The remaining three factors have lower levels of explained variance and simpler structures. Factor four (6.1% of explained variance) differentiates regions in terms of the provision of hospital services (PROHOS) and beds (PERBEA). It can be called a “physical health facility” factor. However, these two variables are not associated with the distribution of doctors (DOCPOP) which constitutes factor five (5.1% of explained variance). Clearly, this is a “medical service”dimension. Somewhat interestingly, the next highest loading on this factor is a positive one associated with small business income (SMABUS). Positive outlier scores are recorded for factor four for Koryak AO, Evenk AO, Tamyr AO, Chukotsk AO, and Moscow, while Ingushetia constitutes the only negative outlier. For factor five, the positive outliers are North Osetia, Moscow, and Koryak AO, with Yamal Nenetsk AO, Ust-Ordynsk AO and Khanty-Mansiysk AO being negative outliers.

The final factor (4.9% of explained variance) combines profitability of assets (PROASS) and investments in fixed assets relative to the previous year (IFAPY).

It could be labeled a “new economic growth” factor. However, since the data relates to conditions for one year only, there is no way of determining if this dimension is indicative of long term underlying conditions in the regions. It should be remembered that the Russian economy had a major setback in 1998 when the Rouble was devaluated by a factor of three against most western currencies. As part of the adjustment process, the subsequent changes of the following year had tremendous regional variation. Positive outliers on this factor are Koryak AO, Vologda oblast, and Gorny-Altai, while Marii El, and Chukotsk AO are negative outliers.

Cluster Analysis

Hierarchical cluster analysis, using Ward’s linkage method, was applied to the set of factor scores associated with the six factors derived in the previous section. Since we considered an appropriate number of clusters for the regions to be between twelve and six, we examined all solutions within this range. The nine cluster solution was found to be the most clearly defined. The size distribution of the nine clusters is reported in Table 4, while the composition of the clusters is reported in Table 5.5.

There are a number of interesting features of this set of groups. First of all, a little over two-thirds (60 out of 88) of the regions are grouped into just two clusters, suggesting a considerable degree of homogeneity amongst many of the regions in terms of the variables summarized by the six factors. Also of note is that three of the clusters consist of individual regions, Moscow, Ingushetia, and Koryak AO. This indicates that these three regions are very distinct relative to the rest of the regions and is also indicated by the appearance of these regions as outliers in the factor scores (Moscow, three times and Ingushetia and Koryak AO twice each).

Since the ultimate purpose of the analysis is to assist in defining clusters to be used for the development of regional policy, it is important to examine the spatial distribution of group membership (see Figure 5.1). As shown in Figure 5.1, only one cluster is spatially contiguous. This is a cluster of three consisting of Tyumen oblast, Khanty-Manslysk AO, and Yarnal Nenetsk AO. However, the two largest groups do show a considerable degree of spatial contiguity, although both are split into several spatial subsets. The most geographically dispersed groups are those with eleven and eight members.

In order to avoid geographically dispersed groups, which are not desirable for the purposes of regional development, we added two variables to the set of factor scores. These were the x and y coordinates of the centroids of the regions. These values were scaled so that the range was typical of the ranges for the factor scores. The size distribution, composition, and spatial locations of the nine groups which result from applying Ward’s method to this data are shown in Table 5.4, Table 5.6, and Figure 5.2, respectively. Several changes are apparent from the clusters derived without the centroids. First of all, the regions are more evenly distributed over the nine clusters. However, Moscow and Koryak AO remain as single region clusters, reinforcing the extent of their distinctiveness. Spatially, the clusters have more integrity with two noticeable exceptions. The first of these is Samara oblast which is two regions removed from the nearest region of the cluster to which it is assigned. The other is composed of the regions of Ust-Ordynsk AO and Aginsk-Buryat AO neither of which are assigned to the same groups as the regions that surround them. However, after closer scrutiny of these anomalies, the clusters shown in Figure 5.2 provide a useful basis on which to define spatially contiguous clusters of regions.

RussiaSmall95 and RussiaSmall99

Since the intention in studying these data sets was to examine the extent of change that has occurred from 1995 to 1999, the results of their analyses are reported together.

Factor Analysis

For the 1995 data set the 14 variables are reduced to four factors with eigenvalues greater than one, which account collectively for 78.1 per cent of the total variance (see Table 5.7). Interestingly, the first three individual factors are relatively uniform in terms of the percentage of variance they explain. The factor loadings are shown in Table 5.8. The first factor (27.4% explained variance) has high positive loadings for PROASS, ACCPER, ELDABA, and high negative loadings for YOUABA and ENTLOS. One might label this factor “the mature socio-economic sector”. It has a substantial number of elderly and few young people, residential space per person is high and businesses are relative prosperous. There are only negative outliers (Ingushetia, Aginsk-Buryat AO, and Tyva) for this factor.

The second factor (23.1% explained variance) has high positive loadings for INFIA and MIGINC. This identifies regions with strong in-migration and higher levels of investment in fixed assets per capita. One could label this as a “human dynamic, big business growth” factor, since little income comes from small business enterprises. There are only positive outliers for this factor (Yamal-Nenetsk AO, Khanty-Manslysk AO, Tyumen oblast, Evenk AO, Nenetsk AO, and Taymyr AO).

The third factor (19.2% explained variance) has high positive loadings for PERBEA and IFAPY and a high negative loading for PROHOS. Interestingly, this factor combines growth in fixed assets in 1995 over 1994 with higher numbers of people per hospital bed and lower levels of hospital services. One could label it a “medically deprived and 1995 economic spurt” factor. This factor has positive outliers for Ingushetia, Stavropol krai, St. Petersburg, and Tyumen oblast and negative outliers for Koryak AO, Evenk AO, Taymyr AO, Chukotsk AO, Komi Permyatsk AO, and Nenetsk AO.

The final factor (8.4% explained variance) has high positive loadings on DOCPOP and RETCAP and a high negative loading on DEMLOD. This dimension represents regions with more doctors, higher retail sales and fewer dependents. Clearly, this is a “health-wealth” factor. High outliers occur for Moscow, Chukotsk AO, Kamchatka oblast, Magadan oblast, and St. Petersburg, while negative outliers are recorded for Ust-Ordynsk AO, Komi Permyatsk AO, and Aginsk Buryat AO.

In contrast to 1995, in 1999 the 14 variables are summarized by five factors, although the total explained variance is almost identical (78.8 per cent) (see Table 5.9). This suggests that interrelationships between at least some of the variables have weakened between the two dates. Comparison with the factors from the 1995 analysis (see below) also indicates that the nature of the relationships between some variables also changed. The factor loadings are given in Table 5.10.

The first factor (27.5% explained variance) in 1999 has high positive loadings for DEMLOD, ELDABA and MIGINC, and a high negative loading for INFIA. There is no similar factor to this in 1995 since it includes variables that were on three different factors at that date. This factor resembles somewhat the second dimension from the 1999M1 data analysis. In this case, one may label it a “depressed living” dimension where the number of dependent (especially elderly) is high, people are still moving in (perhaps young people coming home to live with the elderly parents), investment in fixed assets is low and people have to create small businesses (SMABUS = 0.519) to earn a living. There are no positive outliers for this factor but negative outliers occur for Yamal Nenetsk AO, Chukotsk AO, Khanty-Manslysk AO, Magadan oblast, Tyumen oblast and Kamchatka oblast.

Factor two (19.9% explained variance), with a high positive loading on PERBEA and high negative loadings on PROHOS and IFAPY, is similar to factor three in 1995, except for the change in sign of IFAPY. Clearly, this identifies regions where the number of persons per bed in hospitals is high, other hospital services provisions are also poor, and the investment in fixed assets per person in 1999 is also low. It is a “poor health and poor investment” factor. Positive outliers occur for Ingushetia, Yamal Nenetsk AO and Samara oblast, while negative outliers are recorded for Koryak AO, Evenk AO, Taymyr AO and Chukotsk AO.

The third factor (15.3% explained variance) has a high positive loading for ACCPER and high negative loadings for YOUABA and ENTLOS. These three variables all loaded on factor one in 1995, although YOUABA had the opposite sign. It represents regions with more housing space, fewer children, and fewer businesses running losses. On may label it as a “successful businesses with older workers” dimension. Only negative outliers (Ingushetia, Aginsk Buryat AO, and Tyva) occur for this factor.

Factor four (8.5% explained variance) has high positive loadings for DOCPOP and RETCAP. These two variables were part of factor four in 1995. Again it is a “health-wealth” dimension. Only two outliers, both positive (Moscow and North Osetia), occur for this factor.

The fifth factor (7.6% explained variance) is a single variable one, PROASS, which in 1995 was part of the cluster of variables loading on factor one. Since the next highest is a positive one for IFAPY, this factor could represents regions with profitable assets together with some indication of investment in fixed assets also taking place. It represents an “economic potential” dimension. Positive outliers occur for Koryak AO, Vologda oblast and Gorny Altai, with negative outliers occurring for Marii El and Chukotsk AO.

Finally, it can be noted that the five factors for the 1999 data are consistent with five of the six factors obtained from the 1999 analysis involving 24 variables (see Table 5.3). However, there is no factor equivalent to factor one obtained from the larger data set. This can be explained by the relative absence of income related variables in the smaller data set.

Cluster Analysis

Once more we considered solutions between 12 and 6 groups. For both 1995 and 1997, the most appropriate solution was seven clusters. The size distribution of the clusters for both years is given in Table 5.11.

The most obvious feature of the 1995 solution is that 52 of the regions (almost 60 per cent) are grouped into one>

By 1999 both the composition and the spatial distributions of the clusters had changed considerably (see Table 5.13). The largest cluster now contains 37 regions and there is a second large cluster of 31 regions. However, there are now two single region clusters, Moscow and Koryak AO. Ingushetia is no longer unique but instead is clustered with five other regions. Collectively, there appears to be greater heterogeneity in the regions in 1999. This is also reflected in the spatial distribution of the groups (see Figure 5.4). The largest group consists of three spatially contiguous clusters of regions plus five geographically separated regions, while the other large group is composed of two spatially contiguous clusters plus two separate regions.

Comparison of Tables 5.12 and 5.13 reveals that the major change between 1995 and 1999 was the division of the one large cluster in 1995 into two clusters in 1999. As Figure 5 shows this division occurred along geographical lines (a major north-south split and a more minor east-west one). There are also noticeable geographic trends in the changes in the smaller groups.

Conclusion

Given the above results, we feel confident that a central government regional development policy can be created which would have different objectives, procedures and limits for each of the nine different planing regions of the RF. Each policy would have to identify the major problems in the planning regions and propose solutions for them. Further study and refinements should produce still clearer geographically continuous planning regions. To make sure of the cohesion within these planning regions and major differences between them, more variables in rate formats should be analyzed. Furthermore, one should analyze each year since the early 1990's to see if the pattern of Russian regional groupings is stable or stabilizing.

We recognize that this is only one possible model of a typology of the RF and its regions and the associated regional development policies that could be based on it. There are others, as can be seen in this report, and still others that have not been produced yet. The final choice will depend on the aims and objectives of the RF government.

Table 5.1. Variables Used in the Analyses.

Variable Name

Variable Description

ACCPER*

Provision of accommodation (as of end of year; sq. metres per capita)

AVEINC

Average income per capita (per month; th. roubles; since 1998 – roubles)

AVLSUB

Subsistence level; Ratio of average per capita income to subsistence level; %)

DEMLOD*

Ratio of demographic load (as of 1 January); disable age persons per 1000 persons of able-bodied age; total

Материалы этого сайта размещены для ознакомления, все права принадлежат их авторам.
Если Вы не согласны с тем, что Ваш материал размещён на этом сайте, пожалуйста, напишите нам, мы в течении 1-2 рабочих дней удалим его.