Towards a Swiss Army Knife State? The changing face of economic interventionism in advanced democracies, 1980–2015

Abstract This article systematically reviews trends in numerous economic policy indicators in eighteen OECD countries since the early 1980s, synthesizing findings about the fate of states’ economic interventionism from several customarily separate literatures. Rather than observing any paradigmatic policy shift, the review finds that policies with markedly different ideational foundations currently cohabitate. In line with non-interventionist prescriptions, policymakers have largely abandoned the intrusive heterodox ‘power tools’ of previous eras, while establishing new norms for monetary policy based on monetarist theory. However, this has not led to a full retreat of economic interventionism. Instead, policymakers are gradually developing a new, albeit more constrained, approach to promoting economic activity in line with selected distributional goals – here labelled the micro-interventionist state, or the ‘Swiss Army Knife State’, as it were. The cross-partisan appeal of the ‘multi-tools’ associated with this approach – such as horizontal industrial policy, active social policy, and strategic tax expenditures and procurement – partly stems from their versatility, as policymakers can use them to very different distributional ends. To better understand the politics and distributional consequences of contemporary economic policies, scholars need to take their versatility more seriously, shifting focus theoretically and empirically from how much to how policymakers intervene in the economy.


Introduction
In many academic and policy-oriented communities in the borderlands between political science, economics, and sociology, a popular account of the long-term changes in economic policy in the advanced industrial economies holds that in recent decades, the economy has become increasingly insulated from discretionary interventions by the democratic state. Since the beginning of the 1980s, so it is commonly held, the strongly interventionist Keynesian economic policy paradigm has been gradually succeeded by a paradigm based on monetarist economic theory, which has a skeptical view of the government's active involvement in the economy. In this new eraoften referred to as the 'era of austerity' or the 'neo-liberal era'there is little room for discretionary economic policy interventions, and in turn the prospects for democratically elected governments to influence the distributional outcomes of economic activities are seen as limited at best (e.g. Flinders & Wood, 2014;Sch€ afer & Streeck, 2013;Streeck, 2014).
While the studies that actively advance the stylized narrative outlined above are relatively few, scholars still often take these trends as a point of departure and analyze their implications for various topics, ranging from the functioning of democracy (Bailey, 2015;Flinders & Wood, 2014) to social resilience (Hall & Lamont, 2013), public management systems (Lodge & Hood, 2012), and many others.
For some time now, however, others have raised doubts about to what extent the state has in fact retreated from discretionary economic policy interventions (e.g. Cohen & Centeno, 2006;Hay, 1999;Mazzucato, 2013;Raess & Pontusson, 2015). However, these studiesas well as those that have observed systematic state withdrawal (e.g. Lampe & Sharp, 2013;Schuster et al., 2013) have in most cases focused on particular countries or particular policy areas, and the touchpoints between them have been few. Consequently, a comprehensive and synthesizing review of the fate of economic interventionism across the advanced democracies since the onset of the neo-liberal era until today is still lacking.
This article's first purpose is to carry out such a review in a systematic manner, spanning more than twenty economic policy areas over four decades. Rather than observing any paradigmatic policy shift among the eighteen countries under consideration, this analysis finds that policies with markedly different ideational foundations currently cohabitate.
What this analysis furthermore reveals is the widespread development of what will here be called the micro-interventionist stateor the 'Swiss Army Knife State', as it werein the realm of economic policy. The second purpose of the article is to advance a four-fold argument about this policy approach, which seeks to intervene to discretionarily promote new economic activity in various ways that advance selected distributional goals yet without singling out particular industries or firms in the process.
First, the 'multi-tool' policy instruments associated with micro-interventionist economic policyto which we may foremost count horizontal industrial policy, active labor market policy, work-life balance policy, strategic tax expenditures and strategic procurementare gradually becoming more important, as many of the more contested heterodox 'power tools' of previous erasat least until very recentlyhave been removed from the economic policy menus in the advanced democracies.
Second, although there thus appears to be a growing consensus among political actors about which economic policy tools to use and to what extent, there is, as we shall see, still scope for conflict about how to use them. Importantly, from a democratic perspective, the economic policy instruments subsumed under the microinterventionist heading are all versatile enough to be 're-purposed' and applied as a means to very different distributional ends. This means that policymakers from across the political spectrum, although facing various constraints, can still use economic policy strategically to advance the distinct interests of their core constituents. In other words, rather than having lost their relevance in the post-industrial era, traditional partisan distributional conflicts now extend into the domain of the micro-interventionist economic policies and find new expressions in their detailed programmatic designs.
Third, the growth of these versatile policy interventions has profound implications for our understanding of the relationship between state economic interventionism and redistribution. Indeed, the long-term trends in micro-interventionist public spending are found to be unrelated to the concurrent trends in economic inequality, which cautions against assuming, as many do, that state interventionism is inherently equalizing. Fourth, it is argued, conceptually micro-interventionism has a number of analytical advantages over existing concepts when it comes to capturing the observed policy trends.
Having laid out these arguments successively, the article lastly discusses possible explanations for the changing face of economic interventionism, after which it concludes by spelling out some implications for political economy research. Most importantly, it is necessary to shift our theoretical and empirical focus more from the question of how much to the question of how and for what purpose policymakers intervene in the economy, not only for better understanding the distributional consequences of contemporary economic policy in the advanced democracies, but also for correctly understanding the changing role of the democratic state in economic policy-making.
Long-term trends in economic policy: a comprehensive review As mentioned above, the general retreat of the state from the economy has continued to be a frequent underlying assumption in research conducted in recent decades. And clearly, in line with neoliberal or monetarist prescriptions, there is evidence of widespread state withdrawalat least until the last few yearsfrom certain types of interventionist policies, including tariffs, industrial subsidization and regulation, public enterprises, and central bank operations (e.g. Garriga, 2016;Lampe & Sharp, 2013;Schuster et al., 2013).
However, by now scholars have also observed sufficient incidences of discretionary economic policy interventionswithin particular countries or particular policy fieldsto suggest that the neo-liberal approach to economic policy-making is gradually giving way to something new, or to raise doubts about its spread in the first place. Among country-focused studies, some of the more influential examples include Mettler's (2011), Mazzucato's (2013, and Lindsey and Teles's (2017) analyses of the United States, that highlight ongoing often-overlooked pro-active roles of government interventions into the economy. Observations about the persistence of discretionary state interventionism have also been made by, among others, Hay (1999) regarding the United Kingdom, Levy (2005) regarding France, and more recently by Vail (2018) regarding France, Germany, and Italy. As to studies focusing on particular policy areas, key examples include Raess and Pontusson (2015) on fiscal policy, Aiginger (2007) and Rodrik (2014) on industrial policy, and Bonoli (2013) on so called 'active social policy'.
Thus, to reliably say something about the overall trends in economic interventionism in advanced democracies, one needs to consider a rather wide range of macro-and micro-economic policy tools. To my knowledge, Cohen and Centeno (2006) is the only study to date that does so systematically. Focusing on the period until 1999, the authors found little evidence of a broad scaling back of governments in advanced democracies. However, in reaching that conclusion they drew not only on economic policy but also on the continuously high government spending on traditional social protection programs, such as pensions and health, whichalthough they may increase consumptionare not primarily aimed at affecting economic activity. In other words, as regards economic policy, a comprehensive and long-term cross-national review is still lacking.
To fill this gap, this section reviews the long-term developments in several economic policy sub-fields across eighteen advanced democracies, with the aim of synthesizing the findings from separate research strands concerning the fate of economic interventionism since the early 1980s.
The discussion that follows mostly revolves around Table 1, which reports the developments of a large set of economic policy indicators between the early 1980s and 2015, averaged over a set of advanced democracies. All data are retrieved from existing sources but are presented here in a new, consistent and more comprehensive format. The analysis includes eighteen countries: Australia, Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, the United Kingdom, and the United States. For most indicators, one or a few countries are omitted due to a lack of data for the full set of periods studied. To ease interpretation, the available data are averaged over five periods: the early 1980s (1980)(1981)(1982)(1983)(1984)(1985), the early 1990s (1990)(1991)(1992)(1993)(1994)(1995), the early 2000s (2000)(2001)(2002)(2003)(2004)(2005), the early years of the Great Recession (2008)(2009), and finally, the period from 2012 to 2015, which is the most recent one for which sufficient data are available. Focusing on these five periods is advantageous to reporting individual years because it limits the impact of short-term fluctuations and copes with the fact that not all data are available for all years.

Heterodox economic policy
Let us begin by considering what Pontusson and Raess (2012, p. 503) call 'heterodox' economic policies, which refers to policies that 'seek to resist or steer market pressures for change' and which roughly corresponds to those that in earlier work by Gourevitch (1986) and Hall (1986) are labeled protectionism, nationalization, and economic planning.
First, it is well established that protectionist trade barrierscrudely represented in Table 1 by two partly overlapping indicators on the weighted mean level of tariff on importswere on the decline throughout the post-war period (Lampe & Sharp, 2013) and had become a less important measure for tackling economic and social challenges by the end of the period observed here (Pontusson & Raess, 2012).
Second, whereas economic policy during the Long Recession often involved extensive direct state involvement in economic activity, via the expansion of state enterprises, Table 1 reports that employment in such enterprises has seen a sharp decline since the early 1980s. Substantial declines can also be seen for indirect state involvement through competition-restricting regulation and subsidization of certain  1980-1985, 1990-1995, 2000-2005, 2008-2009, and 2012-2015. Exceptions are noted below. Compliers follow the trend. Remaining at 0 or level just above 0 counts as complying if trend is downward. a Import-weighted average ad valorem tariff (%) (Lampe & Sharp, 2013).  GDP growth. 1981, 1990-1991, 2001, and 2008-2009(Raess & Pontusson, 2015 1980. q % of GDP. Public and mandatory private in-kind benefits. 1980-2014 (OECD, 2017e). r % of GDP. Sum of R&D, training, employment subsidies, labor market services, supported employment and rehabilitation, parental leave, childcare, and care for old age/incapacity. Missing: AU, IT, LU, NZ, US. s % of GDP. Includes aid to research and development, environment and energy saving, small and medium-sized enterprises, commerce, employment, training, or regional development (Buigues & Sekkat, 2009). 2000(European Commission, 2017. Missing: AU, NO, NZ, US.
industrial sectors, includingnoticeablythe financial sector. Third, comparable developments are observed for two other 'protective' measuresearly retirement for labor market reasons, and the direct creation of temporary public jobsboth of which facilitate the withdrawal of labor from the regular market to contain firms' payroll costs. Both are down by two thirds since the early 1980s and were largely absent as a response to the Great Recession. A conspicuous feature of all policies reviewed above is that they have been largely uniformly abandoned. As evidenced by the large share of compliers, meaning countries that follow the average trend, the advanced democracies have converged towards a low level of heterodox economic policyconsistent with neoliberal and monetarist prescriptions.
Given that our observations end in 2015, a caveat is however warranted about the backlash in trade policy liberalization observed more recently. The 2016-2019 period saw a considerable growth in trade distortions compared to 2012-2015 in most of our eighteen countries, and in early 2020 additional measures were implemented in response to the COVID-19 pandemic, including export restrictions on medical supplies in many countries (Global Trade Alert, 2020a, 2020b).

Monetary policy
Let us turn next to monetary policy, where the long-term developments also show evidence of monetarist imprints. Particularly, the delegation of monetary policy to independent central banksemerging in the 1980s as a recipe to prevent negative inflationary effects of shortsighted electoral ambitionshas been widely adopted over the past decades. For our eighteen countries, the average of the central bank independence index compiled by Garriga (2016) roughly doubled during the 1980s and 1990s, and has remained at that level since the turn of the century. The components that constitute the index clarify what this development represents: Central banks have been given greater authority over monetary policy formulation, the ability of governments to influence central bank personnel decisions and to use central bank credit to finance their operations has been curtailed, and price stability has become the more dominant goal for central banks. Consistent with the aims of these policy changes, inflation has decreased in a converging manner and stabilized at low levels. Averaged over the period 2012-2015, inflation in our sample ranged between 0.22 and 2.28 percent, with a mean of 1.47 percent (Armingeon et al., 2019). While central bank independence and price stability targeting of policy rates likely are not the only factors contributing to the current low-inflation regime, Iversen et al. (2000, p. 14) conclusion that 'monetarist policies seem to have triumphed everywhere' still appears strikingly valid.
Trends in two unconventional forms of monetary policy also deserve attention here. The first one concerns depreciating currency interventions, a powerful tool for governments that seek to alter the terms of trade in favor of domestic producers. Represented in Table 1 by an indicator on the average year-to-year percentage change in the exchange rate of the national currency to the US dollar, such interventions have been largely abandoned since the early 1980s. For many of the countries under consideration here, this is obviously in part a consequence of the establishment of the Eurozone in 1999, which meant that they would no longer have currencies of their own.
A contrasting, more interventionist, trend is observed in another unconventional central bank tool, namely large-scale asset purchases to boost money supplyoften called 'quantitative easing'. During and after the Great Recession, when monetary policy rates approached their effective lower bound, unprecedented expansions of central banks' balance sheets took place in the Eurozone, Sweden, the United Kingdom, and the United States (Gambacorta et al., 2014), as crudely represented in Table 1 by an indicator on average year-to-year changes in central bank assets as share of GDP.

Fiscal policy
Next, consider the counter-cyclical fiscal policies associated with Keynesian demand management, which can be said to consist of two components: discretionary fiscal policy and 'automatic stabilizers' that act to dampen fluctuations in real GDP. With respect to discretionary fiscal policy, Raess and Pontusson (2015, p. 18) recently concluded that there has been 'no general retreat from fiscal policy activism' among the advanced democracies since the early 1980s. Table 1 reports the average of the indicator on fiscal policy activism used in their study, which measures the size of discretionary fiscal stimulus per 1 percent contraction in GDP growth for those among the eighteen countries that used fiscal stimulus as a response to a deceleration in GDP growth in any of the four international recessions in 1981, 1990-1991, 2001, and 2008-2009. The indicator, which is based on econometric estimates by the OECD of the so-called underlying government primary balance, 'adjusts for fluctuations in government expenditures and revenues due to the business cycle and, thus, pertains to changes in expenditures and revenues that can be attributed to government decisions' (Raess & Pontusson, 2015, p. 6). Notably, one-off fiscal operations such as the massive bailouts of financial institutions rolled out in a number of OECD countries in 2008-2009 are excluded. Even though the measure is adjusted for the magnitude of the economic downturn, it indicates no downward trend in fiscal policy activism between the early 1980s and the Great Recession.
The pattern noticeably changes when we apply Raess and Pontusson (2015) methodology also to the years 2012-2015, of which it classifies 2012 as a year of international recession. Such an extension, using data from OECD (2017d), shows that none of the countries experiencing a GDP contraction that year -Finland, Italy, Netherlands, Portugal, and Spainused fiscal stimulus; indeed, all responded in a contractionary manner. That observation, but also the remarkable fiscal stimulus packages announced to tackle the economic aftermath of the COVID-19 pandemic in early 2020, which in several countries amounted to up to 9-10 percent of GDP at the time of writing (IMF, 2020), suggests that the jury is still out regarding the long-term trends in fiscal policy activism.
Turning from the discretionary to the automatic component of fiscal policy, it is clear that the size of the automatic stabilizers has increased in the aftermath of the Great Recession compared to previous periods. Needless to say, the rise in the indicator reported here, which measures the cyclical component of the fiscal balance as a share of GDP, is related to the fact that the Great Recession was much deeper than the previous recessions during the observed period (Raess & Pontusson, 2015). Nevertheless, it does indicate that the countries considered here have not lost their capacity for the counter-cyclical stabilization of aggregate demand.

Micro-interventionist economic policy
As noted above, several scholars have questioned the notion of a general retreat of state interventionism by pointing out various policy instruments, through which governments discretionarily influence economic activity, that have shown no sign of contractionand indeed, in several cases, have expanded. Bringing together this research, we next review a number of these instruments and place them under the new common label micro-interventionism.
To clarify the motivation for distinguishing these policy instruments from others, Figure 1 provides aninevitably stylizedclassification of the economic policy interventions considered in this article according to two continuous dimensions. The first dimension concerns whether the interventions operate largely nonselectively at the macro-level of the economysuch as monetary policy operations to boost overall money supply and depreciating exchange rate interventions intended to favor domestic production generallyor more selectively at the micro-levelas, for example tariffs, regulations, or subsidies targeting particular goods, services, or social goals. The second dimension concerns to what extent they are primarily intended to protect specific existing modes of economic activity (e.g. industries, firms, or jobs) rather than to promote economic activity more broadly. This classification is only intended to distinguish between overarching categories of policies and makes no attempt to identify the exact location of any particular policy along the two dimensions 1nor does it cover the universe of economic policy interventions.
Located towards the lower-right corner of this chart, the micro-interventionist policies are labelled as such to enable a two-fold analytical distinction. First, the prefix micro-serves to indicate that these policy instruments are more versatile and can be targeted more selectively towards particular social goals than macroeconomic policies intended to stimulate economic activity more broadly, such as expansionary monetary policy and general tax reductions for businesses or consumers. Second, the term is meant to designate that these policies are less intrusive, in that they tend to bring about less distortion into any given market, than the likewise selective heterodox 'power tools' of previous eras, such as targeted trade barriers and subsidization intended to protect specific industries or national 'champions' from competition. Micro-interventionist policies, as conceived here, rather seek to promote new economic activity in ways that advance selected distributional goals without targeting specific industries or firms, for instance through facilitating certain types of innovation, increasing labor supply and human capital of particular categories of workers, or boosting demand for selected goods and services that comply with some socially desirable standard.
Accordingly, considering the growth of micro-interventionist policies, one might say that the state that we currently see expanding can be portrayed, if not precisely as 'a kind of billiard ball, pushed around by competing interest groups' (Hall, 1986, p. 17), then perhaps as a kind of multi-purpose toola 'Swiss Army Knife State', as it werewhich governments from across the political spectrum may find useful for intervening in the economy to promote economic activity in various ways that are aligned with the distinct social preferences of their core constituents. We return to the versatile and political nature of the micro-interventionist policy tools below, but first let us consider the five policy categories that most clearly belong in that family: (1) horizontal industrial policy, (2) (most) active labor market policy, (3) work-life balance policy, (4) strategic tax expenditures, (5) and strategic procurement.

Horizontal industrial policy
Recent scholarship on industrial policy has indicated that the conclusiondrawn, for instance, by Schuster et al. (2013) that the long-term decline in industrial subsidization reported above can be interpreted as a definitive retreat of the modern state from entrepreneurial activities is somewhat premature (Aiginger, 2007;Mazzucato, 2013;Rodrik, 2014). Aiginger (2007 and later works), for example, documents a renewed interest in industrial policy among especially the European countries in the twenty-first century, reflected in policy agendas such as the Lisbon Strategy and in enacted policy. While different terms have been used to distinguish the 'new' approach to industrial policy from the 'old' one, the distinction between horizontal support, which has a broad impact on many or all industries, and vertical support, which targets specific sectors or firms, provides a good starting point (Buigues & Sekkat, 2009). As reviewed above, traditional vertical industrial aid has indeed lost its appeal among policymakers and experts. It has also become more difficult to promote, due to the expansion of trade agreements, the work of the WTO, and the strengthening of the European Single Market (Aiginger, 2007). In contrast, horizontal aid, aimed at promoting economic activities that are 'common to a large number of sectors and firms that suffer market failures', is gaining popularityand these types of aid are also less restricted by the WTO and the EU (Buigues & Sekkat, 2009, p. 5).
Following the European Commission, Buigues and Sekkat (2009) define horizontal state aid as aid that is targeted at research and development, environment and energy saving, small and medium-sized enterprises, commerce, employment, training, or regional development. Among these objectives, long-term public expenditure data are only available for research and development, employment subsidies, and training. Notably, the two latter policy fields are commonly included also in the definition of active labor market policy (ALMP), which demonstrates a conceptual overlap between industrial policy and ALMP. The time series reported in Table 1 support the notion that horizontal industrial policy is on the rise: Since the early 1980s, public expenditure on research and development has increased by 20 percent and employment subsidies are up by 73 percent. Training expenditure has seen a slight growth over the full period, although it has markedly declined since the early 1990s.
Importantly, not all forms of state aid show up in the public expenditure statistics, but support can also take the form of equity participation, soft loans, guarantees, and tax breaks. The European Commission's (2017) State Aid Scoreboard provides data on cases of state aid among the EU Member States since the year 2000. A shorter time series based on these data, which for the most part overlaps with those just discussed in terms of countries and coverage, corroborates previous findings. For the 14 EU countries under consideration here, horizontal state aid has seen a 37 percent increase since the early 2000sa major part of which represents aid to support energy saving and environmental protection (see also Rodrik, 2014). In absolute terms, it is evident that the long-term increase in horizontal industrial policy does not make up for the decrease in overall industry subsidization reported above. The trend does, however, lend some support to Aiginger's (2007) depiction of industrial policy as a 're-emerging phoenix' rather than a 'dying breed'.

Active labor market policy
Active labor market policy (ALMP) refers to labor market interventions targeted at the unemployed, those at risk of losing their jobs, or certain groups outside the labor forcesuch as housewives and single parentsto facilitate their entry into (or prevent their exit from) employment. As was just demonstrated above, conceiving of ALMP as a distinct category of interventions is problematic, both because of its overlap with other policy areas and because of the great variation among the programs commonly classified as ALMP. Bonoli (2013) notes that ALMPs vary in terms of how pro-market employment-oriented they are, and Farnsworth (2012) argues that some of them are particularly beneficial to private industry. On that basis we mayin addition to employment subsidies and training programsconsider two more subcategories of ALMP as belonging in the micro-interventionist tool-box. The first is labor market services, which comprises placement and counseling services aimed at facilitating the job search activities of non-employed persons, and services to assist employers in recruiting and selecting staff. The second subcategory is sheltered and supported employment and rehabilitation, which covers interventions that aim to promote the labor market integration of persons with reduced working capacity. As reported in Table 1, both have seen a long-term expansion particularly labor market services.

Work-life balance policy
The third major category of micro-interventionist multi-tools is work-life balance policy, the term Bonoli (2013) uses to denote those family policies that promote labor market participation for parents, especially women. The main pillar of these policies is subsidized childcare, but other sorts of support to reconcile work and family life also belong here. One is parental leave policy, which, even though it promotes temporary exits from employment, is 'clearly meant to maximize labour force participation' (Bonoli, 2013, p. 26). Another one is the provision of subsidized services to provide care for the elderly and the incapacitated. Although such policies do not have the human capital-enhancing function of childcare, they are equivalent insofar as they promote labor force participation by making it easier for workersagain, especially womento remain on the labor market longer instead of withdrawing early to care for elderly or incapacitated relatives.
Work-life balance policies can be considered micro-interventionist in at least three regardswhich also distinguish them from traditional social protection programs. First, they are often explicitly aimed at increasing economic activity through promoting paid work (and, indirectly, human capital accumulation) among targeted groups (Bonoli, 2013). Second, they extend state involvement fartherbeyond income protectionto additional activities that 'for Keynesians as much as neoliberals' are seen as private matters, managed either through the family or the market (Jenson, 2012, p. 71). Third, due to their higher versatility, work-life balance policies, more so than traditional social protection, may have markedly different distributional profiles (Gingrich & Ansell, 2015;Lancker & Ghysels, 2012). Moreover, as Bonoli (2013) observes, the philosophy and the policy-making logic that govern work-life balance policies are for the most part rather similar to those that govern ALMP. Table 1 reports time series on the average public and mandatory private expenditures on parental leave, childcare and other in-kind family support, and on care for the elderly and the incapacitated, respectively. Seen over the full period, each indicator has grown considerably, and the increase has been particularly pronounced for the two that represent services.

Strategic tax expenditures
Two additional policy instruments belong under the micro-interventionist umbrella but are more difficult to explore empirically. The first is strategic, or social, tax expenditures (STEs). Tax expenditures in generaldefined as 'departures from the normal tax structure [ … ] designed to favor a particular industry, activity, or class of persons' (Surrey & McDaniel, 1985, p. 3)are by definition interventionist because they selectively alter equilibrium prices and output (Buigues & Sekkat, 2009). But while they receive plenty of attention in industrial policy research, they have 'remained a blind spot in much of the welfare state literature' (Morel et al., 2016, p. 3).
The revenue loss that follows from tax breaks with social purposes today corresponds to a substantial share of traditional social spending in the United States and in several European countries (Adema et al., 2014;Howard, 1997;Mettler, 2011). Nevertheless, there is a lack of exhaustive and comparable data on STEs, especially for longitudinal analysis. A recent literature survey, however, indicated a trend among European welfare states during the last 20 years, towards reforms that make use of STEs, not least in the fields of ALMP, family policy, and child care (Morel et al., 2016). Because of the interventionist and versatile nature of such STEs, they clearly belong in our conception of the micro-interventionist state.

Strategic procurement
The final policy instrument to consider is public procurementthat is, the purchase of works, supplies, and services by governmental bodiesand more specifically, strategic procurement (also known as targeted or social procurement). Whereas direct subsidies as well as the provision of research, education, and training are 'push factors' through which governments can affect markets in a favored direction, strategic procurement can be used as a 'pull factor' towards the same end by increasing demand for specific goods or services (Buigues & Sekkat, 2009). Although the use of public procurement for strategic purposes is to some extent limited by the international rules stemming from the WTO and the EU, it may still prove a powerfuland inherently politicalpolicy instrument (Buigues & Sekkat, 2009).
It is difficult to assess the extent to which strategic procurement is being used today and has changed over time. However, many pieces of evidence point towards a growing importance of this policy tool, while few, if any, indicate a decline. To begin with, available data show that public procurement overall amounts to a substantial and upward trending share of GDP among the advanced democracies. In 2014, the average for the eighteen countries studied here was 13.5 percent of GDP, up from 12.7 in 2007 (OECD, 2017a). With respect to strategic use of public procurement, recent reports from the EU and the OECD point to a substantial and increasing level among member states (European Commission, 2015;OECD, 2013OECD, , 2015OECD, , 2017a. Interestingly, it is only since 2013 that the OECD's biennial Government at a Glance reports contain specific subsections on strategic public procurement. Moreover, three new directives on public procurement adopted by the EU in 2014 have made it 'much easier' for contracting authorities in EU Member States to require that contractors comply with a range of social requirements (Barnard, 2017, p. 4). Barnard's call on labor lawyers to 'start taking procurement law seriously' may well turn out to be useful advice to political economists as well (2017, p. 1).

The strong compliance with the general trends
In summary, rather than observing any paradigmatic policy shift towards noninterventionism, the review above suggests that economic interventionism is undergoing a change of face, and that economic policies with markedly different ideational foundations currently cohabitate. What is furthermore worth highlighting is that the observed policy trends appear strikingly uniform across countries.
As evidenced by the right-most column of Table 1, between the early 1980s and today, each of the heterodox policy indicators has seen a decrease for 75 to 100 percent of the countries. In the realm of monetary policy, there is an equally strong compliance towards central bank independence and a low-inflation regime, well in line with monetarist prescriptions. With respect to fiscal policy, it is illustrative that Raess and Pontusson (2015) found that each of the observed countries considered here, except Sweden, introduced fiscal stimulus as a response to the Great Recession. And with respect to the growing micro-interventionist policies, at least half complied with the general upward trends in each case except training. If we sum up the eight micro-interventionist indicators for the 13 European countries for which they are all available, all countries comply with the upward long-term trend, 2 although Portugal and Spain saw slight reductions in the last period. Conversely, among those countries, only Denmark did not see a decrease in the three heterodox expenditure categories when considered together.
Furthermore, these trends are not unique to Europe. Indicators on micro-interventionist expenditure can also be computed for Australia and the Unites States, disregarding parental leave, and New Zealand, disregarding research and development. Between the early 1980s and the early 2000s, micro-interventionist expenditure increased sharply, especially for Australia and New Zealand, after which Australia and the United States largely levelled off while New Zealand continued upwards. Heterodox expenditure steadily declined in all three countries, albeit with a slight rebound for Australia in 2012-2015.

Who stands to benefit from micro-interventionism?
The distributional versatility of micro-interventionist policy Can the developments described so far be seen as evidence for the notion that there has also been a growing ideological consensus among the politicians under whose watch these policy changes have taken place, and consequently a reduced level of conflict regarding the distributional outcomes of economic policy? To some extent, yes. As summarized in the recent reviews by Stephens (2015) and Potrafke (2017), there is now considerable evidence from studies of long time series data from OECD countries that the impact of partisan politics on economic policy has generally become less pronounced over recent decades. Inevitably, however, these studiesincluding those of micro-interventionist policiestend to rely on aggregate data on public expenditure on different categories of policy instruments. These data are useful for comparing how much a particular policy instrument is being used across countries or over time, but less informative with respect to how and for what purpose.
The argument of this article is not merely that micro-interventionist policies are broadly on the rise, but also that the appeal of these 'multi-tools' partly stems from their versatility, as policymakers from across the political spectrum can design and deploy them as a means to very different distributional ends. Thus, because the same policy instrument can serve different purposes, aggregate expenditure data are of limited use if we want to understand how struggles between political actors affect the design and, in turn, the distributional outcomes of economic policies. To do so, we need to disentangle these policies both theoretically and empirically.
By way of illustration, the brief exposition below serves to demonstrate that for each of the five micro-interventionist policy fields, research supports the notion of a growing consensus about the appropriateness of using the tools in question but that they are being used for different purposes in different countries or by governments with different ideological compositions.
First, as regards industrial policy, Schuster et al. (2013) studied the retreat of the state from entrepreneurial activities in 20 OECD countries between 1980 and 2007, and found no relationship between government ideology and the propensity to privatize public enterprises or to reduce the overall amount of subsidization to industries. However, subsidies and other forms of state aidincluding the expanding horizontal versionsare malleable in their distributional profiles in ways that make them prone to political conflict. For instance, Fukumoto (2017), found that parties in Italy and Belgium whose core constituencies are low-skilled workers are more prone than other parties to favor regional development aid and broad investment incentives, over research and development grants whose distributional profile supposedly favor the high-skilled.
Turning to ALMP, scholars have recently begun to relax the commonplace assumption that these policies primarily benefit labor market 'outsiders' at the expense of 'insiders', and instead argue that they may actually target rather different groups and consequently come with rather different distributional profiles. For instance, studies have shown that right-wing governments are more positive towards ALMP when such policies partly or exclusively target outsiders, whereas left-wing governments are more prone to expand programs that also target 'insiders' (Cronert, 2019;Gingrich & Ansell, 2015).
The work-life balance policies also have no pre-defined distributional profile, but may benefit different groups depending on targeting. For instance, analyzing the distributional effects of childcare in Flanders and Sweden, Lancker and Ghysels (2012) found differentindeed oppositeoutcomes in the two countries. In Sweden, where employment among low-skilled women is higher and where access to childcare is guaranteed and fairly evenly distributed, redistribution is greater. Gingrich and Ansell (2015) observed a political implication of this difference: in countries with lower levels of female employment, right-wing governments spend more on family policy, whereas in countries with higher female employment, leftwing parties are higher spenders.
Strategic tax expenditures (STEs), next, often have a regressive nature and benefit the middle and upper middle classes the most (Adema et al., 2014;Mettler, 2011). However, much like tax policy in general, STEs can actually be used as a means to very different ends, which makes them subject to influence from political parties and party competition (Howard, 1997). Two examples may serve to illustrate this point.
First, earned income tax credits (EITCs) usually serve to increase incomes and work incentives for low-wage workers. In Sweden, however, an EITC introduced by a center-right government in 2007 was designed to increase incomes for workers across the wage distributionalbeit in a slightly progressive mannerto the (absolute) benefit of high-income earners. Interestingly, in 2016, a phase-out mechanism for high-income earners was introduced by the recently elected left-wing government, which altered the EITC's distributional profile more towards those with lower wages (Cronert & Palme, 2019). Second, several countries use tax expenditures to subsidize childcare (Morel et al., 2016). In some other countries, contrastingly, the same instrument has instead been used for carers with low incomes providing services to elderly relatives (Adema et al., 2014). The distributional effects of these two types of schemes, as well as their effect on female labor force participation, are clearly rather different.
Strategic procurement, finally, is an inherently political tool, as any 'procurement decision reflects a political preference and priority' (Buigues & Sekkat, 2009, p. 44). On a general level, strategic procurement can be used both to promote competition, for instance by helping firms overcome entry barriers to oligopolistic markets, and to distort competition by imposing contract requirements that discriminate against companies of particular types (Buigues & Sekkat, 2009). Most countries studied here now have central-level plans that support green public procurement, procurement from small and medium-sized enterprises, and procurement of innovative goods and services (OECD, 2015, p. 139). Yet, commitment and priorities vary across countries. For instance, the European Commission (2015) reports that for the United Kingdom, France and the Netherlands, between 16 and 30 percent of the analyzed procurement value involved green and/or social procurement, while for Sweden and Austria, these figures were in the range of 1 to 3 percent.

The politics of 're-purposing' and its implications for distributional analysis
The multi-tool nature of the micro-interventionist policies illustrated above is important since it is likely to matter for how they are treated in the political arena, especially when inherited through a transition of power between governments with different distributional preferences. To understand how, it is useful to recollect the three central variables of policymaking identified by Hall (1993): the formulation of policy goals, the choice of policy instruments, and the settings of these instruments.
Using this terminology, the more versatile the distributional properties of an inherited set of policy instruments is, the more likely the incoming politicians will find creative ways to 're-purpose' these instruments by re-calibrating their settings or budgetary weights to better serve their preferred distributional goals, instead of facing a potentially difficult and costly process of dismantling the instruments altogether (cf. the concept of 'conversion' in Thelen, 2004). Contrastingly, in cases where policy instruments are less versatile, a change of government is more likely to result in changes in the level, rather than the structure of government spending (for cross-national evidence on this pattern in ALMP spending, see Cronert, 2018).
The distributional versatility of the micro-interventionist policies illustrated above is also important to keep in mind when considering what role the state may have had in the overall growth of economic inequalities during the observed period and particularly so the lower-class constituencies' relative loss compared to the middle-and upper-class constituencies observed in most countries (Iversen & Soskice, 2019, pp. 22-25). Indeed, their higher versatility imply that these state interventions may not be as clearly linked to economic redistribution as some others, such as traditional cash transfer programs that more consistently benefit low-income households more.
A simple illustration of this point is provided in Figure 2, which plots changes in the Gini index of households' disposable income between the early 1980s and 2012-2015 for 17 countries against changes in government spending (as % of GDP) on micro-interventionist policies (panel a) and traditional social protection transfers (panel b) over the same time period. Whereas a substantial negative correlation (r ¼ À0.37) exists between changes in disposable income inequality and social protection, there is no systematic relationship for changes in microinterventionist policy (r ¼ 0.05).
While, for obvious reasons, these simple correlations say little about the causes of economic inequality, they do caution against assuming that all government spending have equivalent and equalizing effects when analyzing the causes and consequences of such spending.
Thus, what the discussions in this section suggest is that given the long-term shifts in the balance of political power between the lower-, middle-and upperclassesin the electoral arena (Sch€ afer & Streeck, 2013, pp. 13-16) as well as among organized interests that indirectly influence policy-making (Pontusson, 2013) the shift towards micro-interventionism may have contributed to the rise of a state that, while not less interventionist, is less catered to the interests of people in the lower part of the income distribution (for a related argument, see Lindsey & Teles, 2017). Focusing analyses only on aggregate expenditure levels thus comes with the risk of overlooking such important dynamics.

What analytic use is micro-interventionism?
Having introduced the concept of micro-interventionism, let us next discuss what advantages it has over some other concepts that have been used to capture recent trends in economic and social policy.
First, it appears warranted not to describe the policy developments in focus here as outright liberalizing or neo-liberal. This is not to dispute that liberalization has  Table 1 for sources (AT and US disregards parental leave, NZ and LU disregards R&D). For panel (b), the indicator includes public and mandatory private social expenditure in cash (Armingeon et al., 2019) less family cash transfers since these are partly included in panel (a).
found other expressions during the studied periodsuch as in the reduction of tariffs (Lampe & Sharp, 2013), the deregulation of industry and finance (Abiad et al., 2010;Schuster et al., 2013), or the transformation of industrial relations (Baccaro & Howell, 2017). Also, the policies we refer to as micro-interventionist do operate under the assumption that the market economy ought to be the primary source of welfare creation, and they are broadly compatible with liberal principles of competition, free trade, openness, and innovation. Some are even based on key neo-liberal or monetarist ideas, for examplein the case of labor market services and trainingthat by removing search frictions and skill shortages, policy may lower the non-accelerating inflation rate of unemployment (NAIRU).
At the same time, however, the micro-interventionist policies ascribe a key discretionary role to the state that goes well beyond what neo-liberals oftentimes supportsuch as addressing market failures and safeguarding essential market institutionsin the sense that they pave the way for imposing selective distributional goals on the economy through the strategic use of taxing and spending. This important common denominator of these policies is better captured by the term micro-interventionist.
As mentioned above, a number of scholars already have observed the expansion of state interventionism in particular policy fields and a number of concepts have been proposed to describe this development (see Bonoli, 2013). So, why the need for the new concept?
First, some previous characterizations are too narrowly defined to capture that a similar approach to state interventionism is evolving in adjacent policy fields. This is, for instance, the case with accounts that describe the 'activation turn' of labor market policy in the 1990s (Bonoli, 2013;Weishaupt, 2011). In this context, Bonoli (2013) usefully brings together ALMP and work-life balance policies under the joint label 'active social policy', having observed that the philosophies and the policy-making logics that govern the two policy fields are, for the most part, rather similar. In a similar vein, by applying the concept micro-interventionism, we may widen our scope further, having observed (1) the conceptual overlap between ALMP and horizontal industrial policy and (2) the growing use of strategic tax expenditures and procurement for purposes that partly overlap with those of active social policy.
Second, some characterizations arguably do not really capture what is distinctive about the observed policy changes. It appears, for instance, not entirely accurate to speak of a general shift from a 'demand-management' to a 'supply-management' strategy (e.g. Jenson, 2012), considering the sustained commitment to fiscal policy activism (Raess & Pontusson, 2015), and the increasing popularity of both employment subsidies, which serve to directly alter employers' labor demand (Cronert, 2018), and strategic procurement, which explicitly boosts the demand for selected goods and services (Buigues & Sekkat, 2009). Similarly, by speaking of a turn towards 'social investment' (e.g. Jenson, 2012), one risks overlooking that whether the expanding instruments are investment-oriented, and whether they can be seen as 'social', crucially depend on their content and targeting.
Third, as we have seen, the concept also must not preclude a scope for distributional conflict regarding the policies to which it refers. This is arguably a risk with some more depoliticized concepts, such as the 'rationalization' or 'recalibration' of policy (e.g. Pierson, 2001).
Being defined solely based on which types of policy instruments it comprises and their versatile applications, the 'micro-interventionist state' can hopefully serve as a concept that is narrow enough to distinguish this particular function of the state from other, less versatile and less intrusive functions, yet wide enough to cover the full spectrum of broadly accepted discretionary economic policy instruments that partisan policymakers have at their disposal.
What factors may explain the turn to micro-interventionism?
This article's primary purpose is to review trends in economic policy and to highlight the growth and the versatility of micro-interventionist policy tools. Whereas a thorough analysis of the causal factors underlying these trends falls outside the scope of this work, it is incumbent upon me to offer some commentary on the broad constellation of phenomena that accompany the observed developments.
Although our discussions have emphasized the continued role of partisan politics in shaping economic policy, governments do not operate in a vacuum. Thus, to understand the systematic policy changes, we need to also consider a number of important, albeit not determinative, constraints under which governments in the advanced democracies operate (Beramendi et al., 2015;Hay, 1999).
To begin with, international bureaucracies have likely played a role in defining what policy instruments are available for policymakers' consideration. More specifically, it is possible that the observed trends have been facilitated by a growing diversity of recommendations stemming from these international organizations. On the one hand, in several cases these have had a liberalizing impact, such as the WTO with respect to limiting tariffs and subsidies (Aiginger, 2007;Farnsworth, 2012) and the EU with respect to promoting monetarist policy (McNamara, 1998). However, as noted by Jenson (2012) and others, whereas in the 1980s the OECD was a firm supporter of neo-liberal policy prescriptions, by the mid-1990s it hadalongside the EU and even the World Bankchanged track and rediscovered the potential benefits of active state interventions. Since then, the EU and the OECD have published recommendations and guidelines that embrace a wide variety of micro-interventionist policies, including ALMP, work-life balance policy, horizontal industrial policy, and strategic procurement (e.g. European Commission, 1997;OECD, 2016). This is not to suggest that interventionism is the only game left in town in Brussels, Paris, and Washington. The point is simply that the growing diversity of policy orientations endorsed by international experts may alter the way in which these endorsements impact national policy-making. An important mechanism through which such an impact may play out is to legitimize reform by like-minded national politicians. Thus, as the international organizations diversify their policy recommendations, more policymakers may find an 'ideational fit' between their preferred policies and those endorsed by international experts.
Second, to stand a chance of winning office, politicians need to be attentive to the preferences of the electorate. Here, it is worth highlighting that voters are generally favorable towards state intervention in the economy. Reported in Figure 3, longitudinal data from the ISSP's (2018) Role of Government surveys available for a subset of countries show that when asked for their opinion on some things the government might do for the economy, respondents have for long leaned more strongly in favor of than against supporting industrial development and financing the creation of new jobs. By contrast, for the non-interventionist optionsto reduce government spending and regulationa substantial decline in approval can be observed since the mid-1990s. Thus, it is possible that the maintained public demand for active intervention may push office-seeking politicians more towards micro-interventionism as the traditional heterodox policy tools are gradually being removed from the feasible policy menus.
Third, politicians' policy proposals and decisions are likely influenced by pressures from the real economy. For instance, the increasing global trade integration and gradual decline of manufacturing in the advanced democracies (although themselves not independent of politics) are likely to have mattered for both the decline of heterodox policy and the rise of micro-interventionism (Aiginger, 2007;Bonoli, 2013).
Fourth, once in place, most policy programs discussed here generate their own support coalitions who exert pressures that policymakers have to navigate (Pierson, 2001). Along those lines, part of the explanation for the shift from heterodox to micro-interventionist policy may be that the multi-tool nature of the latter makes them attractive to a wider support coalition than some of the more narrowly devised heterodox policies.
Fifth, some theories view the development of some micro-interventionist policies as the result of 'social learning' among policy-makers about how economic and social challenges can successfully be met. For instance, Bonoli (2013, pp. 56-60) has suggested that the convergence on active social policy may be understood in part as the outcome of a long process of experimentation with various imperfect policy options that has eventually resulted in a widely acceptable compromise between the protection policies preferred by the left and the non-interventionist approach preferred by the right. From this perspective, the expansion of micro-interventionism could be seen as representing a 're-embedding' of the international markets in a new domestic social compact and thus as a movement towards a new version of the compromise of 'embedded liberalism' that Ruggie (1994) observed in the OECD countries during the postwar era. Yet, even the postwar compromise, as Ruggie pointed out, in fact accommodated a considerable degree of negotiation between competing interests about the orientation of domestic interventionism. Similarly, if a new equilibrium is emerging based on micro-interventionist 'multi-tools', it clearly has not brought an end to distributive struggles about their specific applications.
Assessing the relative merit of each of these factors in explaining the policy trajectories observed above will be an important endeavor for future research. In any event, any theory devised to explain the rise of micro-interventionism must be able to account for a growing agreement on economic policy instruments as well as a sustained scope for disagreement on their specific applications. However, these two things may well go hand-in-hand. As discussed above, the more constrained and consolidated the overall set of feasible policy instruments is, the greater the incentives of incoming politicians to find creative ways to 're-purpose' these policies to their advantage.

Conclusion: towards a Swiss Army Knife State?
Several scholars before me (e.g. Bonoli, 2013;Hay, 1999;Mazzucato, 2013;Rodrik, 2014;Vail, 2018), have observed policy trends within particular countries or economic policy fields that contradict the notion that the economies of the advanced democracies have become insulated from discretionary interventions by the democratic state (cf., Flinders & Wood, 2014;Streeck, 2014). Synthesizing these customarily separate strands of research, this article has presented a systematic long-term review of economic policy trends in eighteen advanced democracies over the past four decades. It has shown that since the early 1980s, in line with non-interventionist, neo-liberal prescriptions, policymakers have largely abandoned the intrusive heterodox economic policy interventions that were used to tackle social and economic challenges during the early post-war erasuch as trade barriers, regulation and subsidization of industry, labor shedding, and depreciating currency interventions.
However, this has not translated into a full retreat of economic interventionism. On the contrary, policymakers have gradually developed a new approach to economic policyhere labelled micro-interventionistbased on horizontal industrial policy, active labor market policy, work-life balance policy, strategic tax expenditures and strategic procurement, which breaks with non-interventionist or neo-liberal prescriptions by strategically using taxing and spending to impose selective distributional goals on the economy. The cross-partisan appeal of the 'multi-tools' associated with this approach partly stems from their versatility, as policymakers canand douse them to very different distributional ends.
While still the traditional social protection functions such as pensions and health make up a larger portion of overall government expenditure, micro-interventionist instruments have expanded across the advanced democracies, and by force of their multi-tool nature they are likely to become an increasingly important arena for distributional conflict. Indeed, if it is generally true that 'new policies produce new politics' (Schattschneider, 1935, p. 288), then perhaps we should also expect that more and more-sophisticated policies produce more and more-sophisticated politics.
Let us conclude by highlighting some important implications of the arguments in this article for international and comparative political economy research.
To begin, as regards economic policy developments in the advanced democracies as a whole it does not seem very fruitful to speak of a paradigmatic shift towards non-interventionism, as it appears that policies with markedly different ideational foundations currently cohabitate. Instead, scholars should recognize that democratically elected governments, although facing various constraints, are still neither unwilling nor unable to use economic policy instruments discretionarily to influence the distributional outcomes of economic activitiesand that furthermore these instruments can serve very different distributional purposes. The last point in turn has two implications. First, scholars concentrating on a particular policy field may do well to keep an eye out for functionally similar policies developing in adjacent fields. Second, the distributional versatility of contemporary economic policy tools generally needs to be taken more seriously in both theoretical and empirical work on the causes and consequences of economic policy.
Theoretically, this means refraining from assuming that discretionary state interventions are inherently redistributive or equalizing, and that contemporary political conflicts stand simply between more and less interventionism. Instead, scholars should approach any policy tool with an inquiring mind, thinking about how it might be re-purposed, rather than revoked, by creative partisan policymakers to serve different distributional goals.
Empirically, it means that scholars need much more detailed data than the indicators on aggregate public expenditure that were used in the analysis above and that still dominate cross-national comparisons of economic policy. Ideally, such data track policy programs and their settings over time and indicate occasions of birth, reform, or termination. At least for analyses of European countries, a number of rich databases fairly recently established by the European Commission, which track individual cases of state aid to industries, labor market programs, and public procurement notices, appear particularly useful in that regard, but they have been only sparsely exploited to date.
In conclusion, taking these insights to heart appears fruitful for improving our understanding of both the distributional consequences of contemporary economic policy and the changing role of different political actors in the policy-making process. Notes 1. For example, tariffs, regulations, and fiscal policy interventions may be more or less broadly applied; when implementing quantitative easing, central banks may be more or less selective regarding the types of bonds they purchase; etc. 2. Even if we omit spending on care for old age and incapacity-the largest and arguably least interventionist policy-the upward pattern is clear. Although the Netherlands and Sweden decreased 5-6 percent from high initial levels, the other countries saw increases between 44 and 270 percent.
Svensson, Kathleen Thelen, and Pernilla Tunberger, as well as participants at the 2019 APSA annual meeting, three anonymous reviewers, and the editorial board of the Review of International Political Economy for helpful guidance and valuable comments on earlier versions of this article.

Disclosure statement
No potential competing interest was reported by the author(s).