Introduction

The entry into force of the Russian Budget Code (Byudzhetnyy kodeks) on 1 January 2000 marked a break from the fiscally tumultuous 1990s.[1] Under the reformed budget system, for two decades until 2022, data was made available online by the financial authorities in accordance with the requirements of the law. Principles concerning transparency and openness were firmly established and the budget system was positively assessed by authoritative observers as advanced in an international comparison (Hughes et al. 2014; Transparency International UK 2011; IBP 2017). At the same time, a significant share of the federal expenditures remained subject to secrecy, a fact that has led to some uncertainty about Russia’s actual resource allocations, in particular related to defence expenditures. In a previous research article, we documented that this issue had received relatively little attention in the literature, and used residual analysis to provide precise quantitative measures of secret expenditures in different chapters and subchapters of the Russian budget; we also estimated lower and upper bounds of Russia’s defence expenditures based on a close reading of the various budget categories used in the Russian budget framework (Andermo and Kragh 2020).[2] While the residual-analysis approach adopted in our previous research makes a useful contribution to the problem of estimating the size and long-term trends of the Russian defence budget, it does not exploit the full potential for analysing the relatively rich data sets made publicly available by Russian authorities.

The purpose of the present discussion is to take the analysis a step further by focusing on the qualitative nature of Russia’s expenditures, including secret expenditures. A key contribution is that we show how budget data (in particular, budget classification codes) can be cross-referenced with budget classification documents from the Finance Ministry to sketch the contours of the secret parts of the budget, for example the general structure of the budget for intelligence and security service activities (Table 8 in the annex), and the of the defence industrial complex (Table 10). The method we develop involves constructing a database that also makes it possible to single out Russian budget expenditures in several different dimensions, including expenditures in any given state programme and how they are allocated between government departments, budget chapters and other categories. The amount of data involved in the analysis requires computerized methods to achieve the desired results, as discussed further below. Although our method is quantitative in nature, our focus here is on the qualitative understanding of the results, in the hope that other researchers might find the method useful for their own purposes. The main empirical contribution is captured by a series of tables documenting Russia’s defence expenditures in 2014-2018 with an unprecedented level of detail (Table 5, Table 6 and Table 7). Finally, the paper discusses the relationship between the federal budget and the State Armament Programme (GPV, Gosudarstvennaya Programma Vooruzheniya); this topic that has received relatively little attention in the literature, with the notable exception of the contributions by Julian Cooper (2013 and 2016).

The enhanced qualitative understanding of the Russian budget system suggested by this paper opens the possibility for a new type of analysis of a large part of Russian state activities since approximately half of the expenditures in the federal budget (possibly as much as 70 percent if secret programmes are included) are associated with one of over 40 state programmes and their corresponding 500 or so subprogrammes in practically all fields of government affairs. State programmes seem to be intricately related to Russia’s institutional structure and can be viewed as a governance institution with a historical tradition in Russia, but to our knowledge, the role of state programmes in contemporary Russia has not been studied in the literature. We show that the State Armament Programme itself is actually not a budget category, but largely operationalized in the budget through the annual State Defence Order (GOZ, Gosudarstvennyy Oboronnyy Zakaz), and that the orders related to the State Armament Programme are only a subset of the annual State Defence Order. We use a methodological argument and new data presented in this text to claim that this observation is correct even for the secret part of the budget which is not directly observable. This is possible because even though most budget allocations in the defence sector are kept secret in the annual budget laws, the generic budget categories describing those allocations are – somewhat surprisingly – published publicly by the Ministry of Finance in the form of regulatory instructions containing detailed descriptions of the principles underpinning the budget classification codes.

The main contribution of our approach is methodological. Although a manual analysis can identify specific items in the Russian state budget, our data-driven method makes it possible to analyse large numbers of data over time and across all budget categories, allowing a greater level of empirical detail than what would otherwise be possible. In this sense, the analysis goes significantly further than the standard reference works on Russia’s defence expenditures from the Stockholm International Peace Research Institute and the International Institute for Strategic Studies (for example SIPRI 2019 and IISS 2019).

The secrecy surrounding Russian defence and security related expenditures requires certain roundabout approaches to circumvent. Although Russia is (even after 2022) operating a budget system that is much more transparent than that of its predecessor the Soviet budget (see Hutchings 1987 and Kontorovich 2009), expenditures classified as secret increased between 2011 and 2016 from around 10 to 20 percent of total federal expenditures, and from 45 to 70 percent of expenditures in budget chapter 0200 National Defence (Andermo and Kragh 2020). Our method uses regulatory documentation that also makes it possible to make certain inferences about the nature of those expenditures that are indeed secret.

The discussion is organized into five parts. The first part introduces the role of state programmes in the Russian budget in a historical perspective to set the stage for the subsequent discussion. The second part draws on previous research to give a brief introduction to the Russian budget system and the sources used to obtain the necessary data. The third part presents the method used to construct the dataset that we need for the empirical analysis, including a brief discussion of the parsing algorithms that we used to extract and structure the data. The fourth part discusses the empirical results related to some defence-related state programmes. The empirical part also leverages the information in the regulatory documentation to make inferences about the nature of secret expenditures. In the last part, we provide a summary of the results. Many of the results are presented in table form; most of these tables have been placed in an annex.

Government programmes and the Russian budget[3]

Various types of programmes have evolved as instruments of governance somewhere between the executive and legislative branches of government since the early days of the Russian Federation: the programmes are typically formulated by the executive but depend on funding by parliament. The first type in modern Russia was the federal target programme (federal’naya tselevaya programma), which was codified in relatively vague terms in a 1994 law; in accordance with this law, federal target programmes are drafted by federal executive entities (including ministries) following procedural guidelines issued by the government.[4] In the budget for 2018, there were 24 federal target programmes, seven of which were related to defence and security. There is a special department in the Ministry for Economic Development in charge of overseeing the federal target programmes which used to operate a web portal dedicated to monitoring these programmes; according to data from this portal, the number of federal target programmes peaked at 57 in 2011.[5]

Media reports frequently suggest that the government fails to reach the stipulated targets in several of these programmes (e.g., Panov 2017). In 2018, Prime Minister Dmitry Medvedev warned that officials might be held responsible for failing to fulfil the programme targets (Kuzmin 2018). In 2010, a new type of so-called state programmes (gosudarstvennye programmy) was introduced by a government decree during Putin’s term as Prime Minister; this decree is amended whenever programmes are added or removed.[6] In 2012, shortly after his return to the presidency, Putin gave a policy speech on budget priorities in which the transition to a budget based on state programmes was one of ten fiscal priorities aimed at a “fundamental increase in the quality of the strategic management of the economy and public finances” (Putin 2012). The timing of this initiative is noteworthy since it came after Putin himself had served as Prime Minister for a four-year period, a role in which the day-to-day mechanics of budgeting are more pronounced than in the presidency. In 2013, the Budget Code was amended to accommodate provisions for state programmes; this was done through an amendment to article 179, which had previously regulated so-called long-term target programmes (dolgosrochnyye tselevyye programmy). The state programmes touch practically all fields of Russian government activity, and during the years covered by this study, 2014−2018, over 40 non-secret state programmes appeared in the federal budget, accounting for about 50 percent of government expenditures in each fiscal year, based on our calculations. If secret expenditures are included, this share rises to 70 percent according to one official source.[7] This focus on a programme-oriented budget process is not unique to Russia but reflects a global trend to introduce performance-based budgeting (Diamond 2003). In his speech, Putin announced that the transition to programme-based budgeting should be completed starting from the fiscal year 2014, which serves to motivate the period chosen for this study; between 2014 and 2018, the structure of the budget classification codes which form the basis of our analysis remained largely unchanged (although he budget code was extended from 17 to 20 characters in 2016, as discussed briefly below).

The most well-known type of Russian state programme is arguably the State Armament Programme (GPV, Gosudarstvennaya programma vooruzheniya). Although this programme gained wider attention in 2010 with the adoption of an ambitious GPV for the period 2011−2020, there were two preceding GPVs adopted in 1996 and 2002, respectively. In the late Soviet era, state planners introduced ten-year armament programmes (programmy vooruzheniya), and it is easy to assume that the GPV would be a legacy from the Soviet planning system, but the concept of armament programmes traces its roots all the way back to Peter the Great. Two noteworthy historical programmes were the 20-year programme adopted in 1882 for the development of a modern navy, and the so-called Great rearmament programme (Bol’shaya programma perevooruzheniya) in the run-up to the first world war (CAST 2015).

The legal basis for the modern GPV is found in the federal law “On defence” from 1996, which states that the GPV is drafted by the government and approved by the president.[8] Before the concept of the GPV was introduced, there was already a law regulating the State Defence Order (GOZ, Gosudarstvennyy Oboronnyy Zakaz) which stipulated that the GOZ should be updated annually and based on, inter alia, a “federal programme for the development, creation and production of weapons and military equipment for a ten-year period”.[9] This division between an annual GOZ and a long-term GPV seems to mirror the routine that was put in place by Soviet planners in the 1980s (CAST 2015). A new law regulating the GOZ was adopted in 2012 where it was stipulated that the GOZ should be drafted based on both the GPV and defence-related federal target programmes, as well as other documents.[10] Although the GPV has been around since the mid-1990s – this is an important methodological point – there is no state programme budget code associated with the GPV in publicly available budget data or in the Finance Ministry’s budget instructions. This contrasts with the state programmes of the 2010 government decree and Putin’s 2012 speech, which indeed have their individual budget codes, as will be discussed further below. This does not rule out the possibility of a separate budget category for the GPV in the secret part of the budget, but an alternative and, arguably, more plausible interpretation is that the GPV does not constitute a budget category at all. This interpretation is supported by its logical relationship to the GOZ defined in the Budget Code: firstly, the main parameters of the GOZ are an integral (but secret) part of the annual budget law (BC article 205), and, secondly, the purpose of the GOZ is to implement the GPV (article 217).[11] Thus, the GOZ serves as a mechanism that links the budget to the GPV and, therefore, there seems to be no need for any special budget category for the GPV itself.  Furthermore, there are examples of budget items related to the GPV that are indeed listed in the Finance Ministry’s instructions even though their associated expenditures are secret in the annual budget laws; one example is the main activity 31 1 07 00000 “Implementation of measures of the State Armament Programme”, which is part of subprogramme 31 1 00 00000 “Construction and development of the Armed Forces of the Russian Federation,” which in turn is part of state programme 31 0 00 00000 “Ensuring the country's defence” (further explored in the empirical section; see also Table 9). This observation suggests that even if the GPV were a secret budget category, it should nevertheless be listed in the instructions, but it is not. The example also shows that for budget purposes, the GPV is interlinked with other state programmes, in this case “Ensuring the country’s defence.” Furthermore, the fact that the GPV is approved by the president, while the state programmes are approved by the government, also indicates a conceptual difference between these types of programmes, in addition to the fact that the GPV predates the other budget-related state programmes by around 15 years. Finally, there are several budget codes in the type code category (further discussed below) that are explicitly associated with the GPV: for example, the code 210 (the “subgroup” category as discussed further below) represents “Development, purchase and repair of weapons, military and special equipment, production equipment and property within the framework of the State Defence Order,” and is further subdivided into several subcategories (element codes) in the range 211−219; some of these codes (211, 214, 216 and 217) are explicitly related to the GPV, and the rest explicitly not related to the GPV; most of these items are secret, but not all of them.[12] Taken together, these observations support the interpretation that, despite its name, the GPV does not have any associated state programme budget code (secret or not) and is not a separate budget category. This discussion also helps to highlight the meaning of secrecy in the context of the Russian budget: while the existence of these budget categories related to the GPV is evidently not secret, secrecy means that we cannot know how they are interlinked, or how actual spending is allocated among them.

Sources and data

All data used for this study come from openly available (at least until 2022), public reports from Russia’s financial authorities in the form of budget execution reports of the Federal Treasury (Federal’noye kaznacheystvo), also known by its acronym Roskazna, and budget categorization instructions from the Finance Ministry (Ministerstvo finansov).[13] The budget execution reports of Roskazna, tracking the implementation of the Russian federal budget, contain four tables for expenditures, designated 2.1, 2.2, 2.3 and 2.4. Each of these four tables corresponds to a specific provision in the Budget Code on how budget expenditures should be presented. Roskazna’s Table 2.1 includes a detailed list of items grouped according to all five budget categories (further explained below) and corresponds to the so-called departmental structure of expenditures (vedomstvennaya struktura raskhodov). Its name derives from the fact that it is the only one of the four tables that – importantly – provides information about the agency affiliation of each budget item; for example, it shows which items of the defence budget that are administered by the Ministry of Defence. In this text we will refer to the agency affiliated with a specific budget item as ‘budgetary manager’, translated from the Russian legal term ‘main manager of budgetary means’ (glavnyy rasporyaditel’ byudzhetnykh sredstv). Certain expenditures in three of these five categories are secret, but not in the other two. Some of the technical terms describing different budget categories were introduced above in the discussion about state programmes. The non-secret categories are ‘chapter’ and ‘subchapter’ (razdel and podrazdel) – the focus of previous literature – while the categories partially subject to secrecy are ‘budgetary manager’, ‘expenditure purpose’ (tselevaya stat’ya raskhodov) and ‘expenditure type’ (vid raskhodov). There are also several subcategories, most importantly state programmes that can be identified by the purpose category. Understanding how these categories relate to each other is the core contribution of the present analysis, as the previous literature has mostly focused on the non-secret categories (chapter and sub-chapter). Roskazna’s Tables 2.2 and 2.3 present the same data as Table 2.1 but exclude the category budgetary manager; thus, they represent subsets of Table 2.1 and can therefore be ignored for our purposes. Table 2.4 presents aggregate data representing only the chapter and subchapter components of expenditures; in this aggregate form, there are no secret expenditures: the sum of expenditures in table 2.4 is equal to the total amount of expenditures approved in the budget, unlike tables 2.1, 2.2 and 2.3, which exclude secret expenditures. Importantly, the size of secret items in the budget can be calculated as the residual between Tables 2.1. and 2.4 for each chapter and subchapter, as explained further in Andermo and Kragh (2020). Furthermore, Tables 2.1, 2.2. and 2.3 also correspond to different appendices to the actual budget law for each year, while Table 2.4 corresponds to an appendix to the explanatory note (poyasnitel’naya zapiska) that always accompanies the budget law.[14] These basic provisions are defined in the Budget Code. A simple but important methodological point is that the secret residual is explicitly acknowledged as such in the Budget Code.

Each budget item in the departmental structure of expenditures is uniquely identified by an alphanumeric code consisting of 20 characters to represent each of the five budget categories and their respective subcategories.[15] The structure of this code is presented in Table 1a and Table 1b below. The following discussion about this code system is necessarily rather technical but is a prerequisite for understanding and interpreting our results. The first three characters of the code represent the budgetary manager and the next four are the chapter and subchapter. The purpose code consists of ten characters in positions 8−17. The purpose code is subdivided into two parts with five characters in each: the first five represent the ‘programme’ code (programmnaya stat’ya, positions 8−12) and the last five represent the ‘direction’ code (napravleniye raskhodov, positions 13−17). The programme and direction codes are always intertwined in the sense that certain direction codes are reserved for certain programmes. Furthermore, the programme code also contains information about the categories ‘subprogramme’ (podprogramma, position 10) and ‘main activity’ (osnovnoye meropriyatiye, positions 11−12). Although the name implies that this code is used to identify programmes, there are also non-programme (neprogrammnyye) codes in this category to identify the one-third or so of total budget expenditures that does not belong to any programme. Finally, the last three characters represent the type code, which is further divided into the subcategories ‘group’ (gruppa, position 18), ‘subgroup’ (podgruppa, position 19) and ‘element’ (element, position 20). We also need to note that before 2016 the budget classification code consisted of only 17 characters rather than 20, because the subcategory main activity was not yet introduced, and the direction category had only four instead of five characters.

Table 1a. Structure of the Russian budget expenditure classification code

Source: MinFin (2018a), Part III, section 1, Table 2.

Table 1b. Structure of the purpose code from Table 1a

Source: As Table 1a and MinFin (2018a) Part III, section 4(1).2.1.

Table 2 (see the annex) shows an example of actual budget execution data from the departmental structure of expenditures (corresponding to Table 2.1 in Roskazna’s report), with sample data for 2017 in the form of the first eleven rows of that report. Columns 2−5 include the components of the budget classification code in each of the five categories (column 3 denotes both chapter and subchapter). The first three rows of Table 2 identify a budgetary manager, a chapter and a subchapter in columns 2 and 3. Columns 4 and 5 are empty on these rows, i.e., the budget code is truncated. It is only rows 7−11 that have complete codes with all 20 characters represented, and row 7 contains the sum of the items on rows 8−11. In fact, row 7 corresponds to the budget item such as it was approved in the federal budget for 2017; the budget law only approves expenditures down to the group level (row 7), while the subgroup and element levels (rows 8–11) are determined at the time of execution by the responsible budgetary manager. That row 7 corresponds to the group level is immediately visible from the two trailing zeroes in the type code (600 in this example). Rows 8−11 correspond to the element level (type codes 611 and 612) and also contain implicit information about the subgroup level, which is not given explicitly in Roskazna’s budget execution reports; from the discussion above it can be inferred in this example as type code 610.

Furthermore, the complete budget codes on rows 7−11 in Table 2 contain all the codes from rows 1−6, which has two important implications: Firstly, it implies that the aggregate data on rows 1−6 can be derived by taking the sum of more detailed items in each category. Secondly, the full meaning of the complete budget code at the most detailed level can be derived by consulting the descriptions of the corresponding item at the aggregate rows. For example, suppose that we wanted to obtain a complete description of the last budget item in Table 2, on row 11. As can be seen in the table, the complete code of this item is 007 0110 0130293987 612. By consulting the other rows of the table, we can conclude that 007 represents the Federal Agency for Scientific Organizations (row 1), 0110 represents General government (row 2) and fundamental research (row 3), and so forth. With only a few rows of data this can be achieved by simply inspecting the table. However, the full version of Roskazna’s Table 2.1, from which our Table 2 is adapted, contains over 13,500 rows, so even if it would be possible in theory to derive a complete description for every budget item by manually inspecting the table, in practice it is not a realistic approach. Furthermore, even if we could complete that task, we would still not solve the problem of calculating the sum in each individual subcategory: for example, on row 4 of Table 2, we notice the state programme “Healthcare Development” (Razvitiye zdravookhraneniya), but, ideally, we would like to obtain a summary of this state programme across all budgetary managers and chapters, not only the ones given in Table 2. In short, to extract all this information, we need to transform the data in the departmental structure of expenditures into a different structure that is more suitable for the analysis.

A commonly used technique for summarizing datasets by calculating sums, averages, standard deviations and other statistical measures across many different categories is the pivot table. A pivot table makes all those calculations feasible but requires that the underlying data adhere to a specific type of pattern. In practice, this means that all categories describing an item must be explicitly indicated as a separate column on the same row as the item itself. In addition to the five budget categories, we also need a column to indicate the fiscal period. Transforming a dataset so that a pivot table can be applied to it is sometimes referred to as normalizing the data.[16] To illustrate the concept, Table 3 in the annex represents a normalized dataset transformed from Table 2. When this transformation is applied to the original data for 2017 of Roskazna’s Table 2.1, the number of rows is reduced from approximately 13,500 to 5,500. Note that additional columns have been added in Table 3 to also cover the subcategories of the purpose and type categories.

A few differences are worth pointing out between Table 2 and Table 3. First, no rouble data is duplicated in column 10 of Table 3, which contrasts with column 6 of Table 2, where several rows contain sums calculated from other rows. Secondly, unlike Table 2, there are no empty cells in Table 3. Thirdly, Table 3 duplicates category data in columns 1–6 and 8 on multiple rows. Finally, Table 3 does not present aggregate data for budgetary manager, chapter, subchapter, state programme and so forth; instead, these aggregate sums can be readily calculated in the pivot table from the data in Table 3. Informally, these differences essentially capture the necessary requirements for a data structure suitable for a pivot table. We can now apply a pivot table to the transformed dataset and thereby extract summarized information for any category.

Computational method

The original data exists in the form presented in Table 2, where each (non-secret) budget item carries its own, unique budget classification code. To extract all the descriptive information contained in this code and transform it to the form shown in Table 3, we need a complete mapping that relates each budget code to the description defined for that code. In other words, we need complete lists of all budget classification codes and their corresponding descriptions. With such a reference list, we can cross-reference every code in the original budget data to that list and thereby construct a complete, normalized dataset for the entire budget, except the secret residual. This can be done in at least two ways, both of which depend on computer power to automate: the most straightforward would be the inspection method as described above when we manually transformed data from the form in Table 2 to that in Table 3. One drawback of the inspection method is that it does not capture the subgroup category, which is only given implicitly by the element code, as discussed above. An alternative way, which we can call the reference method, is to look for the definitions of the budget codes from another source. This gives us the possibility of quality checking the results by combining the two methods. Moreover, as will be clear from the discussion below, the reference method gives us additional information, because it turns out that such a publicly available source does indeed exist, and – importantly – that it also defines the budget codes of the secret budget items which are absent from the Roskazna tables. This in turn provides us with a powerful tool to make inferences about the secret parts of the budget and to relate anecdotal information, such as media reports about the GPV, to the federal budget in a systematic way. It also allows us to say something about the nature of the secret expenditures. This is another of the key contributions of the present analysis, which is based on the reference method.

The alternative data source defining the budget codes which is necessary for the reference method comes in the form of regulatory instructions from the Finance Ministry which contain definitions of all budget classification codes and can therefore be used to construct the necessary reference list. However, although this is straightforward in theory, there are several practical challenges to consider. To start with, the instructions themselves – while readily available through various online sources – run to several thousand pages. The codes are not listed together in one place of the instructions, but in several different sections and appendices. In some cases, the codes are listed along with their corresponding description on one row each, but in other cases the code definitions are interspersed with other text that constitutes noise for our purposes. Table 4 on the next page gives an overview of where the definitions of the codes in each category can be found, together with references to the relevant provisions in the Budget Code and the Finance Ministry’s instructions. The reader might benefit from taking the time to compare Table 4 with Table 1a and Table 1b (page 13) to appreciate how the normative provisions are interrelated.

Finally, there is a third source of data that can be used to corroborate the results obtained by analysing the data from Roskazna and the Finance Ministry, namely the budget legislation itself as published by the duma (ex-ante budget approval), and the annual laws giving ex-post parliamentary approval to the budget fulfilment. However, since legislative data showing actual spending is only available with a delay compared to Roskazna’s operational reports, we do not focus on this data source in the present text (although it could be used to quality check the results).

Table 4. Summary of the regulatory framework defining the Russian budget expenditure classification codes.

Source: MinFin (2018a). Note: Compare with Table 1a and Table 1b. In 2016 the budget classification code was extended from 17 to 20 characters. The subcategory Main activity was introduced, and the Direction category was extended from four to five characters. Due to the extension of the budget code, all versions of the instruction between 8 June 2015 and 29 December 2016 contain duplicate sections covering both the old and the new structures of the code. As can be seen in the last column of the table, for the purpose codes this means that the data must be obtained from two different sections in the document depending on the year.

Having located the data, the question then becomes how to extract it so that we can compile the necessary reference lists. For this purpose, we created an algorithm that trawls through source files containing the Finance Ministry’s instructions and extracts the codes for the different categories. A significant part of the job of the algorithm is to identify and discard information that constitutes noise for our purposes. The format of the data that we are looking for can vary to some degree in the source files. For example, in some cases the definition is given on a single row, but in other cases the definition contains line breaks. Identifying the correct pattern and treating the source data accordingly is another significant part of what the algorithm does.

The source files were obtained from the Russian website Konsultant (https://www.consultant.ru), which contains Russian legislative documents, including instructions issued by ministries. Their service Konsultant Plyus includes a feature that makes it possible to download the content of the regulatory document as text files, and to compare different versions of the same document. The possibility to compare different versions of the text is useful since the instructions are frequently amended, including the definitions of individual budget classification codes.[18] We have limited the analysis to the period 2014−2018, primarily because 2014 is the first year fully represented by the newly introduced state programmes as discussed above, following Putin’s announcement in 2012 that the entire budget should be programme-based, and because a new structure for the classification of budget expenditures was adopted for the fiscal year 2019; all budgets in that period are regulated by the same instructions issued in 2013 (MinFin 2018a, with amendments through 2018; see also the note to Table 4). It would be straightforward to extend the analysis to cover earlier and later years as well, but that would require amendments to the algorithm. An advantage of the inspection method compared to the reference method is that the necessary amendment to the algorithm would be easier to implement since it relies only on the Roskazna tables, not the instructions from the Ministry of Finance.

To capture the frequent changes to the code definitions, the relevant section for each budget category of each version of the Finance Ministry’s instructions (one version for every date when amendments were introduced), as indicated in Table 4, was saved in a separate text file. This generated 329 text files containing a little over 250,000 lines of text in total, corresponding to approximately 70 megabytes of data. In most cases, a code definition remains unchanged between two versions of the instructions (i.e. only a small share of the codes is amended in each revision). We therefore designed the algorithm to keep track of all amendments by collecting metadata describing the type and date of the amendment, including information about when a particular code was discontinued. By running our algorithm on all 329 text files we obtained a reference list of 6,828 unique definitions of budget classification codes, including amendments, and metadata describing the source and the date of validity for each item.

The next step is to perform the cross-reference between the budget data in Table 2.1 of Roskazna’s budget execution reports and the code descriptions in the reference list. For this purpose, we wrote a separate algorithm which takes budget data for one year at a time and loops through the budget classification code of each budget item to identify the corresponding description in the reference list. Since multiple versions of a code can exist in the reference list if that code has been amended, the algorithm also checks the date of validity based on the metadata collected by the first algorithm. Roskazna’s annual budget execution reports are typically dated 1 January of the year following the budget year, so this date was used as the cut-off date in the algorithm. Thus, for each budget year we obtain a complete, normalized dataset with the structure illustrated by Table 3.[19] Finally, we collected data for all years in one table containing our entire dataset in a Microsoft Access database. This dataset contains approximately 25,000 records describing all budget expenditures for the years 2014−2018 with a total of over half a million data points. In the very final step, we manually added records to represent the secret residual for each year in each subchapter (by comparing data from Roskazna’s Tables 2.1 and 2.4 as explained above) so that the sum of all items in the dataset corresponds to the sum of total government expenditure in each year (known from Roskazna’s Table 2.4). We then created a pivot table in a separate Excel file linked to this dataset that was used to inspect the data and obtain the empirical results in the next section.

Empirical Results

The sheer amount of data involved in the analysis obliges us to limit our scope for the purpose of this article to certain areas of particular interest in order to present the results in a suitable form. We have chosen to focus on defence expenditures. These results are noteworthy in themselves, but also serve to illustrate how the method could be employed to obtain data about other fields of government activity than defence. In addition, we also discuss the nature of those expenditures which are secret, in areas such as counterintelligence and the defence industry. As mentioned above in the discussion about state programmes, around 50 percent of total, non-secret expenditures are linked to a state programme, while the share rises to 70 percent if secret expenditures are included. We can therefore conclude that the difference of 20 percentage points represents secret budget items, and, referring to the discussion below about the nature of secret budget items, we can hypothesize with some confidence that these secret expenditures are allocated among three specific state programmes: defence capability, state security, and the military-industrial complex. Based on our data for the non-secret state programmes, the size of the average state programme corresponds to 1.2 percent and the median to 0.5 percent of total federal expenditures. The largest non-secret state programme is “Social support for citizens” (Sotsial’naya podderzhka grazhdan) at around eight percent of total expenditures.

Defence expenditures

The domestic Russian debate about defence expenditures focuses on the budget chapter 0200 National Defence and its corresponding subchapters. This is arguably the most transparent measure of Russia’s defence spending and can be used as a benchmark to compare our results. On the other hand, this measure excludes some military expenditures that are included in the standard measures produced by SIPRI and IISS, most notably the National Guard troops, the border troops and pensions to military personnel. Table 5 (see annex) shows the allocation of expenditures among the subchapters of chapter National Defence, as well as expenditures in subchapters of other chapters that are included in these standard measures. In Andermo and Kragh (2020), we proposed 11 different levels of defence spending using increasingly broad definitions of which budget items to include with the purpose of establishing lower and upper bounds of actual defence spending; Table 5 corresponds to level 4 in that analysis, which is the level closest to the established measures by SIPRI and IISS. These numbers can be obtained by looking only at the chapter and subchapter data, except for the last rows, which require that we also look for budget items associated with the budgetary manager code of the Ministry of Defence; for details, see the original paper. The chapter National Defence corresponds to approximately three-quarters of the total in Table 5 over the period 2014−2018, and the share of the subchapter for the armed forces varies between 55 and 64 percent of the total (row 2). It is also noteworthy that the subchapter for welfare benefits accounts for a substantial increase over the period, from 1.3 to 3.8 percent of the total (row 15); This subchapter (1003) includes items that are of interest because they contain some indirect information about the nature of military activities. For example, budget items in this subchapter containing the purpose code 03 3 03 30550 (03 3 3055 before 2016) represent expenditures on summer camps for children of military personnel in federal service who have been killed, maimed or disappeared while carrying out duties in “armed conflicts of a non-international character.” This amounted to 76 million roubles in 2015 and 94.5 million roubles in 2018 (not shown separately in Table 5 but included in the aggregate on row 15). Before 2015 there were no such non-secret items associated with the Ministry of Defence in the federal budget. Therefore, such numbers could potentially be used to estimate the number of Russian servicemen killed or injured in the conflicts in Ukraine and Syria that started in 2014 and 2015 (which were considered non-international civil wars by Russia).

The rest of this section is devoted to presenting a more detailed breakdown of defence expenditures than the one offered in Table 5, which is focused on those categories which have received the most attention in previous literature. Even if it is possible to construct a table where all categories are included, this would clutter the exposition to the point of sacrificing any meaningful overview. We will therefore deconstruct the military expenditures three categories at a time and present the results in two separate tables (Table 6 and Table 7). In the first step, we look at state programmes associated with military expenditures. Somewhat predictably, since we expect them to be secret, state programmes account for only a small fraction of non-secret military expenditures, as shown in Table 6. We would expect most of the information pertaining to military state programmes to be secret, and, as shown at the bottom of Table 6, approximately half of all military expenditures in these categories are indeed secret (row 55). Nevertheless, we also see from Table 6 that various state programmes account for small amounts of non-secret military expenditures, such as the programme for foreign policy activities (row 28). Of the non-secret budget items, most expenditures occur in the non-programme category “Implementation of functions of other federal government bodies” (row 34), and, at the subprogramme level, to “Other non-programme activities” (rows 38 and 46). There is also a noteworthy budget item related to the non-programme pensions category on row 31 amounting to almost 9 percent of the total. However, these results are only marginally more interesting than the secret residual from Table 5, so the conclusion must be that we don’t get much further by looking only at these categories. In other words, the secrecy effectively ensures that the interesting parts remain shrouded in mystery.

Nevertheless, we can still attempt to go further. The next step is therefore to look at these expenditures from a different perspective by grouping them by other categories. For simplicity and ease of exposition, we will look only at those categories from Table 6 where the bulk of expenditures appear (i.e., “Implementation of functions of other federal government bodies” and “Other non-programme activities”). This is easily achieved with the pivot table approach by adding filters for those categories. In addition to filtering on the main categories from Table 6, we also limit the results to the Ministry of Defence and the Federal Service of the National Guard, and to only the most important subchapters based on Table 5.

The results are shown in Table 7, where military expenditures have been grouped by budgetary manager, subchapter (only the major ones), group and element. [20] Despite the narrowing of the scope, this representation of the data still captures most of the non-secret defence expenditures, as can be seen at the bottom of the table (row 139). Only 1−6 percent of the non-secret expenditures are not captured (calculated on row 140 as the difference between the non-secret subtotals in Table 6, row 54 and Table 7, row 139). It is noteworthy that Table 7 gives us considerably more detail than Table 6, and that expenditures are more evenly distributed across the non-secret categories compared to Table 6. For example, starting with the Ministry of Defence, we see that a significant share of the non-secret expenditures in subchapter 0201 Armed Forces is made up of payments to staff (Table 7, row 3), and that the most significant share occurs in the category “Money allowances to military personnel” (row 12). Furthermore, around seven percent of total military expenditures represent purchases of goods and services (row 15), and, at the most detailed level, we can also see that these expenditures are allocated both within and outside the State Defence Order (rows 16−25). Only a few budget items occur in subchapter 0208 Applied research in the field of National Defence (rows 35-39) since most expenditures in this subchapter are secret. In subchapter 0209 Other National Defence issues we also observe a substantial amount of expenditures going to capital investments within the framework of the State Defence Order (rows 44 and 46). In subchapter 0706 Higher education, we notice a significant share of expenditures on staff belonging to the Ministry of Defence, corresponding to over one percent of total military expenditures (row 53). Subchapter 1001 Pension benefits (row 67) accounts for a steady share of almost nine percent of total military expenditures over the whole period (also captured by Table 6). We further observe a significant item in subchapter 1003 Social welfare in the amount of over 100 billion roubles annually from 2015 to 2017, around three percent of total military expenditures, in the form of subsidies to citizens to purchase housing (row 80). Interestingly, there are almost no expenditures in this category in 2018, while at the same time a similar amount appears in the category on the row above (row 79) representing social security for citizens, suggesting that these expenditures were reclassified in 2018.

The final part of Table 7 concerns expenditures on the Federal Service of the National Guard, which became operational in 2017. Note that the National Guard constitutes a budgetary manager (row 82) but that there is also a subchapter concerning the troops of the National Guard (row 83). A comparison with the Ministry of Defence reveals some noteworthy results. To start with, the overall cost structure of the non-secret expenditures is quite similar between the two bodies. However, the Armed Forces spend a larger share of non-secret expenditures on goods and services (row 15) relative to personnel compensation (row 3) compared with the National Guard troops (rows 95 and 84). One important difference is that while as much as 60 percent of expenditures on the Armed Forces were secret in 2017, the corresponding figure for the National Guard troops was only 7 percent (not shown separately in Table 7 but included in the total on row 141). Furthermore, in 2018 the Ministry of Defence allocated four percent of its total non-secret expenditures on higher education (row 52 as a share of row 1), while the National Guard spent two percent (row 111 vs. 82). Finally, we observe that the National Guard also spends a relatively large share on subsidies to purchase housing (row 137 vs. 82), which corresponded to over one percent of total expenditures of the Federal Service of the National Guard in 2018.

Although Table 6 and Table 7 provide a fairly detailed disaggregation of Russia’s military expenditures, we should be careful in how we interpret the results; the choice of presentation is somewhat arbitrary, and certain categories have been left out to avoid cluttering the exposition. And we should always keep in mind that over half of all expenditures are secret (row 55 in Table 6 and row 141 in Table 7) in the sense that we cannot obtain a disaggregation concerning budgetary managers, programmes and expenditure types.

To conclude this section, we can also note that, as has been widely reported elsewhere, a substantial one-off increase in the defence budget was made in 2016 to settle debts taken by the defence industry since 2011 to accelerate the State Armament Programme (see for example IISS 2018, p. 175). It is noteworthy that the increase in 2016 almost exclusively affects the secret part of the expenditures (row 55 in Table 6 and 141 in Table 7). Unsurprisingly, this observation supports the notion that a significant part of the secret items is related to the defence industry and the State Armament Programme (GPV). Indeed, there are several budget codes in the secret state programme Development of the defence-industrial complex with the explicit purpose of paying subsidies to firms as compensation for interest payments with state guarantees, to compensate for lost revenue, or to prevent bankruptcies (see Table 10 for details). This brings us to the final part of the empirical findings.

The nature of secret budget items

The empirical investigation presented above is primarily concerned with the open, non-secret part of Russia’s military budget. Before we conclude the discussion, however, it is worth pointing out that the sources we have used to construct our dataset have the potential to shed some light on the nature also of the secret expenditures, even if we cannot obtain the actual expenditures in rouble terms. These observations are interesting in themselves but are also helpful in clarifying the precise meaning of secrecy in the Russian budget, which becomes possible because the Finance Ministry’s publicly available regulatory instructions contain budget classification codes for all budget categories, including the secret ones. This is the final key contribution of our paper, which has not been discussed before in the literature. One example is main activity 32 1 01 00000 counterintelligence activities (kontrrazvedyvatel’naya deyatel’nost’). The source of this particular code definition is Part III, section 4(1).2.2(1).30 of the instructions (MinFin 2018a),[21] which details the codes associated with the state programme Ensuring state security. This state programme is largely secret and therefore absent from Roskazna’s budget execution reports, although a very small amount related to state-border protection was openly reported in the budget for 2014 (rows 25 and 27 of Table 6). A complete list of all budget codes associated with this programme based on the Finance Ministry’s instructions is provided in Table 8. Likewise, a similar list is provided in Table 9 for the state programme Ensuring the country’s defence capability.[22]

Table 8 and Table 9 in the annex give the code definitions for the programme, subprogramme and main activity levels of the purpose code of these two state programmes. We can obtain additional information by also looking at the corresponding direction codes (i.e., the last five characters of the purpose code), but in the case of those two programmes this yields little extra information. However, in the case of the third and last of the secret state programmes, Development of the defence-industrial complex, there is additional interesting information in the direction code. Most significantly, these categories contain information about all companies in the defence-industrial complex that receive budget subsidies in the form of contributions to their authorized capital. Table 10 shows the budget codes of this state programme, including the complete list of 95 companies in the defence-industrial complex which receive subsidies from the federal budget. A small number of expenditures under this programme is openly reported (row 30 of Table 6). Table 10 includes several other noteworthy secret budget categories, for example main activity 44 5 07 00000 Research in the field of nuclear weapons and means of protection from the effects of their damaging factors. It is immediately obvious why expenditures in this category are secret, and this observation again serves to underscore the meaning of secrecy in the Russian budget: expenditures are secret if they would otherwise expose detailed information about sensitive Russian government activities. The secrecy also implies that we don’t even know to which budget chapter these expenditures belong – they might not even be captured by the tables presented above but instead be embedded in budget chapters that are outside the scope of the regular military (chapter 0200) or security (0300) budget. We can, however, be fully confident that – even if they are subject to such cross-chapter secrecy – they are captured by the analysis in Andermo and Kragh (2020) since they fall within the definitions used to construct some of the estimates presented there (namely level 8; see the original paper for details).

Summary, policy implications and future research

Existing aggregate estimates of Russia’s defence expenditures, such as the annual publications of SIPRI and IISS, provide only a crude overview of the trends in the Russian budget. More details are needed to detect nuances in the overall trends, such as a shift between external and internal security in the budget priorities. The establishment of the National Guard as a separate entity in 2017 clearly heralded such a shift, as can be seen, for example, in the data provided in Table 5.

The main contribution of this article has been to provide a detailed analysis of the structure of the Russian federal budget, with a focus on expenditures related to defence and security, using a novel, data-driven approach. The empirical tables presented here provide a degree of detail not previously reported in this regard. We have also documented the steps taken to obtain the results. Apart from the empirical results, the study of the structure of the budget is also meaningful since it provides some insights into Russian governance in general. It is clear from the data that the budget structure evolves with political priorities, and it could be argued that the budget has become a more potent tool over the years for transparency and accountability in the Russian government, not least in relation to the numerous state programmes. Although the budget system is clearly relatively sophisticated, our study has not considered its effectiveness or efficiency. The multi-layered and relatively complex structure of the budget also hampers transparency to a significant degree. Presumably this fact is a result of the policy requirement to classify such large shares of the budget as secret.

More specifically, our data-driven analysis has shed new light on two areas of the Russian federal budget. Firstly, we have provided a highly disaggregated image of Russian defence and security related expenditures. Secondly, we have also been able to draw inferences about the nature of secret budget expenditures in the areas of security, defence, and the defence industry, highlighting how certain areas of budget secrecy can be partially circumvented.

Although our empirical results contain many details, they do not tell the whole story. We have deliberately limited the scope of our study to only official data published by Russian government authorities. A more complete analysis would consider other sources of information, such as news reports, financial statements of companies in the defence sector, and comments by Russian officials, to provide more context to the raw data presented here.

References

Andermo, Erik and Martin Kragh. 2020. “Secrecy and Military Expenditures in the Russian Budget”. Journal of Post-Soviet Affairs, 36(4), 297–322. doi.org/10.1080/1060586X.2020.1738816

CAST. 2015. Gosudarstvennyye programmy vooruzheniya Rossiyskoy Federatsii: problemy ispolneniya i potentsial optimizatsii (State armament programmes of the Russian Federation: problems of execution and the potential for optimization). Centre for Analysis of Strategies and Technologies.

Connolly, Richard and Mathieu Boulègue. 2018. Russia’s New State Armament Programme. Implications for the Russian Armed Forces and Military Capabilities to 2027. Research Paper, May 2018. Chatham House, Royal Institute of International Affairs.

Cooper, Julian. 2013. Russian Military Expenditure: Data, Analysis and Issues. FOI report FOI‑R—3688—SE, September 2013. Swedish Defence Research Agency. doi:10.13140/RG.2.1.2548.6480

Cooper, Julian. 2016. If War Comes Tomorrow. How Russia Prepares for Possible Armed Aggression. Whitehall Report 4-16, August 2016. Royal United Services Institute for Defence and Security Studies.

Diamond, Jack. 2003. From Program to Performance Budgeting: The Challenge for Emerging Market Economies. Working Paper WP/03/169, June 2003. International Monetary Fund.

Fortescue, Stephen. 2017a. “Russian federal budget formation: introduction.” Post-Communist Economies, 29(4), 449-456.

Fortescue, Stephen. 2017b. “The role of the executive in Russian budget formation.” Post-Communist Economies, 29(4), 523-537.

Hakvåg, Una. 2017. “Russian defence spending after 2010: the interplay of personal, domestic, and foreign policy interests.” Post-Soviet Affairs 33 (6), 496−510. doi:10.1080/1060586X.2017.1388472

Hughes, Richard., Tom Josephs, Viera Karolova, Vladimir Krivenkov, and Gösta Ljungman. 2014. Russian Federation. Fiscal Transparency Evaluation. IMF Country Report No. 14/134. International Monetary Fund, Fiscal Affairs Department and Statistics Department.

Hutchings, Raymond. 1987. Soviet Secrecy & Non-Secrecy. Totowa, New Jersey: Barnes & Noble Books.

IBP. 2017. “Open Budget Index 2017”. International Budget Partnership. Accessed January 2019. https://www.internationalbudget.org/open-budget-survey/open-budget-index-rankings/

IISS. 2018. “Russia and Eurasia.” In The Military Balance. The Annual Assessment of Global Military Capabilities and Defence Economics, 169−218. London: The International Institute for Strategic Studies.

Kontorovich, Vladimir. 2009. “Secrecy and Western Study of the Soviet Military Sector.” Unpublished; presented at the November 2009 convention of the American Association for the Advancement of Slavic Studies. Department of Economics, Haverford College, Haverford, PA.

Kuzmin, Vladimir. 2018. “Otsenka s minusom” (Evaluation with a minus). Rossiyskaya Gazeta, 5 December, No. 274(7737). https://rg.ru/2018/12/05/medvedev-poruchil-dvum-vedomstvam-povysit-effektivnost-vypolneniia-fcp.html

MinFin. 2018a. Order no. 65n “On the approval of the Instructions on the method for applying the budget classification of the Russian Federation” (prikaz N 65n “Ob utverzhdenii Ykazaniy o poryadke primeneniya byudzhetnoy klassifikatsii Rossiyskoy Federatsii”). Ministry of Finance of the Russian Federation. 1 July 2013, with amendments through 20 December 2018.

MinFin. 2018b. Order No. 132n “On the method for forming and applying budget classification codes of the Russian Federation, their structure and assignment principles” (prikaz N 132n “O poryadke formirovaniya i primeneniya kodov byudzhetnoy klassifikatsii Rossiyskoy Federatsii, ikh strukture i printsipakh naznacheniya”). Ministry of Finance of the Russian Federation. 8 June 2018.

Oxenstierna, Susanne. 2016. “Russia’s defence spending and the economic decline.” Journal of Eurasian Studies 7 (2016), 60−70. doi:10.1016/j.euras.2015.06.001

Oxenstierna, Susanne. 2019. “A new trend in Russia’s defence spending.” In The Russian Economy under Putin, edited by Torbjörn Becker and Susanne Oxenstierna, 87−105. Abingdon: Routledge.

Panov, Pavel. 2017. “Polovina FTsP priznany effektivnymi” (Half of FTsPs recognized as effective). Izvestia, 16 March. https://iz.ru/news/671385

Putin, Vladimir. 2012. Byudzhetnoye poslanie Prezidenta Rossiyskoy Federatsii o byudzhetnoy politiki v 2013–2015 godakh (Budget message of the President of the Russian Federation on budget policy in 2013–2015). 28 June 2012. http://www.kremlin.ru/acts/news/15786.

Roskazna. 2019. “Budget Execution. Federal Budget.” (“Ispolneniye byudzhetov. Federal’nyy byudzhet.”) The Federal Treasury of Russia. Accessed January 2019. http://roskazna.ru/ispolnenie-byudzhetov/federalnyj-byudzhet/

SIPRI. 2019. “SIPRI Military Expenditure Database Sources and Methods”. Accessed 5 March 2019. https://www.sipri.org/databases/milex/sources-and-methods

Transparency International UK. 2011. The Transparency of National Defence Budgets. An Initial Review. Transparency International UK Defence and Security Programme.

Vendil Pallin, Carolina, ed. 2012. Rysk militär förmåga i ett tioårsperspektiv – 2011 (Russian Military Capability in a Ten-Year Perspective – 2011). FOI report FOI‑R—3404—SE, March 2012 (in Swedish with English summary). Swedish Defence Research Agency.

Zatsepin, Vasily. 2008. “Defence in the 2008 federal budget.” Russian Economy: Trends and Perspectives. Monthly Bulletin, November 2008, 56−60. Moscow: Institute for the Economy in Transition.

Zatsepin, Vasily. 2014. “Laws, secrecy and statistics: Recent developments in Russian defence budgeting.” In Proceedings. Forum for Economists International. Conference held in Amsterdam May 30-June 2, 2014, edited by M. Peter van der Hoek. Papendrecht: Forum for Economists International.

Zatsepin, Vasily. 2019. “Voyennaya ekonomika i voyenna reforma v Rossii.” (“Military economy and military reform in Russia”), in Vladimir Mau et al, Rossiyskaya ekonomika v 2018 godu. Tendentsii i perspektivy (Russian economy in 2018. Trends and outlooks). Moscow: Gaidar Institute Publishers. ISBN 978-5-93255-554-5.

[1] Throughout this text, ‘Budget Code’ with capital letters refers to the federal Russian legislation defining how the budget works (Byudzhetnyy kodeks), whereas ‘budget code’, or simply ‘code’, with small letters refers to the budget code associated with each line item in the budget (Russian: kod).

[2] For background on Russian defence spending, see Connolly and Boulègue (2018), Fortescue (2017a, 2017b), Hakvåg (2017), Oxenstierna (2016, 2019), Vendil Pallin (2012), and Zatsepin (2008, 2014, 2019).

[3] This part makes some references to various budget codes (type codes, element codes, etc.); these terms will be further explained in the part on sources and data.

[4] The relevant provision is found in article 2 of federal law No. 60-FZ of 13 December 1994 “On the supply of products for federal state needs” (federal’nyy zakon ot 13 dekabrya 1994 g. N 60-FZ “O postavkakh produktsii dlya federal’nykh gosudarstvennykh nuzhd”).

[5] See http://fcp.economy.gov.ru/ (accessed March 2019; as of 2023, this portal was no longer available).

[6] Government regulation No. 1950-r of 11 November 2010 “On the approval of the list of state programmes of the Russian Federation” (Rasporyazheniye Pravitel’stva Rossiyskoy Federatsii ot 11 noyabrya 2010 g. N 1950-r “Ob utverzhdenii perechnya gosudarstvennykh programm Rossiyskoy Federatsii”).

[7] The government used to operate a web portal with comprehensive information about the various non-secret state programmes. The number 70 percent was one of the main highlights on the main page of this web portal. https://programs.gov.ru/Portal/ (accessed 21 March 2019; no longer available in 2023).

[8] The relevant provisions are found in article 4, item 2.8) and article 6.6) of federal law No. 61-FZ of 31 May 1996 “On defence” (federal’nyy zakon ot 31 maya 1996 g. N 61-FZ “Ob oborone”).

[9] Quoted from article 3, item 1 of federal law No. 213-FZ of 27 December 1995 “On the State defence order” (federal’nyy zakon ot 27 dekabrya 1995 g. N 213-FZ “O gosudarstvennom oboronnom zakaze”).

[10] Article 4, items 3) and 4) of federal law No. 275-FZ of 29 December 2012 “On the State defence order” (federal’nyy zakon ot 29 dekabrya 2012 g. N 275-FZ “O gosudarstvennom oboronnom zakaze”).

[11] The budget law defines the general parameters for the GOZ, not the detailed allocation of funds; the GOZ itself is formally approved by the government within one month from the date when the president signs the budget law, as stipulated article 5 of the law “On the State defence order.”

[12] Most budget items with type codes 211, 214, 216 and 217 are secret, including all items associated with the Ministry of Defence (budgetary manager code 187). The few items that are not secret are associated with the Ministry for Civil Defence, Emergencies and Disaster Management (177), the Federal Drug Control Service (204) and the Federal Penitentiary Service (320). This proves that the GPV is not entirely secret, after all.

[13] More specifically, our data on budget execution comes from Roskazna (2019) and data on budget categorization comes from the Finance Ministry’s 2013 order No. 65n, with amendments through 20 December 2018 (MinFin 2018a), which regulates the budgets for 2014 through 2018. New instructions were issued in 2018 that took effect starting from budget year 2019, see MinFin (2018b).

[14] The State Duma operates a web portal which contains detailed documentation for all stages of the legislative process, including the explanatory notes: https://sozd.duma.gov.ru/ (accessed March 2019; unavailable as of 2023).

[15] Four of the five categories that together make up the budget expenditure classification code are in fact strictly numerical; only the purpose code can contain letters (both Cyrillic and Latin). The need to introduce letters arises from the fact that only one character is used to designate subprogrammes (position 10), and that there can be more than ten subprogrammes related to one state programme. Also, certain letters are always used to designate certain types of programmes, for example the Russian letter П (P) for prioritized projects.

[16] A dataset structured for pivot table analysis is simply a two-dimensional table with a certain structure. The term ‘normalization’ can also refer to the process of structuring a relational database where several tables are linked to each other, which is different from the concept described here.

[17] Refer to Table 1a and Table 1b for the corresponding Russian terms.

[18] The same information is also published on the website of the Finance Ministry, but the information is more easily50 accessible via Konsultant. There are also several other commercial websites specializing in the publication of Russian public documents.

[19] The entry into force of an amendment can be different from the date when the order was issued that executed the amendment, which risks introducing a certain ambiguity regarding which date to use.

[20] Included subchapters are: 0201, 0208, 0209, 0706, 1001 and 1003; other subchapters are treated as residual items included in rows 81 and 138 of Table 7.

[21] The same information is also listed in Appendix 10.1 of the same source but with less contextual information.

[22] The codes for the state programme for state security is given in the instructions governing the budgets for 2014−2018 (MinFin 2018a), but the programme for defence capability is absent from the same source. However, the code definitions for that programme appear in the instructions for 2019 (MinFin 2018b).