A comparison of economic impact analyses which one works best? Lukas van Wyk, Melville Saayman, Riaan Rossouw & Andrea Saayman

A comparison of economic impact analyses which one works best? Lukas van Wyk, Melville Saayman, Riaan Rossouw & Andrea Saayman

Introduction Problem overview Model comparison Empirical comparison Findings and implications Conclusion Outline

Events are under pressure to indicate the economic contribution it makes. Sponsors and government agencies use this info to justify their involvement in events. The dilemma which approach is most effective? o Added to this, the type and size of an event also plays a fundamental role. Three most common regional impact analysis models I-O, SAM, and CGE models were constructed and applied to analyse the economic impact of an arts festival (i.e. Aardklop National Arts Festival). Introduction

Different models have different strengths and weaknesses An important part of the modelling process is to select the type of model that can shed the most light on the issues considered to be most important while ignoring the smallest number of other elements of the problem that might be considered relevant (Denniss, 2012:3). Since SA festivals have developed as a substantial segment of the leisure and tourism industry, the need to assess their economic impact has increased accordingly. This need has been highlighted bearing in mind that festivals usually require financial support to ensure sustainability. Problem overview

All three general equilibrium models used share many similarities in basic accounting structure. The I-O model is imbedded in the SAM all the sectoral and institutional accounts present in the I-O are also part of the SAM model of the same economy. The SAM is the initial starting point for building a CGE structure of the economy under consideration. In contrast to the I-O and SAM, the CGE model is an optimisation model it provides the optimal solution mix of endogenous variables in response to an exogenous shock. Model comparison

Of the three models, the I-O is the simplest, both in construction and implementation. It is the most naive in terms of assumptions and limitations E.g., fixed input coefficient, static models, linearity in the production system implies a strict proportional relationship between input coefficients and output, no supply-side constraints assumes an infinitely elastic supply on inputs, including labour, into the production process, etc. The main benefit of CGE is that the supply and demand sides are explicitly determined with full price response. How generally applicable neoclassical general equilibrium theory is, particularly at the regional level, depends largely on the strength of the small country assumption. Model comparison

External information e.g. elasticities Product Expenditure Behavioural & technical relationships Social accounting matrix (SAM) Core Input-Output framework Industry Employment Other Households Supply of labour Income by income groups Other Government Imports Exports etc. Market-clearing conditions Prices (e.g inflation, exchange rate wage rates etc.)

Theory would suggest (given the aspects taken into consideration) that the CGE model is the most preferred method, followed by SAM and I-O models. Next, an empirical comparison of the models is undertaken. Results should not be regarded as showing definitive differences between the three models but, instead, are indicative of the general differences which may be observed. Of course, different model structures and assumptions, and different applications would produce a different set of results. Empirical comparison

Impact of visitors spending at Aardklop on production (in ZAR) CATEGORY DIRECT OUTPUT MULTIPLIER TOTAL Foreign 1 964 330 1.43 2 811 215 North West 11 632 936 1.35 15 668 888 Rest of RSA 40 352 051 1.38 55 632 691 TOTAL 53 949 317 74 112 794 Output multipliers of 1.43, 1.35 and 1.38 are calculated, all within the range of 1.1 to 1.5 for other studies' SA rural output multipliers previously estimated for other towns in SA. Multipliers are Type-I. Impact of Aardklop on the local economy (in ZAR) CATEGORY DIRECT INCOME MULTIPLIER TOTAL Foreign 1 964 330 0.66 846 885 North West 11 632 936 0.62 4 035 952 Rest of RSA 40 352 051 0.63 15 280 640 TOTAL 53 949 317 20 163 477 Partial I-O model results

Output and income multipliers at Aardklop (in ZAR thousands) CATEGORY CHANGE IN DEMAND CHANGE IN OUTPUT CHANGE IN INCOME OUTPUT MULTIPLIER INCOME MULTIPLIER Foreign 1 496 2 141 987 1.43 0.66 North West 519 699 321 1.35 0.62 Rest of RSA 1 069 1 474 672 1.38 0.63 Sectoral changes in output for Aardklop (in ZAR thousands) CATEGORY FOREIGN NORTH WEST REST OF RSA Accommodation 628 49 238 Retail 724 383 669 Wholesale 58 15 37 Transport / Fuel 361 51 188 Manufacturing 6 2 5 Restaurants / Fast foods 144 96 177 Tourism, cultural and leisure service 106 67 84 Municipal 114 35 76 TOTAL 2 141 699 1 474 Although 18.85% of respondents who took part in the business survey indicated that additional workers are employed by them during the festival, no estimation of the employment impact can be made due mainly to insufficient data. Partial I-O model results

Direct, indirect and induced impact of spending on production (in ZAR millions, 2006 prices) Sector Direct and Indirect North West Rest of RSA International Induced Direct and Indirect Induced Direct and Indirect Agriculture 0.066 0.058 0.285 0.248 0.019 0.017 Mining 0.036 0.029 0.131 0.103 0.007 0.005 Manufacturing 0.756 0.317 2.980 1.240 0.174 0.071 Electricity & water 0.033 0.028 0.130 0.107 0.008 0.006 Construction 0.083 0.032 0.261 0.102 0.012 0.005 Trade & accommodation 12.961 3.793 43.455 12.167 2.073 0.519 Transport & communication 2.445 1.072 12.461 4.439 0.776 0.239 Financial & business services 3.067 1.336 7.874 4.089 0.380 0.188 Community services 1.726 1.242 5.711 4.144 0.100 0.119 Total (in ZAR million) 21.172 7.907 73.288 26.638 3.547 1.170 Typically tourism-related sectors, i.e. trade & accommodation, transport & comms., and financial & business services are the sectors in the region that benefit greatly from visitors to the festival. Through backward linkages, the indirect and induced effects on manufacturing, as well as community & personal services, are also noteworthy. Induced SAM-based multiplier model results

Total impact of the festival on production (in ZAR millions, 2006 prices) Sector North West Rest of RSA International Total Percentage Agriculture 0.124 0.533 0.036 0.693 0.5% Mining 0.065 0.234 0.012 0.310 0.2% Manufacturing 1.073 4.220 0.246 5.539 4.1% Electricity & water 0.061 0.237 0.014 0.311 0.2% Construction 0.115 0.363 0.016 0.494 0.4% Trade & accommodation 16.754 55.621 2.592 74.967 56.1% Transport & communication 3.517 16.900 1.015 21.432 16.0% Financial & business services 4.403 11.964 0.569 16.935 12.7% Community services 2.968 9.855 0.219 13.042 9.8% Total (in ZAR million) 29.080 99.926 4.717 133.723 100.0% The value of the festival to production exceeds ZAR133 million. If local spending is excluded, the value of the festival is ZAR104.64 million. Visitors from the rest of SA, as well as local visitors, are the main contributors to the increase in production. SAM-based multiplier model results

Total impact of the festival on household income (in ZAR millions, 2006 prices) Indirect and induced impacts Sector Total impact Low-income households Other households Total households Percentage Agriculture 0.693 0.02 0.16 0.19 0.3% Mining 0.310 0.01 0.10 0.12 0.2% Manufacturing 5.539 0.17 1.39 1.56 2.8% Electricity & water 0.311 0.01 0.10 0.11 0.2% Construction 0.494 0.02 0.14 0.15 0.3% Trade & accommodation 74.967 3.18 27.46 30.64 55.9% Transport & communication 21.432 0.75 6.25 7.01 12.8% Financial & business services 16.935 0.62 6.29 6.92 12.6% Community services 13.042 0.68 7.49 8.17 14.9% Total (in ZAR million) 133.723 5.47 49.39 54.86 100.0% Festival impact on income of low-income households is key. The low-income categories, which translates into income less than ZAR24 000 per year (i.e. less than US$3 000 per year). Festival generated income of almost ZAR54.9 million for households in the province. Low-income households receive about 10% of this income. SAM-based multiplier model results

Total impact of spending on employment Sector Total impact Labour multiplier Number of jobs Percentage (total) Agriculture 0.693 20.93 15 2.4% Mining 0.310 2.94 1 0.2% Manufacturing 5.539 3.30 18 3.1% Electricity & water 0.311 2.45 1 0.1% Construction 0.494 5.63 3 0.5% Trade & accommodation 74.967 5.65 424 71.5% Transport & communication 21.432 1.45 31 5.2% Financial & business services 16.935 1.96 33 5.6% Community services 13.042 5.18 68 11.4% Total 133.723 593 100.0% Spending stimulus also leads to job creation. Since labour is a key input in the production process, an increase in production is associated with an increase in employment. In the absence of the festival, 593 job opportunities might be lost, most of these in the trade and accommodation sector. SAM-based multiplier model results

Results show: Increase in household expenditure leading to higher GDP growth and employment, higher prices and lower exports, because of less competitiveness leads to a negative impact on the balance of trade. Results to be expected due to the nature of the shock and closure type applied. Why? Prices increased following increase in the demand of goods. The increase in productivity might be ascribed to the increased demand that enabled the output to increase in relation to inputs. Higher prices will have the effect of a weakened real exchange rate, decreased competitiveness and lowered demand for domestic exports. CGE model results: Macro-level

It is clear that the structural effects for the regional model for foreign expenditure is the least influenced of all spending groups. It records a figure of less than 5% of the total increase in output. Financial and business services (0.008%), trade and accommodation (0.007%). The majority sectors reveal a small impact, with the agricultural sector actually experiencing a negative impact. 20.5% of the total increase in output is ascribed to local expenditure. The most significant increase in output is experienced in the financial and business services sector (0.046%), followed by transport and communication (0.029%) and trade and accommodation (0.028%). The rest of SA totals some 74.9% of the total output increase. Financial and business services (0.153%), trade and accommodation (0.111%), transport and communication (0.109%), community services (0.089%) and electricity and water (0.086%). CGE model results: Industry-level

Most industries revealed a positive impact due to the increase in spending, because of the interrelatedness of most industries. The activity in the service industries naturally increased although other industries seem to be less influenced. Possible reasons for these results may be ascribed to the relative share of tourist expenditure in the income base of these sectors. The traded-goods sectors experienced output declines. These industries were negatively affected by the higher domestic prices in their specific industries, as the export of these products declined. Traditional export sectors such as those of agriculture, mining, non-metallic mineral products, basic and fabricated metal products, and import-competing sectors presented by textiles, clothing and footwear sectors serve as such examples. An increase in demand for labour is experienced in most industries. For all other sectors, the employment increase noted is in relation to those of total output. CGE model results: Industry-level

Important considerations when interpreting the observed results: The I-O model constructed and applied is partial in nature and therefore does not capture the induced effect Type-I multipliers. Does therefore not consider other parts of the economy. The I-O is a small-town / place specific model. The SAM (Type-III multipliers) and CGE models applied are both provincial/regional models. Findings & implications

Results using SAM, CGE and I-O models assessing economic impact of Aardklop SECTOR SAM CGE I-O Direct impact as % of total impact 40.4 51.9 72.7 Indirect + induced impacts as % of total impact 59.6 48.1 27.3 Sectors with most total spending Rest of South Africa Rest of South Africa Rest of South Africa Significance of job creation 593 264 Not easily measured Cost of model application Accuracy of measured impact Provincial model is readily available and affordable Relatively not an accurate result Provincial model is readily available and affordable Relatively not an accurate results Development of regional models can be expensive Conservatively measured There is a total impact difference of R30 million when comparing the SAM and CGE results. The almost 29% difference in measuring by the two models, leads to the obvious question as to which is the most reliable. The difference between the I-O model and SAM and CGE models' results is due to the limitations of the partial model already mentioned. The three models applied reveals an exceptional high-low difference of approximately R59 million in total impact. All spending sectors foreign, local and the rest of SA measure the highest impact when applying a SAM model, followed by the CGE and I-O models. Findings & implications

Results using SAM, CGE and I-O models assessing economic impact of Aardklop SECTOR SAM CGE I-O Direct impact as % of total impact 40.4 51.9 72.7 Indirect + induced impacts as % of total impact 59.6 48.1 27.3 Sectors with most total spending Rest of South Africa Rest of South Africa Rest of South Africa Significance of job creation 593 264 Not easily measured Cost of model application Provincial model is readily available and affordable Provincial model is readily available and affordable Development of regional models can be expensive Accuracy of measured impact Possibly grossly overestimated Possibly moderate overestimation Conservatively measured Comparing the calculated results when applying the I-O model, it is evident that a greater economic impact is reflected when SAM and CGE models are used. In view of the significant differences in the measured economic impact when various models are applied to the same event, it becomes imperative that users of the results should ensure that they were obtained by means of a specific model relate accurately to the purpose for which the tool was developed. Finally, a most significant limitation of a partial I-O model seems to be the inadequate measurement of job opportunities created. Findings & implications

The measured economic impact results differ greatly between the various models even when applied to the same festival and given the accompanying assumptions made. The results further support the concern that the economies of hosting communities normally experience high leakages and so the local economy does not benefit to the extent that is generally expected. One area where I-O still has the advantage over more sophisticated models is at the very small-region level (such as small towns), where I-O really provides the only option to planners, despite its known limitations. A key consideration is that the act of selecting what kind of model is to be used to evaluate the economic significance of an event will, inevitably, have an impact on the way that the results will be framed and perceived. Conclusion