Theoretical and empirical study of the efficiency of cities as scientific production units

Размещено на http://www.allbest.ru

Размещено на http://www.allbest.ru

Introduction

To begin with, science as a social institution is under the lens of policymakers lately, but the lens itself yet leaves a lot to desire. Informed decision-making on the science development strategy requires multidimensional analysis of its current performance, however multiple issues obscure this area from being easy to assess. There is a great demand to develop clear and concise framework to tackle those issues, and this thesis is one small step towards this direction. Particularly, the focus is on the scale of city as a major object of study. At this level, research and development activities are assumed to be major representatives of scientific production. The geography of study is limited to Russian Federation. In other words, this work attempts to research R&D progress looking at Russian cities as units of analysis.

Considering city to be a unit of R&D activity is not a mainstream approach for science studies, but neither it is a unique feature of this work. One major justification of using such scale is actual existence of special scientifically intensive territories, such as technopolises, clusters, tech-hubs, science cities etc. They exist all over the world, differ according to size, specialization in R&D, nature of genesis, schemes of financing and legal status. In most cases, they are acknowledged to have higher than average impact on scientific production, which is examined later in this work. In case of Russian Federation, there is a special legal status of “Naukograd”, or science city, indicating high concentrations of research and development facilities in a city or municipal district.

The research problem is that science cities in Russia are unique organizational units which are supposed to have high potential for technology development, but face crisis of its realization nowadays. It means that formation of science cities was driven by historical circumstances which led to establishing of unique ecosystem, that has enormous potential for development, but stays fragile under immense risks of transition from governmentally-subsidized functional organization, to self-sustained science organization unit, competitive in the global market. For instance, due to high level of secrecy many Soviet science cities are located in remote areas, which on one hand creates a problem of transportation accessibility, but on the other hand creates an opportunity to grow and develop surrounding areas without extra pressure and expenses of megapolis areas.

One way to resolve this contradiction between high R&D potential and its problematic realization, is to estimate current situation and identify weak points of science development in each case. This is the main objective of this work, although it is quite challenging to give both comprehensive and data-driven evaluation of scientific performance. Further work is structured the following way:

a) Methodology section draws the particularities of the approach, defines the two hypotheses of the study, justifies the significance of the research question and explains the relevance of the approach chosen to proceed.

b) Theoretical part of the work consists mostly of the literature review and divided into several parts: historical and political context and definitions discussion on what is science city and how it is different from other forms of scientifically intense territorial clusters; theoretical justification of using city as a unit of analysis; worldwide state-of-the-art in studies of cities as R&D organization units, including major research questions and approaches in the recent literature; and finally overview of such studies where same empirical method (DEA) has been implemented.

c) Empirical part consists in the efficiency analysis of scientific production in Russia measured at the scale of territorial units. The study is based on three types of statistical data, aggregated on the city level and analyzed with Data Envelopment Analysis (DEA) - linear programming method of comparative efficiency analysis using multiple inputs and outputs. This method allows to calculate the efficiency score of each city and find out sources of inefficiency among the parameters within each case.

The result of empirical analysis are expected to provide enough information to validate or reject both hypotheses stated in the methodology part, and beyond. From the theoretical point of view, this study contributes to elaboration of methodological framework of assessment productivity of science; from practical standing point, it provide concrete information about productive efficiency of science in Russian cities nowadays.

1. Methodology and approach

1.1 Research hypotheses and their significance

In this thesis, it has been chosen to look at performance of cities in terms of productivity of science. Here are some basic definitions: productivity is understood as a measure of economic performance that indicates how efficiently inputs are converted into outputs. Input is any resource used to create goods and services, output is the quantity of goods or services produced.

Regardless framework, it is possible to break productivity down to ratio of output to input, just measured differently. There is no ambition here to assess the quality of goods and services produced. In case of productivity of science, it is especially important to distinguish quality from quantity assessment. Evaluating how significant are particular scientific works or discoveries is completely different area of studies. For instance, during XX-th century immense amount of resources has been spent on research and development of nuclear weapon. Hundreds of scientists all over the world used to work for years to achieve one goal of building a bomb. In terms of productive efficiency their results would be estimated as inefficient, but it is by no means to say that their job was unimportant or of inferior quality.

Productivity is not overarching, yet an important indicator of scientific development. This measure is broadly used for assessment of economic entities, because it gives sufficient basis for business decisions and satisfies purposes of production planning. Improving productivity leads to increasing profits of enterprises and increasing standards of living in societies. The method used in this work allows to capture economic principles of productivity and extrapolate it to non-economic decision making units, such as cities in this case. Further reaching goal of productivity studies is intended to be in improving efficiency by identifying weak points and targeting it with support measures, such as financial resource re-allocation, legislative support, or human and knowledge capital investments.

The first hypothesis is that science cities are more productive than other cities, those who don't have such legal status. This could be expected basing on the common sense that science cities should be better at science in every possible way. In order to test this hypothesis, it is necessary to estimate productive efficiency of both kinds of cities and compare it with each other. If the efficiency score of science cities is above the rest of cities, then the hypothesis is valid. Other possible outcomes are that science cities are in the middle or at the bottom of productivity ranking of Russian cities.

At the same time, science cities might not really form any cluster of entities with similar productivity situation at all. This case will imply that title of science city doesn't correspond to any concrete productivity efficiency rate and therefore it is insufficient to target science cities as if they were a homogeneous group. On the contrary, this situation will require further case by case analysis of opportunities and threats for any political, socio-economic and other institutional decision making. Detailed case by case analysis, if needed, leads this study further to the second part and opens a room for other hypotheses to be tested.

After calculating the efficiency score for each city and comparing science cities with the rest of the sample, next step is to dig further into the elements that formed the final score. This procedure will be narrowed to the group of science cities because such investigations might have directly applicable outcome only in cities, where R&D performance is already considered to be a matter of policymaking practices. This situation is currently achieved mostly in science cities for objective reasons, such as the necessity to confirm and extend their status every 5 years basing on some historical R&D data on the city level. There are few cases, such as Academgorodok of Novosibirsk or Skolkovo, where science progress is also traced at the scale of territorial unit, but these are exceptions from a basic rule of measuring science at the unit scale of organization (corporate entity).

More specifically, the elements to be analyzed are outputs of scientific production used for calculating efficiency at the previous step of this study. These are outcomes of R&D, represented by bibliometric data, patent activity and commercialization of science. Analysis of each element is aimed in discovering what specific science outcomes could possibly be better with the given amount of resources invested in order to improve cities' efficiency score. In other words, this part is dedicated to investigating sources of inefficiencies in science cities.

The second hypothesis, corresponding to the second part of the thesis, is that the sources of inefficiency differ case by case. This implies that among science cities there is no typical distribution of inefficiency sources all the science cities would have in common. For instance, situation would be that one city is weak on international citation rating, while another's weak point is commercialization of their R&D. If proven right, this entails conclusion that no single policy could be appropriate for all the cities examined, justifying further need to develop individual support measures and discrediting one-for-all incentives. On the contrary, if this hypothesis is proven wrong, this would mean that there is a pattern to characterize common problems of scientific development in all cities, this pattern would justify standardized policies for all science cities even though their efficiency is much dispersed and there is no a single cluster in terms of the score value.

Overall, researches on Science Cities are timely and significant in nowadays Russia. From the theoretical point of view, contribution of this research is seen in the discussion about the origins and essence of science cities, and clarification of its position in broader context of national innovation system both currently and potentially in future. Literature overview will allow to put together various perspectives on the matter and accumulate knowledge for further investigations of this topic. Since public policy on science cities is currently in high gear of development, the discussion on its future might be of interest for every stakeholder of the system. Going through transitional period from closed and largely subsidized towns to autonomous entities on a global technology market scene requires tremendous effort. In order to keep competitiveness under such circumstances in knowledge economy, decision makers are expected to refer to the expert opinion and use accumulated results as a source for their own policy. Elaborated research framework increases overall richness of the discourse inside and outside academia and therefore hopefully leads to more successful strategic development trajectories of science cities.

From the practical point of view, contribution of this study consists in testing non-mainstream, but quickly gaining popularity, model of efficiency estimation on empirical data. Particularly, Data Envelopment Analysis approach will be applied to the results of the previous research on resource endowment in Russian science cities. This exercise is relevant in two senses: firstly, it contributes to methodological discussion on the ways of measuring science on the scale of municipalities as an experimental implication of this method to evaluation real cases; secondly, satisfactory results would contribute to the expert notion itself by giving a new perspective on how efficient Russian science cities are and what are their strong and weak points in terms of productivity.

1.2 Approach description

Following methodology will consist of two parts: theoretical and empirical. Firstly, theoretical will be based on literature review. It will provide all the necessary definitions and justifications for the starting points and assumptions used in further analysis. The essential of them are: why it is relevant to look at scientific production at the city scale? Why are science cities important for overall national science progress? How is productive efficiency measured on the city level and what is state-of-the-art in this area of studies? Answers to these questions will provide an overall picture and comprehensive background for quantitative analysis of the second part.

Basing on the theoretical research, the two hypotheses were formulated:

- science cities are more productive than other cities, that don't have such legal status,

- the sources of inefficiency aren't homogeneous among cities.

The empirical study is designed to test these hypotheses by the following procedure: aggregating statistical information on science performance in Russia on a city level, selecting most relevant indicators for R&D productivity assessment and applying Data Envelopment Analysis (DEA) to obtained data. DEA is applied methodology for efficiency measurement of non-commercial entities. It allows to calculate relative efficiency of each city and to evaluate contribution of each outcome to overall efficiency. In other words, as a result of this analysis efficiency score for all cities will be obtained and inefficient components of R&D production will be identified.

2. Theoretical part

2.1 The historical and political context

The central subject area for this review is academic research on science cities. A formal definition of a science city of the Russian Federation is formulated in the Federal law of April 7, 1999 No. 70-FZ "On the status of naukograd in the Russian Federation", namely “municipal formation with the status of city or district, having high scientific and technical potential, with city scientific-industrial complex”. Here scientific-industrial complex of the science city is defined as “the aggregate of organizations carrying out scientific, scientific-technical, innovative activity, experimental developments, tests, staff training in accordance with state priority areas of science and technology of the Russian Federation”.

In accordance with the above definitions, the characteristics of science cities are not a unique attribute of particular set of Russian cities holding this legal status. Such municipal formations can be found almost everywhere in the world among other forms of territorial entities, such as innopolis, technological hubs, clusters, techno-/science parks and others. Although this study is dedicated to concrete case of Russia, we don't completely exclude other types of territorial scientifically-oriented entities from the review, because it would mean losing a considerable part of the literature, especially foreign sources. However, there are significant differences between these entities, due to the historical circumstances and model of development of these formations which are going to be further observed in this review.

The prototype of the modern technology parks was called "industrial district", which then evolved into research parks in the United States in the 1950s - 1960s, in science parks in the UK in the 1970s and business incubators in the 1980s. The world's First Technopark was established at Stanford University and is now known as Silicon Valley. Development model of the Silicon Valley was initially based on the idea of "Knowledge Cities and Smokeless Industry." Land owned by the University had been leased to companies, often founded by University graduates. The mechanisms of attracting entrepreneurs were low rent, exemption from property tax on the campus, as well as the opportunity to use the University infrastructure. In addition, the proximity to the University for companies meant access to a major source of skilled labor, and for students that meant career opportunities and the chance to apply their new knowledge and ideas.

There are many approaches to differentiate territorial units with high scientific and technical potential. For example, Baronov and Kosunina in distinguish scientific parks, industrial parks and technopolises. According to the authors, scientific parks appeared on the basis of industrial parks, and scientific-industrial parks were based in export-industrial zones. By contrast, technopolises were developed in rural areas in order to increase the level of economic development of a depressed region. Science parks typically focus on the research and development of high technology, promote industrial integration of research and development outcomes. Scientific and industrial parks have as its main purposes the absorption and development of high technology, promotion of the industry growth. For technopolises crucial factors are research, manufacturing, human capital investments. It is the combination of two models: the city based on high technology, and the city based on cultural traditions. Industrial areas of high technology designed to achieve an optimal combination of research, production process, education and standard of living of the population as a basis for the development and production of high technology.

Specifics of the science cities in relation to the above described territorial formations is widely discussed in the Russian literature and public discourse. In particular, these aspects are present in the political agenda of the Committee on local government issues Federal Assembly of the Russian Federation, as evidenced by the report “On the status of state policy on science cities and directions of its development”. Based on this report, regional clusters that subsequently became the science towns were built for the solution of governmental problems of strategic nature related to ensuring competitiveness of the state and its national security. The constructions were carried out since the mid 30-ies, in the times known as "the Soviet period of industrialization." Back then the establishment of settlements was often just a side effect of the construction of industrial facilities, not the main policy goal. As high-tech complexes served the interests of the defense sector, it had to comply with secrecy requirements. which imposed ban on construction in existing settlements. In many cases, scientific and industrial complex would spontaneously become the main pivot of new city, although there are exceptions where the design of the city was carried out ahead purposefully and with higher standards of social comfort.

New systems continued to be established until the middle of 70-ies, and support of those established earlier continued until the middle 80s. They were completely relying on the financing from the state budget allocated through mechanisms reminiscent of a modern state procurement system and Federal target programs. The decline of cities with a high concentration of intellectual and scientific-technical potential began at the turn of 80-ies to 90-ies, when the crisis of the political, economic and social system of the USSR led to the decrease in the level of subsidies, including objects of military-industrial complex oriented to the production of high-tech products.

The term "science city" was proposed in the early 90-ies of to narrow down not quite certain definition of "settlements with a high concentration of intellectual and scientific-technical potential", that could overarch large administrative centers. Currently, the development of science cities and management are regulated by the Federal law of April 7, 1999 No. 70-FZ "About the status of naukograd in the Russian Federation" and a number of presidential decrees and Government resolutions of the Russian Federation.

Bearing in mind above described specifics of the science city, yet it is not necessarily to limit the scope of attention to it as we are aiming to consider the phenomenon broadly. This review examines the research on “cities of science”, understood in a broad sense, so the main driver of literature selection is reflected in the approach to research object. The common point in approaches is that one combines the two major dimensions in consideration:

a) integrity and autonomy of territorial (urban/municipal) systems,

b) development of scientific-industrial complex and R&D performance in it.

The importance of combining both aspects for the development of science cities logically follows from the etymology of the term “Naukograd” (rus. nauka - science, grad - city), and, in addition, is discussed separately in.

Regarding the studies of modern realities of Russian science cities, the most comprehensive and fundamental approach presented in recently published monography of Monahov, Klushnikova, Barsukova, consisting of a) retrospective analysis of the phenomenon of Russian experience of development of science cities; b) country wise comparisons of science cities in innovation system of Russia and the UK; C) a comprehensive analysis of the current state of Russia's science cities; d) theoretical and methodological foundations of intellectual capital management of science cities.

In addition, numerous studies, projects, and events are held at the Higher School of Economics within the framework of advisory support for the development of the `Strategy for scientific and technological development of Russia', namely its part on regional aspects of the development of science and technology. For example, the working group at HSE led the development of socio-economic strategy of Troitsk (one of the science cities). To obtain a general picture of the current state of this work it's best to see the materials of the conference "Development of strategy of development of territories with a high concentration of scientific, technological and industrial potential. From cluster to city of science", held on 5 July 2016. The issues raised were:

- How to develop territories with high concentration of scientific and technological facilities?

- Social aspect and living conditions in the territories of scientific/technological development,

- What should be the main emphasis in the preparation of the Strategy for socio-economic development of the science cities?

- How to coordinate measures aimed at the development of research complex with the rest drivers of growth, such as innovative business, industry, social sphere in science cities?

More theoretical studies aimed to answer these and other questions, as well as the analysis of cluster policy and guidance materials specifically focused on Russia can be found in the electronic library of the Russian Cluster Observatory, NRU HSE ISSEK.

2.2 Why city is a legitimate unit of analysis

The basic unit of analysis is selected to be the city or municipality for several reasons. First, large cities and local labor resource allocation play a significant role in the innovative development and production of knowledge. Secondly, the territorial variable gives the opportunity to assess the level of concentration of scientific production. Thirdly, the geographical proximity has an effect on the intensity of cooperation, including R&D sector.

Fourthly, one of the original motivations of the research was to find an integrated approach that takes into account both endogenous and exogenous parameters of development of science. At the same time, this approach should allow to use quantitative analysis model, without overloading it with hardly measurable data. Consideration of city as a unit of scientific production solves this problem, as it includes two main dimensions: scientific one and spatial one, or dimension of urban environment. In focus of the first are issues that are traditional object of scientometric research, the focus of the second -- urban studies. Within both disciplines there are self-relevant quantitative metrics, and the combination of both enriches each perspective with a new understanding of content and context of scientific development.

Fifthly, the practical advantage of taking city as the unit of analysis roots in its integrity and autonomy. Being factually and legally separate administrative-territorial unit, the city is convenient for collecting and organizing data. At the city level, you can summarize data of a fundamentally different nature, for example construction costs of infrastructure and the number of defended doctoral theses. Thus, it is easy to avoid distortion of data, such as double counting or underestimation of some parameters.

To make judgements about the productive efficiency of the cities that have legally assigned the status of Naukograd, it is needed to compare their performance with other municipal units that do not possess this status. Comparison of the two groups will give an idea about their homogeneity/heterogeneity, and will serve as an actual evidence proving or disproving the unique position of municipal associations, which are recognized as having high scientific and technological potential. That is why the scope of the study is wider than formal set of science cities. It includes 46 Russian cities, of which only 7 are officially science cities (out of the total number 14 of Russia's science cities), and several other cities are candidates to receive this status.

2.3 Methodology state-of-the-art

Most scholar studies on scientifically intense territories on a global scale rise at least one of the three following problematic areas:

1. A synergy effect of the territorial concentration of science and industry, which is the main source of competitive advantage for clusters. Problematics of this topic consists, first, in analyzing stakeholders of R&D on a city level. The aim is to find and crystalize the part of community who is the main driver of synergistic processes. Second way of problematizing this topic is to explore the role of geographical proximity in the mobilization and consolidation of communities and actors in the innovation process.

2. The role of policy and authorities in the development of science in the region. As mentioned earlier, the development of technopolises can follow various patterns, including models of management described in a previous part of this text, and beyond. In practice, the research and development, institutional design and commercialization are closely intertwined with development issues of the municipality, its autonomy and agility of executive office. Tackling these issues implies mostly managerial studies on how to develop of an integrated approach to determine executive structures and processes, areas of responsibility and authority distribution in the management system of science cities.

3. The most practice-oriented direction in the research is complex analysis of experience of the realization of theoretical concepts of urban development. In other words, these are case studies of technology clusters and hubs. They usually put their focus on how harmonization of interests of separate groups of stakeholders is achieved. This includes coordination of stakeholder's vision of the overall picture of cluster development from the strategic point of view, for instance, the choice of strategic positioning: should the city production focus on broad consumption or niche market segments?

In this analytical review, it was decided to focus on the first topic - theoretical studies of synergy in the science cities, because foreign literature sources are more relevant in theoretical discourse than in case studies. The reason is a significant difference between Russian and foreign practices in genesis and governance of science cities. Moreover, practice-oriented work is abundantly performed in the framework of the Russian Cluster Observatory.

The leading role in evolution of scholar discourse on science cities played articles in economics, devoted to the territorial aspect of the industry dissemination. One of the earliest works describing a synergistic effect of technology clusters is Principles of Economics by Alfred Marshall, first published in late XIX century and related to his studies of industrial clusters in the UK. Hundred years later, his concept of “industrial district” formed the understanding of “economic clusters” in works of Michael Porter, one of the main popularizers of this concept in the 90-ies of XX century. The Porter's framework devises influence of a cluster on its surrounding area in three ways: increase of productivity of institutions and companies in the area, thrive of innovation in the field, and stimulation of entrepreneurial activity in the field. His research was empirically based on data about the salary levels, its growth over time, the level of employment and patent information. According to these indicators, he explores the differences between regions, industries and the distribution of knowledge-based territorial units. For example, in The Economic Performance of Regions the following results are obtained:

- There is a correlation between the regions/clusters performance according to selected indicators (salary, employment, patents) and the trade in these regions/clusters.

- All industrial sectors fall into categories a) traded market in the country and the world b) traded on local market C) resource-dependent.

- Technology clusters are mostly categories “traded”, which is about one third of total employment in the United States, and is characterized by significantly higher wages, innovation (patents) and the impact on the average level of development of the region.

- The difference in the level of wages in the regions depends on the performance of clusters and the vitality of innovations born there.

Based on the foregoing, Ellison and Glaeser suggested and still develop, one of the most popular methods of measuring clustering of industry, namely, EG-index. This index measures agglomeration density ratio in order to assess the importance of intersectoral side effects and benefits of cooperation. As an empirical basis, they take data from the U.S. Census Bureau over the previous decade, precisely, the information about the interactions between 122 industries. This data includes conventional economic indexes, such as Herfindahl index of industry monopolization and Gini coefficient of wealth distribution, with which was executed pairwise comparisons of cross-industrial interactions using regression analysis.

This approach gained popularity for wide use of regressions in econometrics and availability of data. Recent example of agglomeration research based on regression analysis, is Nomaler et al., 2014. The authors use bibliometric data from Scopus database to study dependencies between the size of a population of a city and the number of publications per capita in different disciplines. As the result, sublinear dependency has been detected, meaning that the relative performance of very large cities is noticeably lower than expected. Aa expected, small communities make rather small contribution to science, except for some significant spikes, which correspond to science cities, tech hubs or clusters.

Another popular concept which had shaped with the help of Porter's popularization work, is the notion of spillovers, or externalities, which are defined as “non-rival knowledge market costs incurred by a party not agreeing to assume the costs that has a spillover effect of stimulating technological improvements in a neighbor through one's own innovation”. There are several models of evaluating knowledge spillovers, most important among them are MAR (Marshall-Arrow-Romer) externalities, Porter externalities and Jacobs externalities.

The concept of spillovers is closely related to intellectual capital management, namely, the issues of tacit knowledge and know-how transfer. Theoretical research on this topic is usually focused on the importance of cooperation between universities, public sector and industrial centers. Interestingly, recent case studies discuss the phenomena of migration of the traditional scientific knowledge production centers from North-Western Europe and North America to South-East Asia and southern Europe. The empirical basis of this study of externalities is bibliographic and patent data, the approach is co-authorship measurements and citation analysis, aggregated on a city level with the world's hundred largest cities considered.

Network theory is one of the most popular trends of the last 10 years, widely used in many disciplines including studies of the synergetic effects of the territorial concentration of research and development. The idea is simple and holistic: an object of study is a network, modeled as a set of nodes connected by edges. It is applicable at all possible levels: from conceptual decisions on the structure of the national innovation system (a), up to computing precise volume of production and consumption of scientific knowledge within a specific discipline (b).

In the first case (a) we are talking about KIBS-Based Sustainable Urban Development, where the aim is enrichment of sustainable urban environment by applying network analysis to development metrics. The researchers use data on the exchange of knowledge-intensive business services (edges) between city entities (nodes) in China. By the means of network theory, they form a collective understanding of the development of science-based agglomerations in the state. The findings include a significant difference between the cities in terms of their incoming and outgoing connections (in-/out-degree); at the same time, it was founded that the hierarchy rate between the cities is much lower than it was expected initially given the Chinese context; finally, those cities turned to be exposed to some typical network effects (“small world”, “core-periphery”), which gives an additional validation to the approach and results. In conclusion, the authors categorize scientific centers of China according to their network performance and propose to consider obtained features when forming regional policy and strategy for scientific and industrial development of the country.

In the second case (b) the story is about the network analysis of the publication database of the American Physical Society, geographically restricted to the level of urban units and covering a period of fifty years. The study proposes a metric called Knowledge Diffusion Proxy, which is based on the capacity (in-degree) of incoming citations (edges) to different cities (nodes). In addition, this study contains the ranking algorithms for the results of scientific production applied to evaluate spatial and temporal dynamics in particular discipline (network of “knowledge of physics”) in the world. The proposed metric allows to detect cities that play the most significant role in the production and consumption of knowledge of physics (for the time). The potential advantage of this approach is that it reduces all disciplines to a common spatio-temporal denominator, which opens opportunities for advances in cross-disciplinary comparative studies.

Based on the conducted literature review, it is reasonable to summarize the most common types of data used for the analysis of the considered type. These are:

· bibliographic and patent data, including those weighted to the population of the city,

· investments: venture capital / public / investment attractiveness / ROI,

· agglomeration density coefficient (with variations),

· human resources: employment, quality of jobs and wages, including its dynamics.

3. Empirical part: efficiency study

3.1 Objectives and approach

The empirical part of this analysis of science cities as units of organization of science is limited to efficiency assessment. The statistical base for analysis had been gathered on the main indicators of science and technology as well as some indicators of urban development of Russian cities. The received data includes the expenses and performance for research and development, human resources, intellectual property, demographics and construction. All information is collected across 46 cities, 7 of which have the legal status of science city.

Background literature review revealed problematic areas and approaches to studies of development of cities with high scientific and technological potential. According to the results of the theoretical research, there is a necessity to improve and develop approaches to the analysis of efficiency, performance and evaluation of the development prospects of science cities. To contribute to the field, it is relevant to conduct empirical research on the current situation in Russian science cities. Narrowing down the scope, it had been decided to stick to the productivity measurements for the reasons described when formulating research problem. As a reminder, we are starting off two hypotheses:

- science cities are more productive than other cities, that don't have such legal status,

- the sources of inefficiency aren't homogeneous among cities.

The goal of the empirical study is to test these two hypotheses. To do so, it is necessary to proceed through the following steps:

1. preprocess statistical data by systematization and aggregation of available information at the city level, and selection of the most relevant parameters;

2. customize Data Envelopment Analysis instruments to obtain an appropriate model;

3. apply chosen model to preprocessed data;

4. interpret and analyze the results with respect to the context uncovered in the theoretical part of the thesis.

3.2 Data sources, collection and preprocessing

In this section, we are going to establish why and how our database is relevant to validation of the hypotheses. Multiple parameters could contribute to the stated goal, and one of the important points when selecting options is the prospects of active influence on them. In other words, it is preferable to consider those parameters, which values could be governed. Compliance to this criterion allows us to formulate recommendations for specific management steps and policy actions following the study. In this case, the parameters are set based on features of the town/city/municipality as a unit of organization of science. Considering the integrated nature of urban environment and R&D, we take into account two basic groups of parameters: related to research and technological development, and related to the urban development.

The collection of statistics of the first group is governed by regulations at multiple levels, among which the standard guidance is Frascati Manual, originally created by OECD experts to coordinate and facilitate the process of collecting and processing data on research and development at the national level. The guide covers major conceptual and methodological issues, contains detailed categorization of R&D and takes into account specifics of carrying out and funding various types of studies. According to the Manual, the main landmarks for the measurement of scientific and technological progress are innovative activity, introduction of novel technologies, patent statistics and bibliometrics.

Despite significant systematization work done to evaluate the performance of research and development, this issue remains problematic both from theoretical and practical point of view. There is a discrepancy between the input data, based on interviews and locally collected statistics, and outputs that appear in form of universal indices and international rankings. Thus, in absence of a clear understanding of parameter's distribution and weighting, it is impossible to avoid inconsistencies. For instance, Russia took the 14th place in the rating of innovative development "The Global Innovation Index" according to Bloomberg [46]. At the same time, it has taken 85th place in "Global Competitive Index" by to the World Economic Forum according to their criterion of "power of innovation" [47]. Such situation was explained in the comments to the Bloomberg report by negotiating that the high position of Russia reflects the great innovation potential, measured by the number of people who have received higher education and scientific degrees, while the level of actual achievements in innovation is quite low. This example demonstrates that the choice of parameters and distribution of their weights in ranking models remains arbitrary even in practices of the most influential analytical agencies. Often it is led by the possibilities of data collection, that is, depends on what statistical information is available for use.

This empirical study is based on three types of information sources, namely:

1. The results of non-recurring solid statistical survey (inventory) of organizations carrying out activities in science and technology;

2. Web of Science database;

3. Official statistics collected by the Federal Statistics Service (Rosstat) on a regular basis.

The first resource consists of database on more than one hundred performance indicators of R&D organizations (except small businesses) performing research and development. The information was reported by the time of 1st January 2012, collected in the framework of the project "Formation of a system for monitoring the economy of science for assessing science and technology and demonstrating new scientific achievements", conducted by ISSEK HSE in 2011-2013, commissioned by the Ministry of education and science of Russia.

The project consisted of 6 steps: firstly, authors have developed a system of indicators monitor the economy of science. Secondly, they drafted a project of the Unified Registry List of organizations of Russian R&D sector, carried out a pilot study of research and innovation activities of universities, the forms of integration of science, education and real sector of the economy. Thirdly, prepared and conducted a pilot version of the survey prepared an inventory of organizations in the sector of R&D and the survey of small enterprises in scientific and technical sphere involved in transfer and/or use of scientific results and technologies, and formed a system of monitoring the implementation of priority directions of development of science and technology in Russia. Fourthly, they made specialized surveys, including one-time solid statistical examination, which has become a source of information about organizations in the field of research and development used in this work.

Among other results was the formation of the registry of organizations in research and development sector of Russia according to the results of the thoroughly prepared and executed statistical inventory. It has been further used for multiple purposes, such as modeling of innovation activity, policy recommendations, presentations and popularization of the results of monitoring of Economics of science, and other forms of decision-making support for governance and management of R&D sector.

Overall, the above project is based on well elaborated theoretical and empirical studies, in particular, concerning criteria for the selection of parameters. For this reason, further use of the project results in the present study causes no doubt. The result is a database, aggregated at the level of legal entities, that contains information about organizations working in the field of research and development. More specifically, information was provided on:

1.1) The number of employees engaged in research and development,

1.2) Expenditure on research and development,

2) the structure of the organization,

3) the presence and composition of fixed assets of the organization,

4) the characteristics of buildings and premises,

5) the age structure and technical level of machinery and equipment,

6) the presence of unique stands and facilities for carrying out research, developmental and technology intensive work,

7) the use of information and communication technologies,

8) experimental facilities of scientific organizations,

9) the intensity of scientific and technological activities, including,

9.1. the results of scientific research and development,

9.2. the legal support and use of intellectual property,

9.3. technology commercialization,

10) the innovative capacity of the organization,

11) the distribution of researchers by age,

12) the preparation of highly qualified scientific personnel, including,

12.1. the total number and dynamics,

12.2. number of thesis and post-graduate researches.

In the original survey, the inventory unit was a legal entity, therefore for the purposes of this research each organization was assigned to city/municipal entity according to the territory where it is located. Each indicator was aggregated at the level of a) absolute value b) weighted value per organizational unit. The final table in Appendix contains a sample of the 46 lines of the cities, and columns -- their characteristics listed above, including their detailed parts summing up to 100 indicators.

The second resource for the compilation of empirical basis for the current study was the Web of Science database, which served as a source of information on publication activity of Russian researchers. The source is an online citation indexing service for scientific publications. It belongs to a private media conglomerate Thomson Reuters and has a system for granting access by subscription. Historically, the base originated from the Garfield's Science Citation Index (SCI), designed to help researchers navigate the extensive and rapidly growing environment of academic texts. This index implies the idea that the value of information is determined by the impact it has on society and on the professional community within a discipline. In other words, how and by whom the text is used, that is, how often and where a paper was cited.

Currently, Web of Science offers a wide range of analytical tools in four primary areas: interdisciplinary resources, specialized resources, analytical resources, managing bibliography. Inside each direction there are specific indexes, database, key words and network of quotation, tailored to the specifics of the queries to provide the user with the most influential, relevant and reliable information. According to the company's website, the main elements of their value chain are the following: primary interdisciplinary content (publications on fresh topics not formed in a separate scientific school), emerging trends in publications, thematically separate content, separate regional content, research data, analytical tools.

To meet such diverse analytical tasks, meta-data for each publication includes the indicators of different nature and scale of aggregation: the journal title and article author name, organization that supports the research and its address, abstract, individual identifier, bibliography. Through availability of the organization's address it becomes possible to regroup information at the city level for the unified unit of analysis of this study.

The data on publication activity was collected with the help of analytical tools of Web of Science. The main parameters extracted from the database were the number of published articles and monographs, including those indexed in the WoS, including in co-authorship with foreign scientists. In world practice, standards and procedures for the collection of statistics vary depending on the place and institution, level of generalization, the purpose of accounting. The methodology for bibliometric data gathering varies in:

1) databases that are accounted,

2) types of publications,

3) time frame for data on citations, and the need consider:

3.1) the time lag between publication and citation;

3.2) the practice of citation adopted in a specific scientific discipline.

In this study, three decisions were rendered the following way:

1) Database: Web of Science (WoS) -- the main alternative to this database is Scopus (Elsevier). The difference between them lies in the way they structure information, and in coverage of primary sources by discipline. In brief, coverage and structure of the WoS data is more consistent with the purposes of this research.

2) Types of publications: scientific articles, reviews and conference reports -- this sample requires more detailed explanation. Each published in scientific journal work is assigned attributes such as: ID, key words, discipline (not in all cases) and type of publication. There are three ways to categorize scientific publications: publication as a tool for scientific communication, publication as a database document, publication as a unit of statistical accounting. Depending on the selected categorization, all texts fall into different groups, and many journals form these groups individually, therefore historically established categories are often inconsistent. However, for the purposes of collecting statistics in terms of science it is necessary to consider the main types of documents. In absence of a common sense of what "main types" stands for, each time this decision is made by the researcher, usually based on conventional criterion of "opportunity to be quoted" with the aim of subsequent citation analysis.

3) Citation time frame: this study lacks parameters directly related to the volume of citation base. As noted above, the time lag between publication and citation data can vary. In addition, different scientific disciplines differ with accepted citation practices. This study does not focus on a particular discipline, and strives to get the overall picture of the of research and development progress. Citation data would be a fairly accurate measure of the intellectual influence of the city, however, processing a complete network of citations requires unreasonable amount of time and computing power.