4 June 2020
The COVID-19 pandemic is a significant de-globalisation shock at a time when globalisation was already under challenge from President Trump’s trade war. Post-pandemic public policy responses can either compound or alleviate this shock.
A push for greater economic sovereignty aimed at increasing resilience with respect to foreign shocks would continue recent trends which hinder globalisation. At the same time, policymakers around the world are keen to re-boot their economies’ productive potential to recover from the global downturn.
This report shows re-establishing international connectedness can help boost productivity in ways which are consistent with increased resilience to international shocks. The report extends my previous Failure to Converge? publication on the Australian-US productivity gap by examining the relationship between measures of globalisation and productivity across the Organisation for Economic Cooperation and Development (OECD) economies for the period 1970 to 2017.
Importantly, the measure of globalisation used rewards greater diversity among trade and treaty partners so it is consistent with the objective of increasing resilience to foreign shocks and supply chain disruptions. For OECD countries, a one per cent increase in a broad measure of globalisation raises labour productivity by 0.85 per cent, while a one per cent increase in economic globalisation raises productivity by 0.5 per cent in the long-run. Productivity benefits from the social and political dimensions of globalisation, as well as the economic.
The report also finds a negative long-run relationship between the government share of consumption spending and labour productivity. The expansion in state intervention in domestic economies due to the pandemic will also need to be wound back to support post-pandemic productivity growth. A key lesson from the pandemic is that international connectedness need not be a source of vulnerability or lack of resilience. The countries which showed the most effective response to COVID-19 were highly globalised and urbanised economies like Singapore and Taiwan.
The findings of the report can be used to measure the cost to labour productivity and living standards from the COVID-19 de-globalisation shock, as well as the costs and benefits of post-pandemic policy responses aimed at either decreasing or increasing international connectedness. Coordinated diversification of trading partners within the US alliance network can increase resilience to international shocks and reduce dependence on individual countries as a source of supply.
Stockpiling of critical goods in conjunction with alliance partners can go a long way to enhancing resilience, without having to develop domestic production capabilities in which Australia is unlikely to enjoy a comparative advantage or economies of scale. Businesses are likely to increasingly factor security of supply issues into their decision-making about where to locate production facilities and other operations.
The COVID-19 pandemic is a significant de-globalisation shock at a time when globalisation was already under challenge from President Trump’s trade war. Post-pandemic policy responses can either compound or alleviate this shock. A push for greater economic sovereignty aimed at increasing resilience with respect to foreign shocks would continue recent trends which hinder globalisation. At the same time, policymakers around the world are keen to re-boot their economies’ productive potential to recover from the global downturn. This report shows re-establishing international connectedness can help boost productivity in ways that are consistent with increased resilience to international shocks.
Prime Minister Scott Morrison has flagged a post-pandemic effort to boost Australia’s ‘economic sovereignty,’ having previously criticised what he called ‘negative globalism.’[^1] Industry Minister Karen Andrews highlighted manufacturing capabilities as a focus for these efforts. Even in a large and relatively closed industrial economy like the United States, the pandemic exposed a dependence on foreign sources for medical and other supplies. While COVID-19 has exposed potential vulnerabilities to shocks through international connectedness, it has also dramatised the economic benefits of globalisation. As borders closed, economies turned down and jobs were lost in a way unprecedented since the Great Depression of the 1930s.
The Morrison Government points out Australia’s future prosperity will still depend on its economic integration with the rest of the world. His ‘negative globalism’ speech also noted ‘our interests are not served by isolationism and protectionism.[^2] The Australian Government recently joined regional partners in affirming the importance of maintaining open and connected supply chains.[^3] Recent opinion polls find 70 per cent Australians say globalisation is mostly good for Australia, unchanged from 2019.[^4] A retreat behind protectionist barriers looks unlikely, in Australia at least. The pandemic has, however, reinforced the Trump Administration’s already strongly protectionist instincts.
The government has indicated it does not intend to retreat behind a tariff wall or handout subsidies to business. However, this begs the question of how greater economic sovereignty and self-sufficiency can be achieved if market forces do not already lend themselves to domestic production or a greater diversity in offshore sources of supply. Policymakers still need to be wary of the temptations of old-fashioned industry policy, which can undermine both the economy and national security. Australia’s replacement submarine project shows the dangers of allowing domestic industry and employment policy to skew national security priorities.
Rather than economic sovereignty, a more appropriate lens for post-pandemic public policy is economic resilience. This does not require economic self-sufficiency or re-shoring and may well be achieved through even greater diversification among international trading partners. While the global nature of the pandemic caught the world short, domestic and international manufacturers quickly pivoted to address shortfalls. Just as the pandemic does not respect international borders, supply responses should not be ring-fenced.
The author’s previous research shows Australian productivity and living standards do best when Australia is better connected with an increasingly integrated world economy.[^5] While this global connectedness necessarily increases our exposure to foreign shocks, the economic gains from globalisation also afford us the means to better address these shocks. Australia’s fiscal response to the pandemic, for example, was the product of decades of economic growth contributing to a relatively low net debt position, increasing the fiscal capacity to respond. The pursuit of greater economic sovereignty will not enhance resilience if it comes at the expense of international connectedness, productivity and living standards. The promotion of economic sovereignty is also potentially at odds with the government’s post-pandemic reform agenda designed to re-boot productivity.
This report extends our previous Failure to Converge? report on the Australian-US productivity gap by examining the relationship between measures of globalisation and productivity across the Organisation for Economic Cooperation and Development (OECD) economies for the period 1970 to 2017. It shows Australian productivity is not unique in benefiting from increased globalisation. These gains can be found across the OECD economies. The concept of globalisation informing these measures is defined as follows:
the process of creating networks of connections among actors at intra- or multi-continental distances, mediated through a variety of flows including people, information and ideas, capital, and goods. Globalisation is a process that erodes national boundaries, integrates national economies, cultures, technologies and governance, and produces complex relations of mutual interdependence.[^6]
This definition is distinct from concepts such as internationalisation, liberalisation, universalisation or Westernisation with which globalisation is often conflated or confused. The measure of globalisation used is comprehensive in scope and has been shown to have explanatory power in a wide range of other empirical applications.[^7] Importantly, it rewards greater diversity among trade and treaty partners so globalisation is measured in a way which is consistent with the objective of increasing resilience to foreign shocks and supply chain disruptions. Globalisation, productivity and resilience to international shocks need not be in tension.
This analysis is particularly well-suited to evaluating the implications of structural change through greater openness for productivity and living standards. The key result is that, for OECD countries, a one per cent increase in a broad measure of globalisation raises labour productivity by 0.85 per cent, while a one per cent increase in economic globalisation raises productivity by 0.5 per cent. Productivity benefits from the social and political dimensions of globalisation, as well as the economic. This extends the results from a previous report, which also found a positive, although quantitatively smaller, relationship between globalisation and labour productivity for Australia.[^8] The model presented in this report also shows a negative long-run relationship between the government share of consumption spending and labour productivity.
These results have important implications for the policy response to the COVID-19 pandemic. A key lesson from the pandemic is that international connectedness need not be a source of vulnerability or lack of resilience. The countries which showed the greatest resilience to COVID-19 were highly globalised and urbanised economies like Singapore and Taiwan. While globalisation and urbanisation certainly increased their potential exposure to the pandemic, their status as high-income economies with relatively sound governance gave them the capacity to better marshal resources and state capacity to manage the pandemic. Yet this high-income status is itself a function of globalisation. Calls for greater economic sovereignty are likely to reduce productivity and living standards to the extent that they damage economic, social and political integration with the rest of the world.
Another implication of these results is the expansion in state intervention in domestic economies due to the pandemic will also need to be wound back to support post-pandemic productivity growth. While these interventions were necessary for the short-run, they will need to be wound back in the long-run to avoid negative effects on productivity growth and living standards.
The findings of the report can be used to measure the cost to labour productivity and living standards from the COVID-19 de-globalisation shock, as well as the costs and benefits of post-pandemic policy responses aimed at either decreasing or increasing international connectedness. An appreciation of these costs and benefits is essential to crafting policy responses consistent with greater economic resilience as well as economic recovery.
The report also makes policy recommendations for post-pandemic recovery and resilience. Coordinated diversification of trading partners within the US alliance network can increase resilience to international shocks and reduce dependence on individual countries as a source of supply. David Uren’s previous United States Studies Centre report makes recommendations as to how this can be achieved for critical minerals.[^9] Stockpiling of critical goods in conjunction with alliance partners can go a long way to enhancing resilience, without having to develop domestic production capabilities in which Australia is unlikely to enjoy a comparative advantage or economies of scale.
Globalisation has suffered two significant setbacks in just over decade, reversing the trend toward greater economic, social and political integration seen prior to 2008. The global financial crisis of 2008 saw a levelling out in the pace globalisation along a number of dimensions, most notably cross-border capital flows. More recently, the COVID-19 pandemic has caused an unprecedented economic downturn and a closing of borders to international people flows. However, these major shocks are not the only factors at work.
At the same time, globalisation has flatlined, productivity growth around the world has slowed. There is considerable debate over what is causing slower productivity growth, with little consensus on this question among the economics profession.[^10] However, as Hufbauer and Lu note, “the coincidence of the two stagnations is too great to ignore.”[^11] Weaker globalisation has led to a loss of economic dynamism which has reduced productivity growth across the OECD member economies.
Previous United States Studies Centre reports have highlighted the simultaneous slowdown in measures of globalisation and productivity growth since the financial crisis of 2008.[^12] Slower globalisation explains slower productivity growth in ways that competing explanations do not. Most of the alternative explanations for slower productivity growth come up at least somewhat short as a complete explanation.[^13]
Later in this report examines the relationship between the levels of globalisation and labour productivity. Increased openness gives rise to changes in productivity that have level effects, even if they do not have ongoing growth effects. While there are limits to globalisation and the contribution it can make to productivity, most countries are likely to have unexploited potential productivity gains from increased international connectedness. Policymakers enjoy at least some leverage over the openness of their economies by lowering policy-related barriers to trade in goods, services, capital, labour and ideas, although deeper structural factors like geography may be less amenable to changes in policy.
The potential causal effect of slower globalisation on productivity growth only serves to push the question about the underlying causes of slower productivity back one step to the underlying causes of slower globalisation. Large global shocks like the 2008 financial crisis and the COVID-19 pandemic can account for slower globalisation, at least in the short-run. However, in the long-run, these shocks may not be persistent enough to explain a permanent slowing in the pace of globalisation and productivity growth.
Hufbauer and Lu suggest two related candidates to explain the slower pace of globalisation and productivity which may make these one-off shocks more persistent. First, the increase of creeping or micro-protectionism weighed on international trade and investment even before the onset of President Trump’s trade war. Simon Evenett’s Global Trade Alert database has chronicled the rise in micro and other trade barriers since 2009.[^14] Second, the absence of significant multilateral trade and other forms of economic liberalisation since around the mid-2000s. The trade war from January 2018 led to a further downturn in global trade even before the onset of the COVID-19 shock by disrupting global supply chains and increasing economic policy uncertainty.[^15] While growing protectionism could, in theory, be a response to weaker economic conditions, President Trump initiated his trade war against the backdrop of a strong economy. All of these factors could be expected to weigh on both globalisation and productivity.
The relationship between economic openness and productivity has long been of interest to economists. There is considerable empirical evidence for the proposition that economic openness is conducive to growth in both productivity and living standards. However, the robustness and the direction of causality underlying these results has often been called into question. The main issues that arise in the literature are how to reliably measure openness and productivity and how to estimate the relationship between them in a cross-country setting. There is also the issue of causality: does increased globalisation drive higher productivity or does higher productivity lead to greater economic openness? Bilateral causality is also possible.
Globalisation could be expected to benefit productivity in several ways.[^16] International trade increases the size of markets, leading to efficiency gains from economies of scale and increased specialisation. It also facilitates the international transfer of knowledge and human capital. Countries that are more open have a greater ability to absorb technological and other innovations from countries on the global productivity frontier such as the United States.[^17] At the firm level, there may be trade-induced productivity gains as firms learn through exporting. Immigration is also increasingly recognised for its contribution to innovation and productivity.[^18]
Well-established literature starting in the early 1990s examines the role of economic openness in explaining economic growth based on cross-country growth regressions — Frankel and Romer’s 1999 work is a good example.[^19] Investigating the relationship between economic openness and productivity was a natural extension of this research, with the effect of openness on productivity an obvious candidate for the transmission mechanism from increased openness to higher incomes. Much of this literature focuses on explaining total or multifactor productivity (T/MFP). However, cross-country comparisons of MFP are fraught due to problems in measuring capital stocks and the flow of capital services in a consistent way across countries. This compounds problems with measuring trade openness. Rodrik summarises some of the problems with the early literature.[^20]
Edwards finds that more open economies experience faster productivity growth based on a sample of 93 countries, although concedes his research leaves the issue of causality “somewhat open and will require time series analysis to address.”[^21] Alcala and Ciccone find a causal effect of trade on productivity alongside country-size, with both trade and scale effects working their way through to labour productivity via MFP growth (labour productivity can be decomposed into contributions from MFP and capital deepening).[^22] Henry and Milner review much of the relevant theory and evidence through to the mid-2000s, with mixed results.[^23] Irwin and Tervio find trade is not a significant determinant of productivity, controlling for geography and measures of institutional quality, highlighting the importance of other factors.[^24]
Overall, the literature supports the claim that the various elements of globalisation are positive for productivity, although the robustness of this claim has often been called into question due to measurement issues and the difficulty of testing the direction of causality.
The positive cross-country relationship between globalisation and labour productivity in 2017 highlighted in my Failure to Converge? report is shown in Figure 1.
Figure 1. KOF Globalisation Index and labour productivity, 2017
As well as the cross-sectional relationship, this report is also interested in how this relationship evolves over time while controlling for other variables. This report proposes a methodology for testing the average relationship between globalisation and labour productivity across the OECD economies over time conditional on other economic variables which addresses several of the methodological problems arising in the literature already discussed.
The first issue concerns the measurement of economic openness. Previous literature has often relied on measures which are too narrow in scope, such as the traded goods share of GDP. Here I use the KOF Globalisation Index,[^25] which is both comprehensive in scope and has been shown to have considerable explanatory power in a wide range of empirical applications.[^26] It measures economic, political and social dimensions of globalisation and captures both de jure and de facto measures. However, it is likely the measure also proxies for institutional quality given openness is often associated with better governance. Bad governance often does not survive exposure to international competition. The KOF measure also includes an index of human capital accumulation, an important contributor to productivity. Human capital is often traded through education exports and immigration; so it properly belongs in a measure of globalisation. The KOF Globalisation Index may, however, confound the influence of institutions and human capital accumulation with that of openness per se.
The second issue concerns the measure of productivity. Given the problems in making reliable cross-country comparisons of MFP, this report uses labour productivity instead. However, by controlling for the contribution to labour productivity made by the per capita capital stock, the resulting estimates have a similar, although not identical, interpretation to MFP, subject to the caveat about how well capital deepening is measured.
The third issue concerns the direction of causality. The combined time series and cross-section approach taken here allows for causality testing, although only in the narrow sense that past values of globalisation can be shown to predict future labour productivity. This is a standard approach to testing causality in the absence of natural experiments, good instrumental variables or other identifying information for teasing out causal relationships.
The fourth issue concerns the choice of other conditioning variables. The model estimated below includes the per capita capital stock and labour utilisation rate as conditioning variables, along with the government share of consumption. The KOF Globalisation Index accounts for the size of a country by dividing variables by GDP or population size but does not otherwise control for geography.
The fifth issue concerns heterogeneity across the OECD economies. The estimation method is robust to outliers and allows for heterogeneity in the short-run relationships across countries while constraining the long-run response to be the same. The intuition behind this approach is whereas globalisation could be expected to have different dynamic effects on productivity across countries in the short-run, in the long-run, these effects should be more homogeneous. The method is also robust to the order of integration of the data and choice of dynamic specification compared to some alternatives.
The panel consists of 36 OECD economies for the period 1970 to 2017. The maximum number of time series observations (T=47) is greater than the size of the cross-section (N=36). The panel is unbalanced in that not all cross-sections have an equal number of observations due to limited data availability for some OECD economies, but this is not an issue for the estimation methodology. The variables in the panel are as follows. All variables are included in the model in log form, with the exception of the government share of consumption spending:
Labour productivity (lp): GDP per hour worked, current price, current PPP, US dollars. Source: OECD.stat.
Labour utilisation (lu): hours worked per head of population. Source: OECD.stat.
Capital stock per capita (kpc): capital stock at current PPPs (in 2011 US dollars) divided by population in millions. Source: Penn World Tables 9.1.
KOF Globalisation Index (kofgi): Headline country index combining economic, social and political dimensions of globalisation and de jure and de facto measures. Source: KOF Swiss Economic Institute.
Government share of total consumption (g): current PPPs, per cent. Source: Penn World Tables 9.1
A fundamental issue concerns whether the relationship between globalisation and productivity is causal and, if so, in what direction. Economic theory does not provide strong guidance and, in principle at least, causality could run in either or both directions.
A standard approach to testing causality among macroeconomic time series is to ask whether one series effectively forecasts another. In this case, one wants to ask whether past values of globalisation predict future values of productivity or vice versa. This gives greater confidence there is a causal relationship between them, although it is not a definitive test because it tests causality only in a very narrow sense.
There are two basic approaches to testing for causality in a panel setting. The first approach assumes all cross-section coefficients are the same (stacked common coefficients). The second approach, due to Dumitrescu and Hurlin (DH) (2012), makes the opposite assumption by allowing all coefficients to be different across cross-sections.[^27] The first approach makes a stronger assumption than the second, and the second method is arguably better suited to a heterogeneous panel of countries. The test statistics for the null hypothesis of Granger non-causality for both approaches are shown in Table 1.
Table 1. Panel Granger non-causality tests
The first method accepts the null hypothesis that labour productivity does not cause globalisation, but is consistent with globalisation being causal for labour productivity. The second approach rejects the null in both cases, implying bilateral causality. Both tests are thus consistent with causality running from globalisation to labour productivity. The following modelling approach assumes causality running from globalisation to labour productivity and produces estimates consistent with this assumption, while also controlling for other variables.
Table 1 also includes non-causality tests for the relationship between the government share of consumption spending and labour productivity. Non-causality in both directions is accepted by the common coefficients approach. The DH approach points to a causal relationship running from labour productivity to the government consumption share, suggesting this relationship may be cyclical. However, this does not test the long-run relationship and does not control for other variables.
The pooled mean group estimation method for dynamic heterogeneous panels developed by Shin, Pesaran and Smith[^28] is well-suited to estimating long-run relationships given a relatively large number of time series observations on a smaller number of cross-sections. This approach adapts the error correction form of the single equation autoregressive distributed lag (ARDL) model used in a previous report[^29] to a panel setting by allowing the short-run coefficients and error correction (EC) terms of the model to vary across countries while constraining the long-run responses to be the same. The cross-section short-run coefficients for individual economies may still be of interest and are reported below for Australia and the United States, but not for other OECD economies.
The dynamic specification is based on minimising the Schwarz criterion for all possible models with a maximum lag for both the dependent and explanatory variables of up to two years. The dynamic specification chosen is a one-year lag on each of the variables. A constant term is also included.
Cross-section dependence and unit root tests are shown in Appendix 1. They confirm none of the variables are second-difference stationary which would preclude an ARDL specification. The estimation method is robust to the presence of stationary variables and those with a unit root.
The estimated model for labour productivity (lp) in the OECD is shown in Table 2.
Table 2. Labour productivity in the OECD
The long-run equation shows on average across the OECD, a one per cent increase in the KOF Globalisation Index raises labour productivity by nearly 0.9 per cent. Re-estimating the above model with the economic component of the index results in a smaller effect of 0.5 per cent and is less precisely estimated, while leaving the coefficients on other variables largely unchanged. This suggests the political and social components of the overall index are quantitatively and statistically important for labour productivity. The estimated long-run effect of economic globalisation on labour productivity for the OECD as a group is somewhat greater than the 0.3 per cent response found for Australia in the United States Studies Centre’s Failure to Converge? report.
The per capita capital stock and labour utilisation rate have coefficients with the theoretically correct sign and plausible magnitudes. The government share of consumption spending has a large negative effect on labour productivity. This could be due to cyclical effects on both variables, as suggested by the previous causality tests, but given this is a long-run equilibrium relationship, it is less likely to be the result of cyclical effects, which are better captured by the short-run equation, where the effect is also negative.
The mean short-run equation is of interest mainly for the error correction term, which is correctly signed and statistically significant. Globalisation is not economically or statistically significant as a determinant of labour productivity in the short-run, suggesting a heterogeneous response across the panel. The other variables have short-run responses which are correctly signed and of a plausible magnitude.
The model allows us to estimate individual cross-section coefficients for the short-run equation. In Table 3, I show the estimated short-run equations for Australia and the United States.
Table 3. Short-run equations for Australia and the United States
For Australia, the most notable feature is the negative short-run effect of globalisation on labour productivity. This may be attributable to economic liberalisation having negative short-run effects on labour productivity growth as resources are reallocated before the economy moves closer to the efficient frontier. The change in the government share of consumption spending does not have a statistically significant effect on labour productivity in the short-run, although the estimated magnitude of the negative effect is large.
For the United States, globalisation is not economically or statistically significant in the short-run. As noted in a previous report, because the United States is already on the frontier of global productivity and is a relatively closed economy, it is not as dependent on openness to the rest of the world for its productivity.[^30] The government share of consumption spending in the United States is not statistically significant in the short-run.
The results provide a basis for estimating the effects of the COVID-19 pandemic on labour productivity and living standards across the OECD. The loss of global connectedness due to the pandemic will subtract from labour productivity and income. If globalisation in the advanced economies fell to the level seen in 1970, then labour productivity in the OECD would be around 24 per cent lower, with a similar effect on average incomes, all else equal.[^31] While this is an extreme scenario, it highlights the potential damage from the pandemic and pursuit of economic sovereignty at the expense of global ties. The persistence of any effect of the COVID-19 de-globalisation shock will depend on how quickly OECD economies can restore their previous levels of global integration and then build on that connectedness. In the case of international people flows, the shock may persist for a year or more.
As already noted, the measure of globalisation used in this report rewards a diversity rather than a concentration of trade and treaty partners. Because globalisation measured in this way is good for labour productivity; supply chain diversification does not need to come at the expense of economic efficiency.
President Trump’s trade war from 2018 led to an increased focus on potential supply chain vulnerabilities and disruptions on the part of both business and government even before the COVID-19 pandemic. Global supply chains have already shifted from China to places like Vietnam and Mexico, a phenomenon likely to be accelerated in the wake of the pandemic.
The pandemic has dramatised that Australia is exposed on both the export and import front, despite having relatively low levels of economic globalisation for a country of its size. A recent Henry Jackson Society study found Australia was the most strategically dependent of the ‘Five Eyes’ security partners on imports from China.[^32] Yet the pandemic has also shown how the industrial sectors of the United States and its allies can quickly pivot in response to an emergency and address shortfalls in critical supplies. The combined industrial strength of the US alliance network far exceeds that of China. Coordinating diversification of supply chains for critical goods within the alliance network, without overly distorting private sector decision-making, is a key challenge for policymakers. However, even without prompting from public policy, business is likely to increasingly factor security of supply issues into their decision-making about where to locate production facilities and other operations.
A once in a century global shock like COVID-19 is bound to lead to temporary shortages and expose economic vulnerabilities and interdependencies. However, given the uneven and unpredictable path of the pandemic, global free trade still proved the most effective way of getting medical and other supplies to where they were needed, when they were needed. Export controls only exacerbate shortages, especially for developing countries which may lack industrial capacity and be completely dependent on imports, but where the human need for medical supplies is no less pressing. Pandemic preparedness in the United States was particularly harmed by President Trump’s trade war, which imposed punitive tariffs on medical equipment, causing a de-stocking of critical supplies.[^33]
Pandemic preparedness does not require economic self-sufficiency. It is always possible to stockpile essential goods as a precaution against such events. After the SARS outbreak in 2003, Australia stockpiled masks, anti-virals and other supplies, a stockpile drawn down to address the COVID-19 pandemic. Complex medical devices such as ventilators are made from hundreds of individual parts which are typically sourced from a wide range of suppliers located in a number of countries. To produce all these parts domestically would be costly and only serve to further stretch the pandemic preparedness budget.
Stockpiling can also address other potential vulnerabilities. Australia’s dependence on imports of fuel and its failure to meet the International Energy Agency’s requirement to hold oil stocks equivalent to at least 90 days of the prior year’s daily net imports has been recognised an issue by Australian policymakers. Historically, commercial stocks were enough to meet this requirement, but this has not been the case since 2012.
Much of Australia’s fuel inventory is effectively held as stock on the water already on its way to Australia, but this is not counted towards IEA obligations. Stock on the water or located overseas is vulnerable to the same potential disruptions which motivate stockpiling in the first place. Stock held offshore, either in the water or as part of the US Strategic Petroleum Reserve, is not a perfect substitute for stock held onshore when it comes to energy security. For now, however, the world is awash with unwanted oil given the collapse in global demand, making this a less pressing issue, but will require more attention from policymakers over the longer term.
Globalisation, as measured by the KOF Economic Globalisation Index, is shown to be an important long-run determinant of labour productivity for the OECD group of countries. On average, a one per cent gain in the index leads to a 0.85 per cent gain in labour productivity while controlling for the per capita capital stock, the labour utilisation rate and the government share of consumption spending. The contribution of human capital accumulation to labour productivity is captured in the measure of globalisation, reflecting the contribution of international people flows to productivity. Given this set of controls, the estimated effect of globalisation on labour productivity approximates the contribution to multi-factor productivity, although cross-country comparisons are made problematic by issues in measuring the capital stock and the flow of capital services.
Statistical tests are consistent with causality running from globalisation to labour productivity, although there is also statistical support for bilateral causality. It is likely globalisation proxies in part for institutional quality, because poor institutions typically do not perform well when exposed to international competition. While institutional quality, globalisation and labour productivity can also be viewed as being jointly determined by deeper underlying factors such as history and culture, globalisation is likely to be a critical element in this process. Policymakers generally have considerable leverage over economic openness through the choices they make about how they discriminate against or facilitate the cross-border movements of goods, services, capital, people and ideas. These choices can be expected to have at least one-off effects on productivity growth and thus the level of labour productivity, but there may also be ongoing growth effects from increased economic dynamism from increased openness.
The post-pandemic debate about how to build greater resilience to shocks has highlighted global interconnectedness as a potential source of economic vulnerability. However, it is notable that countries with high levels of global economic integration and high rates of urbanisation, such as Singapore and Taiwan, are widely considered to have implemented more effective responses to the pandemic. The greater vulnerability comes not from international connectedness, but poor governance and low incomes. Globalisation is associated with higher incomes due to its effects on productivity and through increased international disciplines on domestic institutions leading to better governance. The increased focus on economic sovereignty in the wake of the pandemic should not come at the expense of international connectedness, as this will tend to weaken productivity, incomes and governance reducing rather than enhancing resilience.
The pandemic has also shown how the industrial sectors of the United States and its allies can quickly pivot in response to an emergency and address shortfalls in critical supplies. The combined industrial strength of the US alliance network far exceeds that of China. Coordinating diversification of supply chains for critical goods within the alliance network, without overly distorting private sector decision-making, is a key challenge for policymakers. However, even without prompting from public policy, business is likely to increasingly factor security of supply issues into their decision-making about where to locate production facilities and other operations.
With a panel of OECD economies, it is likely changes to a variable in any of the panel countries will affect other countries as well due to common shocks or unobserved components. This cross-section dependence will affect the choice of unit root testing strategy. Given the number of time series observations is greater than the number of cross-sections, the relevant test statistics are the Breusch-Pagan LM and Pesaran scaled LM tests. The results of these tests are shown in Table A1. The test statistics reject the null of no-cross section dependence, which is to be expected in a panel of this type.
Table A1. Cross-section dependence tests. H0: no cross-section dependence
Given the existence of cross-section dependence, the unit root testing methodology needs to allow for this dependence. While the estimated ARDL model is robust to ambiguous orders of integration and the presence of I(0) and I(1) variables, it is necessary to confirm that none of the variables are second-difference stationary which would preclude estimating a model of this type. Table A2 reports Im, Pesaran and Shin test statistics for the null hypothesis of an individual unit root process, confirming the panel variables are stationary in levels and first-differences.
Table A2. Unit root tests. H0: unit root