5. Methodology and Framework

5.2. Quantitative Analyses

5.2.2. Taiwan’s Districts

Taiwan’s cities are regrouped into 7 industrial districts. Taiwan’s cities are the first set of regions and tested for correlations since that is how census data grouped. The cluster maps then demonstrate the specialty landscape with the adjacent cities. Those similarly adjacent cities were combined into a broader region (defines district). All the cities are regrouped into 7 districts by its geographic adjacencies with similar cluster landscapes.

This coincidentally matches the local people’s customary divisions, also a part of natural

15 This project is with Professor Porter’s endorsement and the help from by the data manager of Institute for Strategy and Competitiveness, Richard Bryden to define Taiwan’s Cluster Definitions.


division. The national data are divided based on political boundaries of city which may not necessarily reflect the economic boundaries.

From 2011 Taiwan’s Census data, 5 regions were collected from the National Statistics Bureau and are listed below:

North Region – Taipei City, Taipei County, Keelung, Hsinchu City, Hsinchu County, Taoyuan County, Yilan County

Middle Region – Taichung City, Miaoli County, Changhua County, Yunlin County, Nantou County

South Region – Chiayi City, Chiayi County, Tainan City, Kaohsiung City, Pingtung County, Penghu County

East Region – Taitung County, Hualien County Outer Islands – Kinmen County, Lienchiang County

After an evaluation based on the cluster landscape, 7 regions are identified and regrouped, and they are:

North Region – Taipei City, Taipei County, Keelung

Northwest Region – Taoyuan City, Hsinchu City, Hsinchu County, Miaoli County Middle Region – Taichung City, Changhua County, Nantou County

Southwest Region – Yunlin County, Chiayi City, Chiayi County, Tainan City South Region - Kaohsiung City, Pingtung County

East Region – Yilan County, Taitung County, Hualien County Outer Islands – Penghu County, Kinmen County, Lienchiang County 5.2.3. Analytical Framework

With this said, this creates a big learning step for the research not to re-invent the wheel but to adopt to the methodology of user friendly interface. The newly defined US cluster


definitions is the first lesson to learn.15F15F16 The Appendix - Data Preparation is dedicated to describe the first step of the process of this adaptation to normalize Taiwan statistical data retrieved from the Taiwan National Statistics Bureau. After the data is cleaned and

prepared in the growth format, regional data are plotted for analyses. Each city is analyzed with clusters’ growths, and each cluster is analyzed with cities’ growths. From the city cluster map, dominated and strong clusters are identified by the growth analysis.

A dominated cluster has large shares in both cluster and location workforce. For example, Hsinchu has over 50% local workforce working in the IT cluster, and an over 10% of total IT populations working in the Hsinchu. A strong cluster is identified by 5% of total cluster populations in the region, not as intense as the dominated cluster of 10%. The strong clusters tend to grow in a group since many strong clusters have stickiness as a value-chain. More detail analyses will be provided to demonstrate these properties in X Cluster Analyses.

5.2.4. Clusters’ Dynamics with the Desired Presentations

The study of clusters is important to reflect the dynamics of industrial agglomerations either by spillovers or supply-chain expansion. The three calculated growth rates are plotted on 3-Dimentional bubble maps to present the dynamics of clusters in the regions.

International businesses and trades have been going on for centuries, and not until recently the open information act has made competition open on an equal base. The hidden advantage is disappearing due to fast communication and open media, in which limited goods were offered with a price paid, and now goods are in high demanded. The market has become consumer demand driven versus supply driven. Customers are no longer just being provided with the goods and services locally. People are buying and receiving goods all over the world and delivered within days. Data and information of all sorts have also been demanded, so that varieties can be found for a bigger and better consumption. Thus, this makes market behavior directly linking to the industries’

growths with high dynamics. The world transportation and logistics have been operating

16 Categorization of Traded and Local Industries in the US Economy, Mercedes Delgado Richard Bryden Samantha Zyontz; and other articles in US cluster mapping website in cluster definition section. See US Cluster Mapping website for detail definition.


24 and 7 to meet the delivery demands. Studies on cluster shift can show how market is moving and help to create strategies to meet the demanding markets. The circulatory business cycle (Schumpeter, 1939) has shortened its duration and created high dynamics.

Size matters for a particular cluster in a region. The shares of clusters are calculated for the importance of labor productivity. A dominated cluster over 10% workforce nationally has a bigger influence to the GDP. A weak and large cluster can be dangerous for a nation. On the other hand, a small cluster that is growing fast can be potentially agglomerated into a large cluster provided the stimulus is high and suitable for the growths. Therefore, growth rates are important to be measured along with the cluster share sizes. Cluster shares nationally over 5% overalls can be important enough for the cluster. It carries enough weight to become a nationally important cluster. Hence, smaller but potentially fast growing clusters can be the future stars if policy and resources are put into the right demand. Therefore, examining the growth rates and the shares are the first indications to look for in terms of specialization.

5.2.5. Details of Growth Analysis

For the dynamics of clusters, employment, revenue, and wages’ growth rates of each region will be calculated to see the shifts after the external shock among all clusters.

Some clusters grew and some declined. Equally, some regions are growing and some regions are declining. The momentum provides reviewers with indications to search for issues. Along with each cluster size and shares within a region, different categorization may carry different implications. Below is a chart (Figure 2) describing the possible momentums. The last row includes the specializations that are nationally important.

These are the factors when reviewing the growth maps for the understanding of the dynamics. It provides a direction what to look for the next, especially for the microeconomic analysis. Since employment has the strongest social impact, the employment growth is evaluated first. See below guideline.


Figure 2 Cluster Shares and Growth Analysis Guideline.

Cluster Employment Regional Shares

Declining Stagnate or slow growth

Fast growth

Small shares < 5% Little impact to the regional


No significance An indication of potential, but not

5.2.6. Location Quotient – Specialization Calculation

Location Quotient represents the concentration of a certain cluster in a location in reference to the national ratio. If the ratio is over 1 meaning the regional cluster has the national importance. The higher the number, the more concentration of the cluster in the national importance, and it is a highly specialized cluster with the national importance.

Industry LQs are calculated by comparing the industry’s shares of regional cluster factor


with its shares of national cluster factor, and the factor can be employment, revenue, or wages.

Figure 3 Location Quotient Equation.


LQ = a ratio of a local cluster factor shares to the national cluster factor shares.

X = local factor counts of a cluster of a region Y= total regional factor counts of a region X’ = total factor counts of a cluster of a nation Y’ = total national factor counts of a nation

For example: Revenue Location Quotient of the Business Services Cluster in Taipei is calculated:

Figure 4 Revenue LQ Equation.


Below are the implications for LQ and employment shares:

Figure 5 Cluster Shares and LQ conditions.

Employment Shares\LQ

LQ<1 LQ>=1 LQ>=2

34 Regional Shares


No significance Regional specialty and

Regional specialty Region specialty and national

The research will find out where the specialty concentrations are among all redefined clusters for Taiwan, and the cluster map may be evaluated based on the above


5.3. Statistic Correlations

All clusters and all cities’ data have been tested using Pearson correlation. The statistics correlations confirm that Taiwan Traded clusters’ growth rates are highly correlated. The result shows the high correlation on all three growth rates across the board over all cities on traded clusters with exception of the Outer Islands. Additionally, by running the larger redefined district-region correlations which also confirm the correlation. Followed by running K-mean cluster prediction. The result shows there is only one cluster in Taiwan.

The initial correlation started with 6 clusters’ definitions, and it remerged into one after several runs. That means Taiwan has only one dominated industry. The statistical correlations cannot tell in precision which city has which industries actively growing or declining on what factors. In order to find out the answer, a scoring system is developed to identify the dynamics of the clusters in a region.

35 Cluster Scoring Method

A scoring system is developed to identify Taiwan’s regional economic landscape, and specializations, shares, and growth rates are analyzed. The system evaluates on four important factors, location quotient (LQ), employment shares, employment growth rates, revenue growth rates, and wages growth rates. Correlation has been done to confirm the all three growth rates are highly correlated in the Traded Clusters of all Taiwan cities and regions, and very little correlation among the Local clusters’ growth rates. Therefore, only the Traded cluster growth rates are used for the ranking system.

The scoring system includes the following:

Location Quotient: LQ represents the concentration of a cluster in a region. It gives the indication that the regional cluster correlates with the national ratio. A ratio of 1 meaning the region has the same concentration of the national ratios. LQ over 1 means the

concentration is higher than the normal national and is an important location, and LQ over 2 means the cluster is highly specialized in this region.

Employment Shares: any cluster over 5% shares has important presence in the region. It carries a good weight in employment participation.

Employment Growth Rate: The growth rate indicates whether the cluster is actively growing and hiring.

Revenue Growth Rate: The revenue growth indicates the market demand for the cluster.

More revenue normally means more work. Of course there may be other factors influencing the revenue rise. Technology breakthrough with a better production efficiency can also raise more demand such as machine automation with a faster throughput can give more capacity on production and receive more orders.

Wage Growth Rate: The wage is an important gauge for the sustainability of the

workforce. The higher the wages, normally indicates the work skills and knowledge may be high, or there might be more workflow from the businesses. It can also imply that there is a shortage in labor pool.


A scoring system is designed for easy reading of cluster’s characteristics. The scoring system is digitally sliced numerical groups by every three digits to read out the sum of each group for the accumulation of the group counts. Below are the detail definitions of the group representations.

Figure 6 Cluster ID Scoring Digit-Sliced Evaluation System DSE.

Factors Conditions Assigned Value Note

LQ LQ>=2 100,000,000,000,000 Highly specialized cluster and concentrated in a region LQ 2<= LQ >=1 100,000,000,000 Specialized cluster in national

standard Employment

Shares (ES)

ES>=n% 100,000,000 The cluster has a significant presence in a region. N is normally a 5 to find strong clusters.

Positive numbers n means there are n clusters growing.

The higher n is the more

100 Revenue growth means better income for the businesses in general. The number

represents numbers of clusters in growth.

1 Wage growth indicates a sustainable workforce and sustainable businesses. Wage drop can mean oversupply.

Number represents the number of clusters have wage rising.

This Score Card will be applied in the analysis of Taiwan cluster evaluations in the Regional Analyses.


5.4. Types of Clusters – Analyzing the Growth:

In Figure 7, ES (employment shares),we see that New Taipei has 2 clusters over 10%

employment shares. Other cities have 1 cluster are the cities with a higher concentration of one dominate cluster such as IT for Hsinchu City, or the area with less population. The zero indication means there is no one cluster dominates the city. When LQ is =1, this means the regional employment ratio if equal to the national ratio to other clusters. But when LQ >2, it means the ratio is twice as much as the national ratio. The right-hand 3 groups are the accumulations of positive growth rates (EG=employment growth, RG=Revenue Growth, WG= wage growth).

Figure 7 Cities’ Cluster ID Scores.

Cluster ID Score

New Taipei 2,019,002,031,043,030 2,015,002,022,034,020 4,000,009,009,008 Taipei 7,014,001,032,042,040 5,007,001,024,032,030 2,007,000,008,010,010 Taichung City 6,017,000,040,047,040 6,011,000,030,035,030 6,000,010,012,012 Tainan City 8,021,000,035,039,030 8,014,000,024,029,020 7,000,011,010,008 Kaohsiung City 3,020,001,035,041,030 2,011,000,024,029,020 1,009,001,011,012,010 Yilan County 11,014,002,032,039,000 10,004,001,024,029,000 1,010,001,008,010,000

Taoyuan City 4,012,001,042,043,030 4,011,001,029,031,030 1,000,013,012,009 Hsinchu County 5,009,001,042,039,030 5,007,001,031,030,020 2,000,011,009,007 Miaoli County 7,010,001,035,040,020 7,007,001,024,030,020 3,000,011,010,005 Changhua

County 14,009,000,039,043,000 14,005,000,028,032,000 4,000,011,011,008 Nantou County 9,016,000,033,035,020 8,007,000,024,028,020 1,009,000,009,007,000

Yunlin County 7,022,000,038,040,030 7,012,000,026,030,020 10,000,012,010,005 Chiayi County 9,019,000,038,040,020 9,013,000,027,030,020 6,000,011,010,008 Pingtung County 10,015,002,038,042,000 7,008,000,029,032,030 3,007,002,009,010,010

Taitung County 10,006,002,021,027,000 5,001,000,014,021,010 5,005,002,007,006,010 Hualien County 8,010,003,033,036,020 4,002,001,024,026,020 4,008,002,009,010,010 Penghu County 8,010,002,030,022,010 4,001,001,017,015,000 4,009,001,013,007,010 Keelung City 4,017,002,033,038,030 3,008,001,023,029,020 1,009,001,010,009,010 Hsinchu City 2,003,001,033,035,030 2,002,001,020,025,030 1,000,013,010,006 Chiayi City 6,010,001,028,038,020 3,004,000,020,029,020 3,006,001,008,009,010 Kinmen County 8,012,003,034,033,020 6,004,002,022,022,020 2,008,001,012,011,000 Lienchiang

County 6,006,003,012,017,010 4,002,002,007,007,010 2,004,001,005,010,010

* Employment Shares >=10% for the boundary


2LQ=location quotient is 2; LQ=location quotient is 1; ES=cluster employment shares;

EG=# of clusters with employment growth; RG=# of clusters with revenue growth;

WG=# of clusters with wage growth

5.4.1. Dominated Clusters

With the 10% dominated clusters’ tests, the ES digit group shows the numbers of dominated cluster in one city. From Figure 7, and New Taipei, Kinmen, and Lienchiang have two dominated clusters; 9 cities have one dominated cluster. Distribution and Electronic Commerce dominates in New Taipei and Taipei. Distribution and Electronic Commerce dominates in New Taipei, Taoyuan, Hsinchu City and County, and Miaoli.

Construction Products and Services dominates in Yilan, Hualien, Penghu, Kinmen, and Lienchiang. Food Processing and Manufacturing dominates in Kinmen. Water

transportation dominates in Keelung and Lienchiang. Other cities do not have dominated clusters.

5.4.2. Highly Specialized Clusters

Next we are interested in finding out which city has certain highly specialized clusters.

This is by looking at the 2LQ digit group of the DSE Score. Changhua has the most (14) highly specialized clusters, and New Taipei, Kaohsiung, and Hsinchu city have the least (2) highly specialized clusters. It is interesting that Changhau does not have a dominated cluster, but clusters are very specialized. Conversely, New Taipei and Hsinchu City with highly dominated clusters has the least highly specialized clusters. We can conclude that a city with dominated clusters with an employment mass over 10%, the specialized clusters are reduced to the minimum. The dominated clusters have high growth

momentum, which attracts labor pool to converge. This also differentiates two types of cluster agglomeration. One is the dominated cluster type, and the other is the

specialization cluster type. We will later evaluate New Taipei and Hsinchu for the dominated clusters, and Changhua for the specialization type by how each dynamic behave. First let’s understand the growth rates by taking the next cluster size from 5%

shares and up and interpret the indicators below.

39 Figure 8 Digit Sliced Evaluation System.

Cluster ID Score










NewTaipei 2,019,003,031,043,030 2,015,002,022,034,020 4,001,009,009,008 Taipei 7,014,007,032,042,040 5,007,003,024,032,030 2,007,004,008,010,010 TaichungCity 6,017,005,040,047,040 6,011,003,030,035,030 6,002,010,012,012 TainanCity 8,021,005,035,039,030 8,014,002,024,029,020 7,003,011,010,008 KaohsiungCity 3,020,005,035,041,030 2,011,002,024,029,020 1,009,003,011,012,010 YilanCounty 11,014,005,032,039,000 10,004,001,024,029,000 1,010,004,008,010,000 TaoyuanCity 4,012,003,042,043,030 4,011,002,029,031,030 1,001,013,012,009 HsinchuCounty 5,009,003,042,039,030 5,007,002,031,030,020 2,001,011,009,007 MiaoliCounty 7,010,003,035,040,020 7,007,001,024,030,020 3,002,011,010,005 ChanghuaCounty 14,009,006,039,043,000 14,005,004,028,032,000 4,002,011,011,008 NantouCounty 9,016,007,033,035,020 8,007,003,024,028,020 1,009,004,009,007,000 YunlinCounty 7,022,006,038,040,030 7,012,003,026,030,020 10,003,012,010,005 ChiayiCounty 9,019,006,038,040,020 9,013,003,027,030,020 6,003,011,010,008 PingtungCounty 10,015,006,038,042,000 7,008,002,029,032,030 3,007,004,009,010,010 TaitungCounty 10,006,006,021,027,000 5,001,002,014,021,010 5,005,004,007,006,010 HualienCounty 8,010,007,033,036,020 4,002,003,024,026,020 4,008,004,009,010,010 PenghuCounty 8,010,007,030,022,010 4,001,003,017,015,000 4,009,004,013,007,010 KeelungCity 4,017,007,033,038,030 3,008,003,023,029,020 1,009,004,010,009,010 HsinchuCity 2,003,002,033,035,030 2,002,001,020,025,030 1,001,013,010,006 ChiayiCity 6,010,005,028,038,020 3,004,001,020,029,020 3,006,004,008,009,010 KinmenCounty 8,012,007,034,033,020 6,004,003,022,022,020 2,008,004,012,011,000 LienchiangCount

y 6,006,007,012,017,010 4,002,004,007,007,010 2,004,003,005,010,010 total size 5,040,051,042 3,028,039,031 2,012,012,011

*Employment shares >=5

5.4.3. Strong Clusters

The strong cluster is defined: When a cluster with local specialization agglomerates, the cluster starts to converge with the employment growth, and if the market demand continues to be strong, it will become a strong cluster. The strong cluster will further agglomerate into a dominated cluster with concentration and high growing market demand with innovations and a highly skilled labor pool. As long as the market demand remains strong, the specialization continues to evolve and deepen, and the growing labor participation start to dominate the region. The strong clusters are defined to be with large


shares of labor participation with a high concertation of specialization. A dominated cluster is also the strongest cluster. In this case, active strong clusters are defined to be specialized clusters (LQ) that have employment shares (MS) over 5% of regional employments and growing (EG), and the formula is as below:

Figure 9 Strong Cluster Definition with Specialization and Growth Rate Qualification.

1 5% 0

Taiwan’s strong clusters’ landscape are mapped below.

Figure 10 Taiwan’s Industrial Agglomeration – Strong Clusters in all Cities.

Source: Taiwan National Statistics Bureau

Below are the detail active strong clusters in Traded and Local clusters:

Water TransportationInsurance ServicesConstruction Products and ServicesBusiness ServicesHospitality and TourismInformation Technology and…Textile ManufacturingFood Processing and ManufacturingMetalworking TechnologyUpstream Metal ManufacturingProduction Technology and…Local Food and Beverage…Local Health ServicesLocal Hospitality EstablishmentsLocal Community and Civic…

0 50 100 150 200

New Tapei Taipei Keelung City Taoyuan County Hsinchu County Hsinchu City Miaoli County Taichung City Changhua County Nantou County Yunlin County Chiayi County Chiayi City Tainan City Kaohsiung City Pingtung County Yilan County Hualien County Taitung County Penghu County Kinmen County Lienchiang County Empolyment SizeThousands

LQ > 1, Employment Growth >1, Size > 5%

Taiwan's Industrial Agglomeration Landscape ‐

Dominated Growing Clusters in all Cities

41 Figure 11 Active Strong Clusters in All Cities.

Regions Traded Dominated Clusters Local Dominated Clusters

North New Taipei IT Community and Civic

Organizations Taipei Business Services


Food, Hospitality

Keelung Business Services Food, Health Services, Hospitality

Northwest IT Food in Miaoli

Middle Taichung Production Technology Health, Community and Civic Organizations Yunlin Textile Manufacturing

Construction Products

Chiayi Health Services,

Hospitality, Community and Civic Organizations Tainan Upstream Metal


Food, Health Services,

South Kaohsiung Food, Health Services,

Community and Civic Organizations

42 Pingtung Construction Products

and Services

Food, Health Services, Hospitality, Community and Civic Organizations East Yilan Construction Products

and Services

Food, Health Services, Hospitality, Community and Civic Organizations Hualien Business Services

Construction Products

Taitung Construction Products and Services

Food, Health Services, Hospitality

Outer Islands

Penghu Business Services Construction Products

Kinmen Business Services Construction Products

Lienchiang Hospitality and Tourism

Water Transportation


One special note, Nantou is located in the middle of mountain ridge area; although it appears to be in the middle region, but the specialization develops upon its geographical advantage of scenic sites. Its specializations and activities are closer to the Southwest District where the mountain ridge connects the Sun-Moon-Lake in the middle at Nantou and Ali Mountain is at Chiayi County in the south. Often tour buses move from one to the next, and those two sites are rarely missed. Furthermore, Construction Products and


Services are mostly dominated in all regions expect for the middle region and up to the north tip. Below is the map for Taiwan’s Strong Clusters.

Figure 12 Taiwan’s Strong Clusters.

Source: Taiwan National Statistics Bureau

Spillovers can follow the agglomerations on reach a saturation point. Either competition or the labor shortage will trigger the spillovers. The following chapter analyzes some clusters and their spillover potentials.

5.5. Setting up a Web Tool for Users

As an example, the US cluster mapping platform was formally released in September 2014. See below web link:

Construction Products and ServicesDistribution and Electronic CommerceFood Processing and ManufacturingHospitality and TourismInformation Technology and Analytical…Production Technology and Heavy MachineryWater TransportationLocal Food and Beverage Processing and…Local Health ServicesLocal Hospitality EstablishmentsLocal Real Estate, Construction, and…

0 50 100 150

New Tapei Taipei Keelung City Taoyuan County Hsinchu County Hsinchu City Miaoli County Taichung City Changhua County Nantou County Yunlin County Chiayi County Chiayi City Tainan City Kaohsiung City Pingtung County Yilan County Hualien County Taitung County Penghu County Kinmen County Lienchiang County

New Tapei Taipei Keelung City Taoyuan County Hsinchu County Hsinchu City Miaoli County Taichung City Changhua County Nantou County Yunlin County Chiayi County Chiayi City Tainan City Kaohsiung City Pingtung County Yilan County Hualien County Taitung County Penghu County Kinmen County Lienchiang County

在文檔中 台灣經濟形貌: 增進公共政策效益之動態群聚模式 - 政大學術集成 (頁 42-0)