August 23, 2016

High–Stakes Exams: The A–Level Results are Out, but the Debate Goes On

Writing about web page

On 29 June the Social Market Foundation published our CAGE policy briefing, "The long term consequences of having a bad day: How high-stakes exams mismeasure potential," by Victor Lavy (Warwick and Hebrew University of Jerusalem), Avraham Ebenstein (Hebrew University of Jerusalem), and Sefi Roth (Royal Holloway University of London). Below is their summary of findings; the full version is available here:

Examination candidates can have a bad day for many reasons unrelated to knowledge, skill, or cognitive ability. Possible causes include minor infection, migraine, hay fever, menstruation, disturbed sleep, and atmospheric pollution. When the stakes are high, as with exams used to rank students or admit them to further training or employment, there can be permanent consequences for the individual and for society.

There is a lack of evidence on these consequences. Empirical challenges include the difficulty in identifying the return to cognitive ability separately from the return to doing well on the examination.

Our solution to this problem is to examine the consequences of fluctuations in a random factor on exam performance. We use fluctuations in air pollution for this purpose. When the same student takes multiple exams, and exposure to ambient pollution varies randomly from day to day, we can use the associated variation in performance to measure the component of a student’s score which is related entirely to luck. The results show that transitory ambient pollution exposure is associated with a significant decline in student performance.

We then examine these students during adulthood (8-10 years after the exams) and we find that pollution exposure during exams causes lasting damage to post-secondary educational attainment and earnings later in life.

Our analysis highlights that high-stakes exams provide measures of student quality that may be imprecise and misleading. These measures can lead to allocative inefficiency because students who have had a bad day because of factors outside their control are mis-ranked. After that, they are inefficiently assigned to further training and to different occupations and this reduces labour productivity overall.

As well as illustrating these problems with high-stakes exams, our findings also expand understanding of the costs of pollution. They imply that a narrow focus on traditional health outcomes, such as hospitalisation and increased mortality, is not enough. The full cost of pollution includes loss of mental acuity, which is essential to productivity in most professions. The use of high-stakes exams to measure mental acuity then multiplies this loss over many years.

After that, Simon Lebus of Cambridge Assessment published a critical comment, "High stakes in adverse conditions." He pointed out that exams are designed to discriminate among candidates, for example by setting time constraints. He questioned "whether it is meaningful to suggest that the 'fairness' of an exam (in any event always a form of sampling) is compromised by random external factors any more than by the peculiarities of its own design." Below are his concluding sentences; you can read the full version here.

If you really want variation to play a role then go with 'local' assessment in schools – where a much greater range of variables is likely to apply, ranging from the physical environment of the classroom (noise, light, discipline and so on) to the nature of the tasks, the facilities, the 'fairness' of the teacher's judgement, possible support from the teacher and so on. Remedies to exclude such variation range from the expensive to the draconian to the impossible. If we were to try to mitigate every external variable (and not just pollution) we would end up chasing myriad effects and influences, with no necessary improvement in the predictive validity of qualifications – which currently are at levels which enjoy public confidence.

Despite legitimate questions about the design and role of high stakes assessment, it is important to recognise that trying to design assessments around the principle of excluding every element of randomness would (perhaps counter-intuitively) end up likely introducing even more randomness, with a consequent adverse impact on both equity and attainment.

Below, Victor and Lavy and his co-authors reply to Simon Lebus, in a comment that appeared first on the SMF blog on 19 August.

Dear Simon,

Thank you for your thoughtful comments. You raise three key issues that are critical to high-stakes testing, and related to our research findings. The first issue relates to exactly what disturbances should or should not a student be protected from. For example, in your example of the concerned grandmother, she perceived the time constraint as an unjust factor that prevented her from demonstrating her actual knowledge. In your opinion (and in ours), exams with time constraints are a legitimate way to distinguish weaker or stronger students, since the speed with which you can perform a task is a reflection of aptitude. However, we would submit that sensitivity to pollution (or lack thereof), is not a factor that should not be used to rank students. As our findings demonstrate, pollution targets asthmatic students and other students with health ailments, and it seems likely that the optimal tests would be designed to minimize the influence of this type of external factor.

A second issue worth noting is that our study finds that temporary disturbances like pollution have a very large impact on student performance. If students were relatively insensitive to pollution, there would be little need to worry about the testing conditions. However, we find that pollution has a very large impact on performance, and this is observed even when exploiting variation across the same student’s exams across different days – and so it is likely that we are identifying a causal association, rather than simply observing a correlation between scores and pollution. As you suggest, it is likely that noise and temperature affect student performance as well, and all these factors should in an ideal world be eliminated from the testing environment. Our findings suggest that efforts to eliminate these factors are not wasted.

The third issue our study raises is a question of fairness. If pollution were truly random and all students had an equal chance of being exposed, it is likely that it could be justly ignored by the testing administration. However, the parts of cities that are most exposed to pollution are generally the poorest areas of cities which are near factories and other dis-amenities. And so, it is likely underprivileged students who are most likely to be affected. Therefore, to ignore the potential impact of pollution on test-takers is to open up the possibility of stacking the deck further against students from more modest backgrounds.

In light of these factors, we submit that high stakes exams like A levels or the Israeli Bagrut should be held in controlled environments, with all possible attention paid to making these settings consistent and fair for all students. We propose that there are policy interventions that could be made at relatively low cost. These including rescheduling exams to different days or different testing sites, concentrating exams during seasons with less pollution, and reducing the weight of any single exam by having students judged by their average score over multiple administrations. In light of the great academic and financial stakes of these tests, it is no trivial matter.

Best wishes,

Victor, Avi and Sefi

May 10, 2016

Brexit and the Consequences for International Trade: Three Points that Deserve More Attention

Writing about web page

This column is by Dennis Novy, Associate Professor in the Department of Economics at the University of Warwick, a Research Fellow at the Centre for Economic Policy Research (CEPR) and an Associate at the Centre for Economic Performance at the London School of Economics. Dennis worked as the Specialist Adviser to the House of Lords for their inquiry into the Transatlantic Trade and Investment Partnership (TTIP). He tweets at @DennisNovy. His column first appeared on the Social Market Foundation blog on 4 May 2016.

Trade is an important part of the referendum debate but those advocating Brexit are hugely underestimating the challenges for the UK in the event that it leaves the EU. As an expert on international trade, I believe there are three issues that need more attention in the public debate.

1) The EU isn’t just a free trade area

First, the European Union isn’t simply a free trade area. It is a customs union. This means that every EU country has the same external trade policy with all non-EU countries such as Brazil, China or the U.S. If Britain left the EU, then it would have to renegotiate those external trade relationships. This process would take a very long time, given that there are dozens of such agreements in place or under negotiation. Ten years would seem an optimistic time frame.

But it would not stop there. Once a post-Brexit Britain introduced new agreements with third countries, complicated administrative barriers might kick in such as ‘rules of origin’ to prevent tariff arbitrage with the remainder of the European Union. In addition, the EU Single Market is a deeper form of economic agreement than the typical free trade area. The Single Market provides further benefits such as common regulation and non-discrimination for public procurement. It took decades of political work to reach this level of economic integration.

2) UK capacity constraints

Second, the UK’s capacity to do new deals beyond the EU post-Brexit will be limited and stretched thin by simultaneously needing to agree a new deal with the EU itself. Britain has not negotiated a free trade agreement in over four decades. Britain simply does not have the administrative expertise to carry out some of the functions that the EU currently fulfils on its behalf.

Take the example of the current negotiations on the Transatlantic Trade and Investment Partnership (TTIP) with the United States. The team hammering out the agreement under the guidance of the EU Trade Commissioner in Brussels consists of some of the most experienced and seasoned operators in the business, drawn from all over Europe. These are the best and brightest in the game that can stand up to the venerable team working for the US Trade Representative. It would take Britain years – and serious government resources – to try and match that expertise. For many years, Britain would have to heavily rely on contracted consultants.

3) Obama was polite

Third, President Obama was being polite when he talked about the prospects for a post-Brexit US-UK trade deal during his visit in April 2016. The US would be in no hurry and Britain would find itself at “the back of the queue.” He explained there was a fixed cost to negotiating trade agreements. The U.S. would therefore give priority to larger negotiating partners such as the EU and Pacific partner countries.

Negotiating power largely depends on economic size. If Britain left the EU, its negotiating power would be decimated overnight. The U.S. would therefore likely revert to its default negotiating position, which is to make Britain a “take-it-or-leave-it” offer on terms less favourable than to the remainder of the EU. This is in essence how it negotiates trade agreements with smaller countries. Britain should not expect give-aways.

March 03, 2016

How are India’s New States Faring?

Writing about web page

In the year 2000, three new states – Jharkhand, Chhattisgarh and Uttarakhand – were carved out of the large states of Bihar, Madhya Pradesh and Uttar Pradesh respectively. This column analyses the performance of the new entities before and after breakup, and in relation to their respective rump. Written by Amrita Dhillon, Pramila Krishnan, Manasa Patnam, and Carlo Perroni, it first appeared on Ideas for India on 2 March 2916, and is reprinted here with permission.

Is small beautiful? Since the breakup of the three large states of Bihar, Madhya Pradesh (MP) and Uttar Pradesh (UP) in 2000, and the most recent breakup of Andhra Pradesh in 2014, there has been much debate on whether the new, smaller states have done better in terms of both governance and economic performance. For instance, the particular success of Bihar relative to the past has been attributed to the zeal of the Chief Minister, Nitish Kumar, while the weak performance of the breakaway state Jharkhand compared to its rump (remnant of the old state post breakup), Bihar, has been attributed to the poor and corrupt leadership and the large mining rents available. Speculation about how these states have fared has ranged from the Great Man theory,1 and the natural resource curse,2 to the greater efficiency of governing smaller regions.

New vs rump states

We began this project with some scepticism about ascribing change entirely to the zeal of a particular leader. The starting point of our analysis was investigating the average performance of each of the new entities before and after breakup. In particular, we were interested in the relative performance of the breakaway states vis-à-vis their rump both before 20003 (when the new states did not exist) and post 2000, in order to understand how the breakup affected each part. In what follows, we compare Jharkhand to its rump, Bihar; Chhattisgarh to its rump, Madhya Pradesh; and Uttarakhand to its rump, Uttar Pradesh, over the period 1992 to 2010. We use data on luminosity4 as a proxy for the evolution of economic activity at the level of the assembly constituency (AC), for the period 1992-2010. We also obtained data on the location and value of mining deposits across the three states, with a view to examining the importance of rents from mineral resources.5

In order to identify the effect of state breakup on both economic activity and inequality, we rely on geographic discontinuity at the boundary of each state pre-breakup, to estimate the causal effect of secession on growth and inequality outcomes. Figure 1 shows the states before and after breakup.

Figure 1. The old and new states in India

Our first result is that while all six states experience an increase in luminosity, it is clear that, on average, new states did better than the rump states: the increase in luminosity post-breakup for new states was 35% more than that for rump states. Moreover, the effects vary across pairs of new states and rumps; the average positive effect is driven by Uttarakhand while Jharkhand fares substantially worse than Bihar and Chhattisgarh does about the same as MP.

Natural resources

This led us to explore reasons for the variation across pairs of states and an obvious candidate to focus on seemed to be the variation in the distribution of natural resources. The change in distribution of resources post breakup is striking: almost all of the mineral resources accrued to Jharkhand upon separation, leaving Bihar with very little. In contrast, while Chhattisgarh gained a large share, MP retains considerable resources. Finally, since pre-breakup UP had very little in mineral resources, the division left both the rump and Uttarakhand with even less. Figure 2 shows the distribution of mineral deposits across states after breakup.

Figure 2. The distribution of mineral deposits across states, 2002

The main change due to breakup is the creation of a new political entity and the consequent changes in attribution of rents in the breakaway states. We thus conjecture that the negative outcomes observed in the new states of Jharkhand and Chhattisgarh might be related to the change in political outcomes brought about by the breakup, along with the distribution of natural resource rich ACs between the rump and new states. To examine this conjecture, we investigate differences in outcomes at the AC level across the three states. We find that ACs with an abundance of natural resources do better on average, post breakup, relative to other ACs. However, all ACs do worse in states which end up with a larger fraction of resource-abundant ACs. For example, in Jharkhand, which ended up with a very high fraction of mineral-rich ACs, all ACs do worse than ACs in Bihar. This suggests that the concentration of natural resource rich ACs in states can lead to weak economic outcomes, lower growth and higher local inequality in ACs, post breakup. This effect is especially pronounced for resource rich ACs.

Role of institutional arrangements

We tie these results to recent literature on the economic effects of natural resources via politics. The creation of a new state creates differential incentives for local elites to capture mineral rents, depending on the spatial distribution of natural resources. ACs that have large potential rents generate incentives for rent capture, with attendant spillovers on the citizens. These rents and spillovers in turn, create incentives for vote buying with a corresponding lack of accountability. Thus, when the new state comes with a high fraction of ACs which are rich in mineral deposits (as was the case with Jharkhand, and less so with Chhattisgarh), we suggest that this may cause a loss of electoral accountability and worse outcomes in all constituencies in the state, but especially in the ACs which are prone to rent-grabbing activities. Of course, there are positive effects from breakup: potentially the tax base goes up in states which get the lion’s share of natural resources and the smaller size of states may also imply better political representation and administration.

It is instructive to compare the results here with a recent study by Loazya et al. (2013) examining the impact of a sharp increase in revenues from mining across districts in Peru. They find that while economic outcomes improved in mining districts, inequality increased both within and across all districts. They attribute the better outcomes in mining districts to the legal requirement that revenues are distributed back to mineral-producing districts, while the increase in inequality within mineral-producing districts is attributed to capture by local agents. Viewed from the prism of our theoretical framework, this result can be explained by the predominantly higher effect of increased local revenues, as opposed to the case of India where redistribution happens at the state level and not at the local level. The increase in local inequality in mineral-producing districts, on the other hand, resonates with our findings. The different institutional structures in India imply that political power resides at the state level rather than the local level, so that the political bargain between local- and state-level elites is a key part of the story.

The previous government at the Centre had proposed a draft Mines and Mineral Development and Regulation Bill, 2011, which provided for a 26% share in mining profits for local communities, which would have been a substantial change in policy. The current government has instead proposed an amendment to the original Bill of 1957, which in turn has a rather convoluted provision for sharing of benefits with local communities. It proposes the establishment of District Mineral Foundations (DMFs) in areas affected by mining-related operations. The objective of this foundation would be to work for the "interests and benefits of persons and areas affected by mining related operations.” Holders of mining leases are required to pay the DMF an amount not exceeding one-third of the royalty in the case of new leases and equivalent to the royalty in case of old leases. The amendment allows state governments to set the rules for the foundation and determine its composition (Narain 2015).

These new institutional arrangements might well be the key to the improved performance of areas with high concentration of resources that might succumb to a local natural resource curse otherwise. However, the incentives for local capture of the DMFs cannot be readily dismissed, especially if de facto power continues to reside at the state level. Our research suggests that future state secession in the presence of substantial and spatially concentrated natural resource endowments6 may be particularly susceptible to the natural resource curse. A number of authors (for example, Robinson et al. 2006, Mehlum et al. 2006) have pointed out that in the presence of abundant natural resources, poor institutions lead to poor economic outcomes. Our research highlights the possibility that a particular institutional feature that matters in the face of large rents is where the authority to grant licences and redistribution of surpluses resides.


  1. This is attributed to the historian Thomas Carlyle (1840) who wrote, "The history of the world is but the biography of great men."
  2. This refers to the fact that countries endowed with natural resources such as minerals and oil have tended to grow more slowly than countries without these resources (see Frankel (2012) for a survey).
  3. Here, the comparison is between the region that later broke away and the remaining part of the state.
  4. Luminosity refers to measures of night-time lights visible from space that are increasingly used as a proxy for economic activity/output, particularly in contexts where reliable data on these indicators are sparse (see Henderson et al. 2011, Chen and Nordhaus 2011). The night-time image data is from the Defense Meteorological Satellite Program Operational Linescan System (DMSP-OLS).
  5. We exclude forest resources from our analysis since ascribing rents from them to specific areas is difficult, unlike in the case of minerals. Data on the location, type and size of mineral deposits are from the Mineral Atlas of India (Geological Survey of India, 2001). We are grateful to Sam Asher for sharing his version of the data with us.
  6. In the sense that the fraction of ACs with substantial potential for rent grabbing increases.

Further reading

  • Carlyle, T (1840), ‘On Heroes, Hero-Worship, and the Heroic in History’, Lectures, Yale University Press, Re-published in 2013.
  • Chen, X and W D Nordhaus (2011), “Using luminosity data as a proxy for economic statistics”, Proceedings of the National Academy of Sciences, 108, 8589–8594.
  • Dhillon, A, P Krishnan, M Patnam and C Perroni (2015), ‘The Natural Resource Curse Re-visited: The Case of New States in India’, Working Paper.
  • Frankel, J F (2012), ‘The Natural Resource Curse: A Survey of Diagnoses and Some Prescriptions’, In Commodity Price Volatility and Inclusive Growth in Low - Income Countries, Arezki, R, C Pattillo, M Quintyn and M Zhu (eds.), International Monetary Fund.
  • Henderson, J V, A Storeygard and D N Weil (2011), “A bright idea for measuring economic growth”, The American Economic Review, 101, 194.
  • Kale, S and L Bhandari (2010), “Small States – Large States”, Analytique, VI(4).
  • Loayza, N V, A Mier Y Teran and J Rigolini (2013), ‘Poverty, Inequality and the Local Natural Resource curse’, IZA Discussion Paper No. 7226.
  • Mehlum, H, K Moene and R Torvick (2006), "Institutions and the resource curse", The Economic Journal, Vol. 116 (508), 1-20.
  • Narain, S (2015), ‘The Making of Mining Policy’, The Business Standard, 19 July 2015.
  • Robinson, J A, R Torvik and T Verdier (2006), “Political Foundations of the Resource Curse”, Journal of Development Economics, 79 (2): 447–468.

February 23, 2016

The CAGE Effect

Writing about web page

CAGE researchers have two papers in the current issue of the Economic Journal that has just been published. These papers are The Empire Is Dead, Long Live the Empire! Long-Run Persistence of Trust and Corruption in the Bureaucracy, by Sascha O. Becker, Katrin Boeckh, Christa Hainz, and Ludger Woessmann, in the Economic Journal 126(590), pages 40–74; and Poverty and Aspirations Failure, by Patricio S. Dalton, Sayantan Ghosal, and Anandi Mani, in the Economic Journal 126 (590), pages 165–188.

The abstracts of the two papers are as follows:

The Empire is Dead:

We hypothesise that the Habsburg Empire with its well-respected administration increased citizens’ trust in local public services. In several Eastern European countries, communities on both sides of the long-gone Habsburg border have shared common formal institutions for a century now. We use a border specification and a two-dimensional geographic regression discontinuity design to identify from individuals living within a restricted band around the former border. We find that historical Habsburg affiliation increases current trust and reduces corruption in courts and police. Falsification tests of spuriously moved borders, geographic and pre-existing differences and interpersonal trust corroborate a genuine Habsburg effect.

Povert and Aspirations Failure:

We develop a theoretical framework to study the psychology of poverty and ‘aspirations failure’, defined as the failure to aspire to one's own potential. In our framework, rich and the poor persons share the same preferences and same behavioural bias in setting aspirations. We show that poverty can exacerbate the effects of this behavioural bias leading to aspirations failure and hence, a behavioural poverty trap. Aspirations failure is a consequence of poverty, rather than a cause. We specify the conditions under which raising aspirations alone is sufficient to help escape from a poverty trap, even without relaxing material constraints.

February 11, 2016

Trade and Tasks over Three Decades

Writing about web page

This column, by Sascha O. Becker and Marc Muendler, first appeared on VoxEU on 6 February 2016.

Offshoring and global value chains have reshaped global trade patterns. This column describes how the German economy has been exposed to significant offshoring for at least three decades. The authors find an increasing importance of high-end tasks in the country. Organising and consulting activities under deadlines, changing business conditions, and tougher performance standards are an increasingly common reality in German workplaces. Labour market institutions in German trade partners are largely unrelated to the changing content of German imports.

Offshoring has been considered an opportunity for industries and firms to bring down costs and gain foreign market share, as well as for consumers to get access to lower-priced products of high quality. To whom do these welfare gains from globalisation 2.0 accrue? Workers in high-income countries are particularly worried that their jobs are at risk when production stages move abroad. In a recent study based on survey answers to questions about job characteristics, Blinder and Krueger (2013) conclude that one-quarter of US jobs are potentially offshorable. While this may seem a high number, potential offshorability is not the same as actual offshoring, as rightly pointed out by Blinder himself in an earlier VoxEU column (Blinder 2009).

Trade in task: Theory and empirics

Recent empirical research documents that trade and offshoring affect labour demand by shifting work between occupations (e.g. Jensen and Kletzer 2010, Crinò 2010, Ebenstein et al. 2014). Within occupations and sectors, theoretical research posits (e.g. Grossman and Rossi-Hansberg 2008) and empirical evidence suggests (e.g. Becker et al. 2013) that tasks matter for offshoring. On the one hand, top trained radiologists in low-income countries interpret computer tomography images at a distance, so this high-skilled occupation may well be at risk to be offshored from high-income countries. On the other hand, a less-skilled janitor’s occupation is bound to the workplace and cannot easily relocate. How reflective are these examples of general changes in the economy? How has the composition of workplace tasks changed over the past three decades? Is there a link between offshoring and workplace changes? Do labour market institutions matter for the evolution?

New evidence from German data

In a recent paper (Becker and Muendler 2015), we use worker-level survey data from Germany, covering the period 1979-2006, to take on these important questions. Quite uniquely by international comparison, the survey data allow us to look at workplace changes for the same occupation within the same industry over time. Much of the existing evidence, in contrast, used to be based on a US occupational classification that required the implicit empirical assumption that the task content of occupations is time invariant.

We use five waves of the German Qualifications and Career survey (1979, 1985-86, 1991-92, 1998-99, 2005-06), the so-called BIBB survey for short. The survey offers detailed information about the tasks a worker reports for the job, and we combine the survey evidence by industry with bilateral trade data for Germany and its trade partners. We have carefully connected the survey information to comparable task information over time and can document two particular dimensions of tasks.

  • First, the ‘what?’ dimension of tasks describes 15 activities that a worker may perform or not, such as produce goods; develop, research, construct; organise, plan; or oversee, control.

  • Second, the ‘how?’ dimension of tasks captures 9 performance requirements that a worker may need to perform frequently or rarely in a given occupation, such as improve/adopt techniques; face new situations/activities; have work procedures prescribed in detail; or confront deadlines/pressure.

Much of the existing empirical literature has focused on the how dimension of performance requirements, and in particular on the ‘codifiability’ of job prescriptions (work procedures prescribed in detail) and the ‘routineness’ (rarely face new situations/activities). How have tasks at German workplaces evolved over time? Merely counting up the average number of tasks that a worker performs on the job, performance requirements have changed little and imply that these how requirements remain similarly focused in number (but their nature shifts as we will see). In contrast, there is much action in the so far less prominent what dimension of tasks:

  • We find that the average German worker in 2006 performs six times more activities than in 1979, documenting an increasing trend towards ‘multi-tasking’.

During the same time, Germany’s import and export flows have grown strongly (see Figure 1).

Figure 1. German imports, 1979-2006

Focusing on the impact of intermediate imports

Given our foremost concern with offshoring, we pay particular attention to trade in intermediate goods—arguably the most distinctive feature of the recent globalisation wave. More than a century ago, during the second half of the 19th century and up to World War I, the world economy underwent a first wave of globalisation (‘globalization 1.0’) but foreign trade involved mostly the exchange of final goods. Trade then contracted strongly between the wars, and only resumed after World War II. As trade rose to new heights in the latter part of the 20th century, it also took new forms.

A crucial novelty of the recent expansion in globalisation is the rise of trade in intermediate goods. This second wave of globalisation is associated with the formation of value chains across borders, and a large share of world trade is now transacted within the boundaries of multinational firms but between their affiliates across country borders. That is what we call globalisation 2.0 in short.

Beyond OECD standards, the German import matrix is not based on the import proportionality assumption, by which imports of intermediate goods would be inferred from the proportion in which an industry uses domestic intermediate inputs. Instead, the German Statistical Office breaks down imports of intermediates and of final goods, varying by sector. This distinction between outsourcing—an industry’s use of intermediate inputs from any source—and offshoring is important.

  • Across German industries, the share of outsourcing (intermediate inputs from outside the firm) in the production value has remained fairly constant at about one-half (it dropped from 51 to 47% between 1979 and 1992 but has since risen back to 51% in 2006).

In other words, German industries have been outsourcing about half of their production value for the past three decades.

However, in 1979 only 14% of the intermediate inputs came from abroad (that is about 7% of the production value were imported inputs), but by 2006 close to 22% of inputs are foreign sourced (that is about 11% of the production value are now imported). In other words:

  • German industries have replaced about one in eleven domestic supplies with foreign supplies over three decades.

In this sense, offshoring is far from a recent phenomenon, it has just intensified.

Interestingly, nine out of the top-ten import source countries are the same in every single sample year since 1979, and eight of those nine countries are also among Germany’s top-ten export destinations in every sample year. The one new top-ten importer country for Germany in 2006 is China, replacing Spain from 1999. A hypothesis consistent with this evidence is that intermediate inputs previously sourced from low-income regions within western Europe are now being sourced from east Asia.

The intensified offshoring of intermediate inputs offers scope for globalisation 2.0 to affect the tasks performed by German workers, over and beyond the effects that globalisation 1.0 continues to exert. In fact, consistent with the idea that offshoring leads to more ‘high-end’ tasks being performed at home over time, we find that, in the ‘what’ dimension, activities other than ‘produce’ (which we consider the most offshorable task) have gained detectably more prominence in German workers’ task sets. Importantly, this shift towards more high-end tasks occurs within sectors and occupations, so is a universal phenomenon, as shown in regressions conditioning on sector and occupation effects.

Figure 2 illustrates this point. We show the contribution of 15 activities (relative to the ‘produce’ category which is normalised to 1.0) to the work performed within given occupations and industries over time. The numbers are from the regression analysis (see paper for details). The visibly striking upward turn of the task profile implies that every single activity gains in importance after the base year 1979, relative to the most offshorable reference category ‘1 produce’. Most of the shift away from the benchmark activity ‘produce’ has taken place already by 1986. This early shift in the task profile coincides with descriptive evidence on trade documenting that a strong increase in intermediate imports occurred in the early sample years between 1979 and 1986. Third, the shift away from the benchmark activity ‘produce’ affects both high-end activities such as ‘14 organise/plan’ or ‘15 oversee/control’ as well as lower-end activities such as ‘2 repair/maintain’.

Figure 2. Activity content of German work

In the ‘how’ dimension, shifts are somewhat less pronounced, but we do find that performance requirements such as ‘meet deadlines’ and ‘improve techniques’ gain relative to the reference category ‘prescribed work’.

In regression analysis we assess how tasks have evolved across sectors over time, as predicted by different types of trade flow: intermediate imports, final goods imports, and exports. Across sectors and occupations, exports and imported inputs work mostly in the same direction, raising task frequencies. There is some heterogeneity, however. The change in tasks is slightly stronger in weakly unionised sectors, consistent with the idea that where unions influence the speed of change in the workplace, the transition to high-end tasks proceeds more slowly. Similarly, we find the change in tasks to be slower in tighter labour markets, where arguably the employer side of the labour market is in a weaker position to pursue workplace changes.


In conclusion, we find that the German economy has been exposed to significant offshoring for at least three decades. We document an increasing importance of high end tasks in Germany; organising and consulting activities under deadlines, changing business conditions and tougher performance standards are an increasingly common reality in German workplaces. The German labour market has shown to be capable of adjustment, and we detect no systematic market failure that policymakers would need to address in order to make German workers fit for globalisation. If anything, to be prepared for a future increase in offshoring, employers and workers will likely need to emphasise flexible skills and continued training that enable coordination beyond the immediate realm of the individual workplace.


  • Becker, S O, K Ekholm, and M-A Muendler (2013), “Offshoring and the Onshore Composition of Tasks and Skills,” Journal of International Economics 90(1): 91-106.
  • Becker, S O and M-A Muendler (2015), “Trade and Tasks: An Exploration over Three Decades in Germany,” Economic Policy 30(84): 589-641.
  • Blinder, A S (2009), “On the measurability of offshorability,”, 9 October.
  • Blinder, A S and A B Krueger (2013), “Alternative Measures of Offshorability: A Survey Approach,” Journal of Labor Economics 31(S1): S97-S128.
  • Crinò, R (2010), “Service Offshoring and White-Collar Employment,” Review of Economic Studies 77(2): 595-632.
  • Ebenstein, A, A Harrison, M McMillan and S Phillips (2014), “Estimating the Impact of Trade and Offshoring on American Workers using the Current Population Surveys,” Review of Economics and Statistics 96(4): 581-95.
  • Grossman, G M and E Rossi-Hansberg (2008), “Trading Tasks: A Simple Theory of Offshoring,” American Economic Review 98(5): 1978–97.
  • Jensen, J B and L G Kletzer (2010), “Measuring Tradable Services and the Task Content of Offshorable Services Jobs,” in K G Abraham, J R Spletzer and M J Harper (eds.), Labor in the New Economy. University of Chicago Press, ch. 8: 309–35.

February 08, 2016

Onwards and Upwards: American Productivity Growth during the Great Depression

Writing about web page

This column, by Gerben Bakker, Nicholas Crafts, and Pieter Woltjer, first appeared on VoxEU on 5 February 2016.

The Great Depression is considered one of the darkest times for the US economy, but some argue that the US economy experienced strong productivity growth over the period. This column reassesses this performance using improved measures of total factor productivity that allow for comparisons of productivity growth in the Depression era and in later decades. Contrary to Alvin Hansen’s gloomy prognosis of secular stagnation, the US economy was in a very strong position during the 1930s by today’s standards.

Alexander Field (2003) described the 1930s as the “most technologically progressive decade” of the 20th century for the US. He argued that total factor productivity (TFP) growth peaked at that time and was spread widely across the US economy. In making these observations, Field relied primarily on the classic study by Kendrick (1961), but took the 1930s to comprise the years 1929 to 1941, to cover fully both downturn and recovery.

Field’s conclusions may strike today’s economists as surprising. Everyone knows that the US experienced a massive banking crisis during the 1930s and most also remember that Alvin Hansen (1939) famously diagnosed ‘secular stagnation’ as the prognosis for the US economy largely on the basis of pessimism about technological progress. Recently, and in a similar vein, Robert Gordon (2016) has claimed that WWII saved the US from secular stagnation and that in the absence of the war, US growth prospects would have been dismal at best.

Our new paper revisits the measurement of TFP growth in the US before WWII (Bakker et al 2015). We have constructed estimates which improve on those of Kendrick in several important ways. In particular, we provide a much more detailed breakdown of TFP growth at the industry level and we take account of improvements in labour quality and capital services. This type of thorough growth accounting controls for heterogeneity across the factor inputs. Consider labour input. Simple hours-worked data combine hours of lawyers, managers, barbers, bricklayers as well as farm labourers, disregarding the differences in education, work-experience, and gender across these workers. As noted by Fernald (2014), these various professions have very different wage rates, which is likely to correspond to differences in marginal products. In a similar vein, a light truck (that is generally written off in about 10 years) needs to have a higher marginal product than a hospital (which might provide services for 50 years). In our paper, we weight the different factor inputs using observed or estimated relative factor prices to control for these implicit differences in marginal products. For 1929 to 1941, this yields a new set of estimates constructed on a similar basis to that used by the US Bureau of Labor Statistics nowadays, allowing us to compare the growth of TFP during the Depression to technological change in later decades.

An overview of these new estimates of the sources of labour productivity growth is given in Table 1. Compared with Kendrick, we find that labour quality contributes more and TFP growth less. For this period as a whole, TFP growth accounted for about 60% of labour productivity growth rather than the 7/8th famously attributed to the residual by Solow (1957).1 Contrary to secular stagnation pessimism, TFP growth was very strong both in the 1920s and the 1930s, at 1.7% and 1.9% per year, respectively – well ahead of anything seen in the last 40 years. Regardless, even though the 1930s saw the fastest TFP growth in the private domestic economy before WWII, it was not the most progressive decade of the whole 20th century in terms of TFP growth. Both 1948-60 and 1960-73 were superior at 2.0% and 2.2% per year, respectively (see Figure 1).

Figure 1. TFP growth in the private domestic economy, US, 1899-2007 (% per year)

Note: the post-war break points are chosen on the basis of NBER business cycle peaks.
Sources: Bakker et al (2015) and Bureau of Labor Statistics, “Historical multifactor productivity measures”, (October 2014); National Bureau of Economic Research, “US business cycle expansions and contractions,” (accessed 28 November 2015).

Table 1. Sources of labour productivity growth: US, 1929-1941 (% per year)

Notes: growth accounting estimates for the private domestic economy. For 1929-1941, growth accounting is based on estimates of capital and labour services constructed on a similar basis to those provided by the BLS for the post WWII period. For 1899-1929, estimates are on the basis of the capital-stocks concept used by Kendrick (1961) but labour quality is on a labour-services basis which, unlike Kendrick, takes into account improvements in education within occupations.
Source: Bakker et al (2015).

That said, as Table 2 reports, Field’s emphasis on the very broad-based advance in TFP during the 1930s is amply justified. Strong TFP growth is consistent with the recovery of R&D spending after the early 1930s, with the volume and range of technical publications in the late 1930s (Alexopoulos and Cohen 2011), as well as with the influx of foreign technology and knowledge in the interwar period, which led to a 20-30% spillover-induced increase in domestic invention in affected scientific areas (Moser and Voena 2012, Moser et al 2014). Nicholas (2003) found that a fifth of patents assigned to quoted industrial firms in the 1920s were still cited in patents granted in the last quarter of the 20th century.

Table 2. TFP growth and TFP contribution to labour productivity growth, US private domestic economy, 1919-1941 (% per year)

Note: Contribution = value-added share * TFP growth. ‘Chemicals’,’ Electricity’, ‘Internal combustion engine’, and ‘Communications’ are the ‘One big wave clusters’ listed in footnote 2.
Source: Bakker et al (2015).

The key ‘one big wave’ technology clusters2 of the second industrial revolution highlighted by Gordon show up strongly, but certainly do not dominate. Manufacturing contributes well below half of all TFP growth and the largest sectoral contribution comes from distribution. This was not an economy whose TFP growth was dominated by one general purpose technology, even one as important as electricity – unlike the late 20th century, where ICT loomed much larger. R&D was much more sectorally concentrated than TFP growth, which suggests that the benefits of technical progress spread widely rather than being confined to sectors in which they originated.3 The ability of large ‘unexciting’ sectors such as agriculture, distribution, or financial services to use new technology effectively seemed far more important, as they had a greater impact on growth than the small ‘exciting’ high-tech sectors.

It still seems reasonable to believe that rapid TFP growth in the 1930s was in spite of – rather than because of – the Great Depression. Bank failures and the disruption of lending that resulted were an adverse shock which hurt innovation although the localised nature of bank distress mitigated the impact on R&D somewhat (Nanda and Nicholas 2014). The resilience of TFP growth in the 1930s reflected US success in creating a strong ‘national innovation system’ based on world-leading investments in human capital and R&D (Goldin and Katz 2008, Mowery and Rosenberg 2000) and a market economy in which creative destruction could flourish, which had become well established by the first quarter of the 20th century.

This is reflected in impressive TFP growth in the 1920s and the much greater success of the US in exploiting the opportunities of the second industrial revolution compared with rivals like the UK. US TFP growth was about three times UK TFP growth in the interwar period, and was at least twice as fast in every major sector except agriculture and construction.4 The strength of the US at the time lay in good horizontal (rather than selective) industrial policies which underpinned private-sector innovative effort while at this point federal government R&D was unimportant.

The US still had a significant unemployment problem in the late 1930s, but it surely would not have faced long-term secular stagnation in the absence of WWII. As Kevin O’Rourke (2015) has recently emphasised, based largely on rapid TFP growth, the ‘natural rate of growth’ was high in the US at this time and this would have underpinned high levels of investment to allow capital stock growth to keep up. WWII may have been helpful in offsetting hysteresis effects in the labour market (Mathy 2015), but it was not required to rescue the economy from low trend growth. By today’s standards, the Depression-era US was in a very strong position even when Alvin Hansen was so gloomy.


  • Alexopoulos, M and J Cohen (2011) “Volumes of evidence: Examining technical change in the last century though a new lens”, Canadian Journal of Economics, 44: 413-450.
  • Bakker, G, N Crafts and P Woltjer (2015) “A vision of the growth process in a technologically progressive economy: the United States, 1929-1941”, CEPR, Discussion Paper No. 10995.
  • Fernald, J (2014) "A quarterly, utilization-adjusted series on total factor productivity", Federal Reserve Bank of San Francisco, Working Paper, No. 2012-19.
  • Field, A J (2003) “The most technologically progressive decade of the century”, American Economic Review, 93: 1399-1413.
  • Goldin, C and L F Katz (2008 The race between education and technology, Cambridge, MA: Harvard University Press.
  • Gordon, R J (2016) The rise and fall of American growth, Princeton, NJ: Princeton University Press.
  • Hansen, A H (1939) “Economic progress and declining population growth”, American Economic Review, 29: 1-15.
  • Mathy, G P (2015) “Hysteresis and persistent long-term unemployment: Lessons from the Great Depression and World War 2”, American University Department of Economics, Working Paper 2015-02.
  • Matthews, R C O, C H Feinstein and J C Odling-Smee (1982) British Economic growth 1856-1973, Stanford, CA: Stanford University Press.
  • Moser, P and A Voena (2012) “Compulsory licensing: Evidence from trading with the enemy act”, American Economic Review, 102: 396-427.
  • Moser, P, A Voena and F Waldinger (2014) “German Jewish emigrés and US invention”, American Economic Review, 104: 322-355.
  • Mowery, D C and N Rosenberg (2000) “Twentieth century technological change”, in The Cambridge Economic History of the United States, vol 3, S L Engerman and R E Gallmann (eds), Cambridge: Cambridge University Press.
  • Nanda, R and T Nicholas (2014) “Did bank distress stifle innovation during the Great Depression?”, Journal of Financial Economics, 114: 273-292.
  • Nicholas, T (2003) “Why Schumpeter was right: Innovation, market power and creative destruction in 1920s’ America”, Journal of Economic History, 63: 1023-1058.
  • O’Rourke, K H (2015) “Economic impossibilities for our grandchildren?”, CEPR, Discussion Paper No. 10974.
  • Solow, R M (1957) “Technical change and the aggregate production function”, Review of Economics and Statistics, 39: 312-320.


1 TFP growth and labour quality together account for 85% of labour productivity growth so the contribution of physical capital deepening is much the same as in Solow (1957).

2 These clusters were based on electricity, the internal combustion engine, chemicals, and information, communications and entertainment.

3 In fact, very few industries were research-intensive. Only chemicals, petroleum and electrical machinery had more than 1% of the labour force employed as scientists and engineers.

4 Based on a comparison of Bakker et al (2015) and Matthews et al (1982).

January 13, 2016

Farewell to the Vultures? Argentine Debt Restructuring and Bargaining Theory

Writing about web page

This column, by Marcus Miller and Sayantan Ghosal, first appeared on VoxEU on 8 January 2016.

Shylock's insistence in The Merchant of Venice that his “pound of flesh” be paid as per the contract, regardless of the extreme and grotesque cost to the debtor, is an apt parallel with vulture funds holding out on Argentinian debt pay-outs. This column assesses the Argentinian debt situation and develops an accord that would create a compromise between the extremes on both sides.

The stand off between the Argentine sovereign state and those holdout creditors who have not yet settled has lasted for more than a decade. The gap between what the holdouts claim (no write-down and full compensation for all interest rate and legal costs incurred) and what the sovereign has – until now – seen fit to offer (namely no payment whatsoever) could hardly be wider.

December 18, 2015

Shocking Language

Writing about web page

Shocking language: understanding the macroeconomic effects of central bank communication, by Stephen Hansen and Michael McMahon, is no. 258 in the CAGE working paper series. The authors write:

On December 16, 2015, the US Federal Reserve’s Federal Open Market Committee (FOMC) raised interest rates for the first time in nearly ten years. Markets spent the days and weeks before the FOMC meeting in anticipation of the outcome. But the decision to raise key policy interest rates by 25 basis points was not a surprise and it was not that decision that generated the weeks of anticipation. Instead, it was the statement that the FOMC would release. The signal as to how they, the 19 members of the FOMC, saw the state of the economy and the guidance that they would give on likely future interest rate changes.

The study of monetary policy has typically focused on the effect of interest rate changes on the economy. Since the FOMC first accompanied their decision with a statement in February 1994, central bank communication has emerged as a key tool for central banks in their attempts to influence the economy. In this paper, acknowledging the multidimensionality of monetary policy, we examine the effect on markets and the real economy of both changes in the stance of policy and the effect of communication.

But we also take the multidimensionality a step further. We distinguish between communication about the current state of the economy and communication about the likely path of future interest rates, which we call forward guidance.

The major challenge in measuring the effects of communication is that most communication is words which do not lend themselves readily to quantitative analysis. We overcome this challenge by using novel techniques from computational linguistics to quantify the different dimensions of FOMC communication about monetary policy.

Another challenge, one that is standard in macroeconomics, arises because communication might both affect, and be affected by, the behavior of the economy. This inter-relatedness makes disentangling the effects of communication hard. To overcome this, we use a factor-augmented VAR which takes account of the two-way causation between the variables of interest.

We find that FOMC forward guidance seems to have bigger effects than the FOMC's view of the current economic situation, especially on market yields. However, the main effects of monetary policy still come from changes in the stance. As is typically found, most indicators of macroeconomic activity are not driven by any of the dimensions of monetary policy over the medium term.

December 03, 2015

On Plague in a Time of Ebola

Writing about web page

On Plague in a time of Ebola, by Cormac O Grada, is no. 250 in the CAGE working paper series. The author writes:

Ebola and plague share several characteristics, even though the second and third plague epidemics dwarfed the 2014/15 Ebola outbreak in terms of mortality. This essay reviews the mortality due to the two diseases and their lethality; the spatial and socioeconomic dimensions of plague mortality; the role of public action in containing the two diseases; and their economic impact.In certain ways the two are rather similar--worries about transmission, relatively short incubation, high lethality rate, quick and unpleasant death--so much so that a few historians argued (falsely) a decade ago that the Black Death was a form of Ebola. The paper then discusses the proportion who died in the first wave of the BD, how lethal was it, the effectiveness or not of prevention methods. For insight into the socioeconomic gradient in mortality it focuses more on the last outbreaks of plague in London, where the data are much better than in the 14th century. I also compare the lethality rates of plague and Ebola, and comment on the considerable variation in the latter between the three countries affected. An added concern with Ebola was the very high mortality rates of care workers; their death rate during plague outbreaks is not known. Finally the role prevention and cures in both cases is discussed.

December 01, 2015

Quality and the Great Trade Collapse

Writing about web page

Quality and the Great Trade Collapse, by Natalie Chen and Luciana Juvenal, is no. 249 in the CAGE working papers. Between the third quarter of 2008 and the second quarter of 2009, world trade fell by 30 percent in nominal terms, and 18 percent in real terms. The crisis disproportionately affected trade in a narrow range of products. As these goods represent a large share of world trade, and only a small fraction of world GDP, the world trade-to-GDP ratio collapsed dramatically. This paper digs deeper into what was happening. The authors write:

We explore whether the global financial crisis has had heterogeneous effects on traded goods differentiated by quality. Combining a dataset of Argentinean firm-level destination-specific wine exports with quality ratings, we show that higher quality exports grew faster before the crisis, but this trend reversed during the recession. Our results suggest that a one-unit increase in quality strengthened export growth by about two percentage points before the crisis, and subsequently lowered it by two percentage points during the crisis. We checked if our results could be driven by mean reversion by including lagged dependent variables, but they continue to hold even after controlling for mean reversion.

Quantitatively, the effect is large: up to nine percentage points difference in trade performance can be explained by the quality composition of exports. This is the difference between the growth in the value of exports that would have prevailed if we assume that the quality of all exported wines had increased at the highest level observed in our dataset and the growth in the value of exports that would have prevailed if the ratings of all wines had instead fallen at the lowest level of quality in the sample. These two counterfactuals provide us with upper and lower bound estimates of the hypothetical performance of trade during the crisis due to changes in the quality composition of exports.

The flight from quality that we observe was triggered by a fall in aggregate demand, was more acute when households could substitute imports by domestic alternatives, and was stronger for smaller firms’ exports.

The importance of our findings is, first, to show that the composition of trade matters for the responsiveness of trade flows to economic downturns. This is helpful to infer how different countries are likely to perform in recessions. Given that higher income countries tend to be more specialized in the production of higher quality goods, our results imply that these countries’ exports might suffer more in recessions.

Second, our results suggest that the real effects of financial crises can be large.

Finally, our findings have implications for understanding the distributional effects of crisis episodes, given that a flight from quality resulting from a negative income shock can be costly for consumers in terms of welfare. Under the assumption that consumers love not only variety, but also quality, a decline in the quality consumed reduces welfare.

About CAGE

CAGE is Warwick’s Centre for Competitive Advantage in the Global Economy. Our blog carries news and discussion of CAGE research on long-run growth and development, the influence of beliefs and cultures, political economy, and economic policy.

March 2017

Mo Tu We Th Fr Sa Su
Feb |  Today  |
      1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31      

Search this blog



Blog archive

RSS2.0 Atom
Not signed in
Sign in

Powered by BlogBuilder