What Explains Location Disadvantage?

The provincial dummies from Table 7 [ PDF 87.7KB | 3 pages ]give a summary measure of provincial location effects after controlling for measurable variables at the enterprise level and for measurable and unmeasurable effects at the sector level. However it is clearly desirable to try to go behind these dummies to understand what is driving the process of locational disadvantage. We do not have adequate data to replace provincial dummies by accurate continuous variables in these regressions, but we can compare the values of the dummies with proxies for locational effects in the wider literature cited in the earlier section.

Demurger et al (2002) provide the most ambitious attempt to disaggregate provincial effects by replacing provincial dummies with two continuous variables, one based on geography (the proportion of the population within 100 kms from the coast) and the other on policy (using a scoring system based on the type of Special Zones in a province). Demurger et al (2002) only provide an average value for the geography variable averaged across regions of the country. Table 8 [ PDF 83.9KB | 1 page1 ]shows the regional averages for their geographical variable and our provincial dummies. For ease of exposition our dummies are given for the profitability indicator only. They are shown as negative numbers, so a higher negative value of the dummy indicates increasing provincial disadvantage. As coastal location is treated as an advantage a rise in the geography measure is an indication of geographical advantage.

Table 8 [ PDF 83.9KB | 1 page1 ]reveals a consistent pattern with lower negative values for the dummies in the coast and central regions, which have high values of the geographic variable. The north east has the highest score of any region by our dummy measure, indicating the greatest negative location effect. This does not seem to be related systematically to the geographic variable however, since the two western regions score well below the north-east in terms of distance from the coast. Furthermore it will be recalled that in their analysis of provincial GDP growth Demurger at al (2002: table 9) report only a modest impact on north-east growth from geographic effects.

It is possible to test for the impact of the Demurger et al (2002) policy variable on our dummies. Their variable is based on a score averaged over the long period 1978-98 (Demurger et al 2002: table 4). Most provinces had introduced some for of special zones by the mid-1990's with little change over the period 1990-1998. As our data refer to 1995-2002 we take the single year score for the policy variable for the year 1995, although for most provinces the score is constant during the 1990's. When the province dummies are regressed on this policy variable no significant relationship emerges and the adjusted R2 is close to zero (see Table 9 [ PDF 52.9KB | 1 page ]).

Of the other factors affecting provincial growth that have been examined in the literature we also test for a relation between our provincial dummies and simple measures of barriers to inter-provincial trade and infrastructure. Data from Poncet (2003 Appendix B table 2) on inter-provincial trade flows in total province absorption are used as a proxy for internal barriers to trade, on the crude assumption that the higher is the ratio of intra-province expenditure to expenditure on goods from elsewhere in the country the higher are internal trade barriers. There is no relation however between this measure and our provincial dummies, again with an adjusted R2 of close to zero (see Table 9 [ PDF 52.9KB | 1 page ]).

We try two alternative measures of infrastructure - telephones per capita and road density (road length/area). There is no relationship between the provincial dummies and the latter variable by province. Where we do find some relation is in a regression of the provincial dummies on telephones per capita. The dummies are positively and significantly related to the former (so regional disadvantage falls with more telephone communications) (see Table 9 [ PDF 52.9KB | 1 page ]). Good communications are normally seen as an important part of the business environment so the first positive relation is not unexpected. It should be borne in mind that in her analysis of the impact of infrastructure on provincial GDP growth, Demurger (2001) finds that her telecommunications variable had a positive effect on provincial growth relative to the national average in all three of the north-east provinces. On the other hand, her transport variable had a negative effect in two out of the three. In both cases infrastructure variables are not the dominant explanation of relative provincial growth in the north-east.

This analysis has offered little help in opening the 'black box' of the provincial dummies. The better known measures on geography and policy do not seem important explanations. Also the basic data on intra and inter-provincial trade flows shed little light. As might be expected infrastructure appears to matter, but different measures give conflicting results. This leaves the key explanation likely to lie with the 'investment climate' in the region. Infrastructure provision may play a role, but given the fact that the region is not particularly poorly endowed with infrastructure by national standards this is unlikely to be the key. Lack of 'marketization', defined as the limited spread of market relations in the north-east, is a widely cited explanation for its relatively poor performance and prospects.¹¹ Marketization is often measured crudely by the share of non-state or foreign-owned firms in economic activity. However our analysis, which controls for these structural features, shows that there are other factors at work. Even allowing for a lower than average role for non-state or foreign - owned firms in different sectors performance is still poor.

This leaves as the key explanation more fundamental features of the investment climate relating to institutional quality via the enforcement of property rights, the application of regulations and the development of financial norms and institutions. At present although work on the investment climate in the region is ongoing, published data refer only to a limited number of cities. The results are also somewhat ambiguous, showing a very mixed picture for the north-east. Out of a ranking of 23 cities nationally Changchun and Dalian are in the top half of the list of 23 (7th and 10th respectively), whilst Benxi and Harbin are in the very bottom group with the worst investment climate (23rd and 21st, respectively).¹³ There is clearly scope for further research in this area to link findings such as our's on firm- level estimates of performance with surveys on the quality of the investment climate.

In general, however, our results leave little doubt that changes in ownership and industrial structure that are often put forward as solutions for the region's problems are unlikely to be sufficient to raise its performance to national levels for comparable activities.

Download this Discussion Paper [ PDF 168.5KB| 25 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.