1 per cent of GDP) corroborating the results obtained from studies which apply input-output modelling techniques. 3

Scope and Methodology
The underlying aim of this paper is to identify and clarify the main causes behind the observed variation of the contribution (or impact) of the tourism sector to the Maltese economy which may be observed amongst the various studies. The methodology employed is primarily a literature review of all studies on the impact of tourism in Malta. The paper will identify estimates of the tourism contribution from various sources and compare and contrast the dierent methodologies employed thus providing a critique of the analysis and a comparison of results. 1 For the latest main tourism indicators refer to Appendix Table 1. 2 WTTC is a forum for travel and tourism industry made up from the global business community and works with governments to raise awareness about the importance of tourism. 3 The Tourism Satellite Account for the Maltese economy was constructed by Sacco (2016) for the reference year of 2010. 2 When evaluating such estimates, it is important to acknowledge that the methodologies 4 upon which they are derived may vary extensively in terms of both modelling complexity and the overall quality of data utilized for the study itself. The comparative assessment provided in this paper, also aims to provide a deeper understanding of how this contribution may have evolved over time as the tourism industry faced various challenges and exploited various opportunities to maintain its key role as a key driver of growth.

Contribution of the tourism sector based on Tourism
Satellite Accounts Within the context of national accounting, an industry is typically measured from the output side by calculating the value added (namely total turnover less intermediate cost) of each producer within that given industry. However, unlike traditional sectors such as agriculture and transport, there is no industry classication under the European Statistical Classication of Economic Activities (NACE) Rev.2. that is specic to the tourism industry (or sector). This issue stems predominately from the underlying nature of tourism activities, in that the sector's supply and demand engages a wide range of activities to and from multiple sectors (Hara, 2008). It should be noted, that it is nonetheless possible to assess those industries which are mostly associated with tourism activities which do have a NACE classication, such as the accommodation industry and the food service industry. Together, these two industries amounted to 5.2 per cent of GVA in 2015. But not all the value added generated by these industries is attributable to tourist demand.
Likewise, the value added generated by museums and heritage sites is classied under the recreation and culture. But most of the visitors are tourists and thus can be classied under the tourism industry.
In order to appropriately measure the full extent of tourism activities from both the demand and the supply side, national statistics oces generally compile what are known as Tourism Satellite accounts (TSAs). TSAs analyse in detail all aspects of the demand for goods and services associated with the activity of visitors; to observe the operational interface with the supply of such goods and services within the economy; and describe how this supply interacts with other economic activities (TSA: RMF 2008). TSAs are not a modelling but an accounting tool that records annual activities of tourism as an industry (Hara, 2012). The TSA is an extension to the system of national accounts which enables an understanding of the size and role of economic activity related to tourism which may not be clearly captured by national accounts.
The TSA is made up of a unique set of inter-related tables that show the size and the distribution of the dierent forms of tourism consumption in a country and contributions 4 For an extensive overview on the various methodologies that can be applied to assess the impact of the tourism sector on an economy refer to Hara (2008) and Dwyer et. Al (2012).
to GDP, national income, employment, and other macroeconomic measures of a national economy. Outputs derived from the TSA can be directly compared to main macroeconomic aggregates produced by the system of national accounting for other industries within the economy (Hara, 2008).
A study by Sacco (2016) presents the rst attempt to construct a set of TSA tables for the Maltese economy in which estimates for the direct contribution of the tourism sector to the total GVA generated in the Maltese economy are presented based on two methodologies. Sacco (2016) estimates that Malta's tourism industry directly generated an amount of GVA equal to ¿330.4 million during 2010, which equates to approximately 5.7 per cent of total GVA. This estimate is generated by subtracting the proportional value of intermediate consumption from the tourism share of output for each tourism related sector. Given that this method is however subject to criticism (OECD, 2000), the net ratio approach, which is an OECD recommended methodology (OECD, 2000), is also applied in order to calculate total direct gross value added (TDGVA). One of the key characteristics of this approach is the application of specic intermediate consumption to output ratios, relating to the main characteristic industry of each product, in such a manner that each product is now allocated a quantity of intermediate inputs which are characteristic to that of the associated industry, rather than the specic product type. The TDGVA derived by Sacco (2016), based on the net ratio approach, estimates that Malta's tourism industry directly generated an amount of GVA equal to ¿330.1 million during 2010. This implies that both methods estimate an approximately consistent level of direct contribution to total GVA generated by the tourism sector, which in 2010 amounted to 5.7 per cent. It is important to note, that tourism satellite accounts assess the direct contribution of the tourism sector but do not include the indirect eects on other industries and induced eects from consumption generated through the generation of salaries and wages. 4. The Input-Output modelling approach to measuring tourism's contribution To capture indirect and induced eects, an input-output framework is required. An inputoutput model is a quantitative economic technique that represents the interdependencies between dierent sectors of an economy. Input-output tables track the output generated by an industry as the intermediate input in the production process of another industry or the nal purchase by the various consumers (Miller and Blair, 2009). Nevertheless, it is important to note that the input-output framework assumes xed prices and the use of capital and labour in xed proportions. In reality, an increase in tourist demand may not necessarily increase output of other industries in the presence of supply constraints in those industries which would compel them to raise prices rather than meet the excess demand through higher production. Thus, the assumptions underlying input-output models may not always hold, therefore leading to possible overestimations of the contribution of a given sector. In addition, the input-output framework assumes that a change to nal demand, and thus also tourism expenditure, will not lead to any changes in the technology of production as well as no changes in the proportion of capital and labour used within the production process. Although one acknowledges the empirical usefulness of input-output methodologies, this framework's underlying modelling assumption implies that empirical results need to be evaluated with caution, particularly in the presence of signicant changes in the composition of tourism expenditure over time and the possibility of potentially overestimating the indirect contribution of tourism.
Through the input-output methodology it is also possible to capture the fact that to generate its output, the tourism industry employs workers in return for wages and salaries, which income they spend on goods and services. Thus, the initial tourist expenditure also generates these`induced eects'. In this respect, another limitation is that the standard input-output framework ignores the impact on savings when indeed an increase in tourism related income need not be entirely consumed and could even be associated with an increase in the marginal savings rate and hence a lower induced eect than the one portrayed by the input-output framework. It should be noted that although three studies which are to be discussed in the subsequent sections compute various types of multipliers, such as the output multipliers 5 and employment multipliers 6 , it was decided that emphasis should be placed primarily on value added multipliers. These are the multipliers which adhere closest to the statistical concept of GDP and thus lead to a better understanding of the contribution of tourism to the Maltese economy.

Industry Linkages and Multipliers in Tourism derived from a IOT for 2001
Despite the limitations highlighted above, the use of input-output models remains important in understanding the linkages between tourism and the rest of the economy and in comparing the multipliers of the dierent industries with those of tourism. As a predominantly service-based industry, it is often presumed that tourism generates large multipliers compared to, say, manufacturing which necessitates the importation of raw materials and energy products which in turn tends to diminish multipliers. 5 A type I output multiplier for a given sector j may be dened as the total value of production in all sectors of the economy that is necessary in order to satisfy a ¿1 worth of nal demand for sector j's output (Miller and Blair, 2009). This multiplier is primarily an indicator of the degree of structural independence between the industries in the economy. 6 The employment-output multipliers derived in Cassar (2015) and Blake et. al. (2003) are also referred to as physical employment-output multipliers (Miller and Blair, 2009). They assess the eects, in terms of monetary income, that changes in the nal demand for a sector have terms of the physical amount of jobs created.   Table 1.
Multipliers measure the impact on the total economy as a result of an initial increase in the nal demand of a specic industry. The value added multiplier measures the value added generated for every Euro spent by tourists in the economy. Value added Type I multipliers are often less than unity as value added excludes the intermediate costs involved in the output generated to meet the tourist demand. Multipliers to a high degree depend on the inter-industry linkages. The more an industry is integrated with the other domestic industries, the higher the multiplier. It is important to note, that the higher the level of leakages in the economy, such as imports, taxes and savings, generally, the lower the overall sectoral multipliers will be (Miller and Blair, 2009).
The tourism value added multiplier derived in Blake (2003) is less than unity (0.63) 6 when taking accounting of both the direct and indirect eects. Direct impacts reect the value added generated domestically by the tourism sector. Indirect impacts represent the value added remaining after several rounds of spending by industries linked with tourism. A multiplier which includes both the direct and indirect eects is referred to as a Type I multiplier. When one includes the change in household consumption generated by changes in the wages and salaries earned as a result of the direct and indirect eects of economic activity taking place in the tourism sectors (i.e. the induced eects), the gross value added multiplier for tourism is still below unity. This is called the Type The study by Cassar (2015) is based on a Symmetric Input-Output Table (SIOT) for 2008 which follows ESA95, which compared to Blake et al., (2003) provides a more updated assessment of the impact of tourism-related industries. Cassar (2015) constructed an industry-by-industry SIOT for Malta for the year 2008, based on the xed production sales structure assumption. In his study, Cassar (2015) was able to derive industry specic multipliers based on the input-output methodology framework at a highly disaggregated 59 industry level. In the absence of tourism satellite accounts, the industries that we consider to be an integral part of tourism are the land transport, water transport, air transport, and accommodation and food services activities. It is important to note, that there may be elements within these industries that do not specically pertain to the tourism industry. Furthermore, elements of the tourism industry may be included in other sectors of the economy, but are not included as part of this analysis.
The value added multipliers generated by Cassar (2015) for the four sectors identied as representative of the tourism sector in Malta are reproduced in Table 2 below and are utilized to compute a weighted average tourism multiplier 8 in order to allow a comparison with the estimate presented within Blake et al. (2003). Notwithstanding the diering reference years of the datasets utilized, and to an extent the methodologies applied, both multiplier estimates seem relatively consistent. 7 A summary of the full set of multipliers relating to the tourism related sectors derived by Blake et al. (2003b), including the employment multipliers, may be found in Appendix Table 2. 8 The weighted average tourism multiplier is derived by multiplying each individual industry value added multiplier to the sectors' proportional percentage share of GVA (or weighting) and aggregating across the respective four sectors.   Type I value added multiplier for land transport is ranked amongst the median of the 9 For the purpose of his study, Cassar (2015) also derives the respective output, employment and income multipliers. A summary of the full set of multipliers relating to the tourism related sectors, including the Type I and Type II income and employment multipliers, may be found in Appendix Table  2. When one also considers the induced eects it is interesting to note that the industry ranking by strength of Type II value added multipliers improves signicantly in the case of accommodation and food services and in air transport, reecting their labour intensity of production. Figure 2 illustrates the value added multipliers of the tourism related industries.  Given the high level of aggregation presented in the input-output analysis provided by NSO (2016), it was only possible to identify two industries which may be considered to be an integral part of tourism, namely the transport sector and hotels and restaurants sector. It is important to note that the aggregated sector for transport incorporates all the activities which fall under the classication of land transport, sea transport, air transport, warehousing and support activities for transportation and postal and courier activities and thus may not necessarily reect the expenditure patterns of solely tourism related activities. Table 3 illustrates the value added multipliers of the tourism related industries as presented within NSO (2016) together with the derived weighted average tourism value added multiplier. 10 The Type I value-added multiplier for the hotels and restaurants sector of 0.63 and that of 0.56 for the Transport sector implies an increase of ¿0.63 and ¿0.56 in value added generated, respectively, per Euro increase in the nal demand for each sector. In terms of their overall ranking, the hotels are restaurants sector is ranked 10 th and the transport sector is ranked 14 th which would indicate an average to low impact when evaluated within the context of a sectoral aggregation equal to 17 industries. It is interesting to note however that the Type I value added multiplier for the hotels and restaurants sector is only marginally lower than that obtained from Cassar (2015), indicating some level of 10 The weighted average tourism multiplier is derived by multiplying each individual industry valueadded multiplier to the sectors' proportional percentage share of GVA (or weighting) and aggregating across the respective two sectors.  Within this evaluation, the contribution of tourism to the Maltese economy (C t ), for the given reference year of the input output table (t), is dened as the total expenditure by tourists 11 (T E t ), for the same year, multiplied by the applied weighted average tourism value added multiplier (W AT M t ) and expressed as a percentage of the total GVA of the respective reference year of the input-output table. The results are presented in Table 4.  Table 4 it is very interesting to note that, notwithstanding the divergences in  Table 5 shows the overall eects of a 10 per cent increase in the 2001 tourism demand (equivalent to Lm32 million). It is clear that signicant crowding out can occur from tourism expansion when proper account is taken of behavioural eects. The results indicated that in the short-run, for every Lm32 million of additional demand, only Lm18.8 million was spent in Malta, thus implying that a proportion of the stimulus from the increase in tourism demand is crowded out by higher prices. High elasticity of demand by tourists is a signicant contributor to this crowding out. In the long-run, far less is crowded out primarily because there is greater labour and capital mobility between economic sectors to meet the additional demand (it improves from Lm18.8 million up to Lm29.1 million). This increase in extra demand also led to an increase in the GDP for Malta. However, this eect is greater in the short-run due to the crowding out eect through labour constraints in the long-run, thus suggesting that the supply constraints in the labour market were deemed to hamper the industry's ability to fully meet the additional tourism demand. 12 For further reading on the application of CGE analysis towards tourism impact assessment see Dwyer, et al., (2003) and Narayan, (2004). 13 For further reading on the interpretation and uses of a SAM refer to Pyatt and Round (1985).  for Malta, unlike the other input-output based studies discussed in this paper, are based on some rather strong assumptions which were required due to lack of available data.
At a sectoral level, an analysis of the three input-output studies assessed in this paper indicates that the tourism-based industries (namely accommodation, food and transport services) are rmly interlinked with other sectors of the economy in general, such that the tourism-based industries generate relatively high value added multiplier eects compared to many manufacturing-based industries in Malta. Nevertheless, such multiplier eects are not amongst the highest in the Maltese economy and are closer to the median observed and quite typical of a service-based industry.
Notwithstanding the importance of input-output analysis as a tool for assessing and monitoring the size and evolution of tourism activity over time, more sophisticated models exist, such as CGE models, which better capture the underlying economic reality. Such models allow for the relaxation of certain assumptions made by input-output models such as wage and price rigidities as well as those relating to supply constraints. The results obtained by both Blake et al. (2003) and by Sinclair et al. (2005), which apply CGE modelling to the Maltese economy in order to estimate the impact of tourism, suggest that the tourism multipliers derived from traditional input-output models could be overestimated. It is however important to evaluate the results obtained from the various studies discussed in this paper keeping in mind the numerous discrepancies which exist between each study in terms of statistical methodologies employed and overall data quality.
The analysis undertaken within this paper indicates that there is signicant need for fur-