Explaining Performance Gaps

We use two categories of performance indicators related to productivity and to profitability. For productivity we will use gross output per person (Y/L) and value added per person (VA/L). For profitability we usethe ratio of operating profits to total assets (OP/TA) and the ratio of imputed profits to total assets (IP/TA). Operating profits are taken directly from the accounting reports of enterprises, whilst imputed profits are defined as above (IP = VA – W – FC – D).

OP is a number reported by the enterprises based on their accounting profits and is subject to a very complicated set of rules that could vary across ownership, location, industry, and time. OP is also subject to manipulation by the enterprises when they attempt to hide or inflate profits. IP can be regarded as a proxy for underlying economic returns. Hence when we compare the performance of enterprises across ownership, region, industry, and time it is more useful to use IP.

The above four performance indicators are the dependent variables for the regressions. Table 1.10 [ PDF 373.2KB | 73 pages ] and Table 1.11 [ PDF 373.2KB | 73 pages ] shows the gaps of performance between the north-east region and the rest of PRC by industry at the 2-digit level for the period 1995-2002. Table 1.10 [ PDF 373.2KB | 73 pages ] is ranked by the productivity gaps. Among the 37 industries, the north-east region has only three industries that have higher value added per worker than the rest of PRC: food production, petroleum processing, non-metal mining. The gaps in productivity are 29%, 24% and 3% respectively for the three industries. Clearly the north-east region is on the whole well behind the rest of PRC in industrial productivity.

Table 1.11 [ PDF 373.2KB | 73 pages ] has the same contents as in Table 1.10 [ PDF 373.2KB | 73 pages ] but is ranked by the gap in the return on total assets as measured by imputed profits. As shown in the table, only 3 out of 37 industries in the north-east region have higher returns than in the rest of PRC: petroleum extraction, petroleum processing and furniture with the gaps of 5.5%, 1.5% and 0.3% respectively. Hence, it is clear that the north-east region is well behind the rest of PRC in industrial profitability on the whole.

The above gross performance indicators reflect exactly the impression we have about the north-east region from newspapers and research reports, but seems contradictory to what we have presented in the last section on the improving performance of the north-east . The problem is that the large performance gaps may be caused by many identifiable factors in the region, such as the higher share of SOEs, differences in capital intensity and in capital structure, such as a higher welfare (or non-productive) component of capital. In the following regression analysis using panel data estimation methods, we attempt to explain the performance gaps, by identifying statistically the causal factors involved. We use four basic regression models that are variants of a simple ‘structure-conduct-performance’ approach. These explain performance by variables that reflect scale, factor intensity, competition, ownership and various dummies for type of industry, region of location and time.

Model I: The overall performance gap specific to the north-east region

The regression equation for model I is :

The key variables are as already defined in the previous sections. Subscripts t and i refer to time and enterprise, respectively. All performance variables are explained in the same way, so here output per worker (Y/L) stands for the four alternative performance indicators discussed above. The independent variables include the following:

• ln(L): size of the firm (as measured by number of workers);
• ln(Kp/L): intensity of production fixed capital;
• ln(Kf/L): intensity of non-production/welfare capital;
• ln(M/L): intensity of intermediate inputs;
• Ind3Concentration: Herfindal index for industry concentration at 3-digit industry level as defined in Table 1.4 [ PDF 373.2KB | 73 pages ];
• FIE_ind2MKT_Share: market share of foreign invested enterprises at 2 digit industry level as shown in Table 1.5 [ PDF 373.2KB | 73 pages ];
• D^ind2 Industry dummies at 2 digit industry level;
• Interaction terms between industry dummies (D^ind2) and ln(Kp/L);
• Ownership dummies (D^type);
• Year dummies (D^year);
• North-east region dummy (D^nep).

The main purpose of this regression model is to identify the performance gap between the enterprises in the north-east and in the rest of PRC after controlling for other factors that are not specific to the location effects. In other words, the regression coefficient for the north-east region dummy (D^nep) indicates the performance gap for enterprises in the north-east that is specifically due to the location effects of the region. As shown in Table 3.1 [ PDF 373.2KB | 73 pages ], this is significantly negative, indicating that if all the other controlling variables have the same value, enterprises located in the north-east are likely to have lower Y/L by 10 percentage points, lower VA/L by 44 percentage points, lower OP/TA by 2 percentage points, and lower IP/TA by 4 percentage points compared with enterprises in other locations in PRC. This performance gap is specific to the north-east region and cannot be explained by the other controlling variables included in the regression.

Most of the controlling variables have systematically significant effects on performance. These effects can be summarized as follows:

• The size of the firm has small but significantly positive effects on both measures of profits and output per worker (Y/L), but strong and significantly negative effects on productivity as measured by VA/L;
• The intensity of production and non-production fixed capital and intermediate inputs on the whole have a strong and significant contribution to all performance indicators, so more capital and intermediate inputs per worker raises performance;
• The higher is industry concentration the lower is performance;
• The presence of foreign invested enterprises improves the performance of all enterprises in an industry;
• Performance varies significantly across industries;
• The non-state enterprises perform much better than the SOEs. In particular, in terms of VA/L, the gap is as large as 36% to 80%. The details of the performance gap due to ownership effects are shown in Table 3.2 [ PDF 373.2KB | 73 pages ];
• The performance of all enterprises has improved over the years, particularly since 2000, as shown in detail in Table 3.2 [ PDF 373.2KB | 73 pages ].

While standard reforms such as privatization, the introduction of market competition and foreign direct investment can certainly help the north-east region to improve enterprise performance, the region still has a performance gap that is specific to its own location. As our regression results are based on a very large and representative data set and on a rigorous econometric approach , the conclusion here is much more robust and systematic than that drawn from casual observation or limited case studies.

Model II: The performance gap by ownership, industry and year within the north-east, whilst removing its negative location effect.

This model controls for the same scale, technology, competition and ownership used in Model I, whilst in addition removing effect of the negative location-specific impact of the north-east region. The specification of regression Model II is the following:

The control variables are the same as in Model I, but here in addition to the north-east regional dummy (Dnep), we include a set of four interaction terms:

between D^nep and ln(Kf/L);
between D^nep and D^ind2;
between D^nep and D^type;
between D^nep and D^year.

The first set of interaction terms (D^nep * ln(Kf/L)) is designed to check for the effect of nonproductive or welfare capital on performance, that is specific to the north-east region. We often hear the claim that the enterprises in the north-east have larger social burdens, such as the provision of employee housing and fringe benefits like schools and hospitals. If this effect is important , it will show up as a statistically significant negative coefficient for this interaction term. The results in Table 3.3 [ PDF 373.2KB | 73 pages ] show the coefficient is positive, although it is statistically not very significant.

The second set of interaction terms is designed to compare the performance gap between the north-east region and the rest of PRC by industry, after controlling for systematic factors, like sacle and ownership, included in Model I. It should be noted that the control variables here include in addition the north-east dummy (D^nep) and the standard (that is the noninteractive) industry dummies (D^ind2). Thus the coefficients on the interactive terms (D^nep * D^ind2) will indicate the performance gap after controlling for both industry and location effects. In another words, they will indicate the gap in productivity or profitability between enterprises in the north-east and those in the rest of the country, after (and only after) the negative impact of the north-east region’s locational effect is taken way. The regression method allows us to do this counter-factual exercise to decompose the impact on performance arising from different sources. We should read the results here together with our comparison in tables Table 1.10 [ PDF 373.2KB | 73 pages ] and Table 1.11 [ PDF 373.2KB | 73 pages ] on the gross performance by industry. In Table 1.10 [ PDF 373.2KB | 73 pages ] and Table 1.11 [ PDF 373.2KB | 73 pages ], we only find 3 out 37 industries in the north-east, which are performing better than in the rest of PRC. In Table 3.3 [ PDF 373.2KB | 73 pages ] and Table 3.4 [ PDF 373.2KB | 73 pages ], after controlling or taking away the impact of various factors, there are 26 out of 37 industries in the north-east region, which are doing better than the rest of PRC, using imputed profits (IP/TA) as the measure of performance. Among the 26 better performing industries, for 12 the differences with the rest of the country are statistically significant. In another words, the poor performance as shown by the north-east region in the industry by industry comparison of performance indicators is largely due to a range of systematic factors, for example relating to ownership and competition, and to locationspecific effects. If the north-east region can catch up in these location and non-location specific areas of reform and development, these results suggest that the region has the potential to have many more nationally competitive industries. However for this potential to be realized all of these disadvantageous features of the region will have to be corrected.

The third set of interaction terms (D^nep * D^type) is designed to check if the non-state enterprises in the north-east region are doing exceptionally better or worse than the nonstate enterprises in the rest of PRC after controlling for location and non-location factors, since there is a hypothesis that the north-east region does not have a good environment for the development of non-state enterprises. The results confirm this perspective. The private enterprises in the north-east region seem to perform much worse than those in other regions even after controlling for the negative impact of the general location effect that is associated with the north-east region. This unfriendliness towards private ownership however does not apply to the foreign and Hong Kong, China invested firms, which seems better able to deal with local business environment than the purely domestic private enterprises.

The fourth set of the interaction terms (D^nep * D^year) is designed to examine the timing of the performance of enterprises in the north-east region. The results show that relative to enterprises in the rest of PRC, enterprises in north-east region have improved their performance significantly since 1998. It seems that the central government’s policy of invigorating the north-east region has had positive effects, in addition to the general cyclical recovery since 1999.

The magnitude of the effects of the above interaction terms are shown in detail in Table 3.4 [ PDF 373.2KB | 73 pages ] for convenience of comparison. Table 3.5 [ PDF 373.2KB | 73 pages ] ranks the 37 industries based on the north-east region’s specific industry performance advantage, taking away the impacts of systematic factors such as industry, ownership, market competition, and locational effects. This table should be used when considering the potential comparative advantage of industries in the north-east region. For investors already operating in the region (and therefore already affected by its locational disadvantages) Table 3.5 [ PDF 373.2KB | 73 pages ] should be much more helpful than Table 1.10 [ PDF 373.2KB | 73 pages ] and Table 1.11 [ PDF 373.2KB | 73 pages ]. However there is still a broad similarity in ranking between actual profitability performance (as in Table 1.11 [ PDF 373.2KB | 73 pages ]) and potential performance (as in Table 3.5 [ PDF 373.2KB | 73 pages ]); for example, Petroleum activities are the most profitable relative to similar activities elsewhere in the country in both tables and Leather activities are the relatively least profitable in both tables.

Model III: The overall performance gap specific to the three provinces

Regression model III is similar to model I, except that the north-east region dummy is replaced by three provincial dummies: D^Liaoning, D^Jilin, and D^Heilingjiang. Hence the model is

This regression is designed to check if the three provinces in the north-east region have performed differently relative to the rest of PRC. The results in table 3.6 show that the provinces performed similarly and have almost the same performance gap with the rest of PRC. Hence there is no significant ‘within north-east effect’ with all three provinces sharing similar locational disadvantages.

Model: IV: Performance gap specific to location by province

In regression Model IV, the three provincial dummies are replaced by 28 provincial dummies in order to get a ranking of the performance gap across PRC’s provinces that is specific to location effects, in another words, taking away the systematic impacts from the control variables, like scale, technology, ownership, competition and so forth. The specification of the model is the following.

The results are shown in Table 3.7 [ PDF 373.2KB | 73 pages ] and Table 3.8 [ PDF 373.2KB | 73 pages ]. Relative to the rest of the country the three northeast provinces performed best when using imputed profit (IP/TA) as a measure of performance. By this measure the ranking out of 28 provinces is 21 for Heilongjiang, 22 for Jilin, and 27 for Liaoning. Controlling for everything else, the locational disadvantage creates a lower profit rate of about 7 percentage points for Heilongjiang 22 and Jilin and 8 percentage points for Liaoning.

² This table shows that there is a long way to go for the three provinces in improving their business environment, in addition to the standard reforms such as privatization, market competition, and attracting FDI. In other words even if they brought their situation up to the national level in terms of ownership and competition they would still have substantially lower enterprise profitability due to their locational disadvantages. Although we are not quite sure what accounts for the location-specific barriers to performance as a broad explanation we speculate that institutional infrastructure and the openness of the local economies, are likely to be the main factors.

Download this Discussion Paper [ PDF 437.4KB| 90 pages ].

Comment(s)

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