Defence spending is rising whether voters like it or not. The UK has committed to 2.5% of national income and aims for nearer 3.5% over the next decade, £30bn a year for each percentage point. What does the country get back? Can defence spending be pro-growth?

In this week's VoxTalk, John Van Reenen (LSE) argues that getting a return on investment based on innovation need not be left to luck. For example nuclear power, GPS and the internet all began as military projects. The spillovers can be planned for; the trick is to make defence spending innovation-rich, and make procurement work better.

Traditional top-down procurement mostly produces lock-in: the same firms winning over and over. Van Reenen's study of a project at the US Air Force shows the difference: when it asked firms what they could build, rather than telling them what to make, the competitions brought in startups, generated more original patents, and spilled ideas into the civilian economy. 

The research behind this episode:

Moretti, Enrico, Claudia Steinwender, and John Van Reenen. 2025. "The Intellectual Spoils of War? Defense R&D, Productivity, and International Spillovers." The Review of Economics and Statistics 107 (1). An ungated version is available as NBER Working Paper No. 26483.

Howell, Sabrina T., Jason Rathje, John Van Reenen, and Jun Wong. 2025. "Opening Up Military Innovation: Causal Effects of Reforms to US Defense Research." Journal of Political Economy 133 (11). An ungated version is available as NBER Working Paper No. 28700.

To cite this episode:

Phillips, Tim, and John Van Reenen. 2026. “Making defence spending pay.” VoxTalks Economics (podcast).

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

John Van Reenen is the Ronald Coase School Professor at the London School of Economics and Director of the Programme on Innovation and Diffusion at the Centre for Economic Performance. He chairs the Council of Economic Advisors to the Chancellor of the Exchequer and is a Research Fellow of the Centre for Economic Policy Research and the NBER. His research spans innovation, productivity, industrial organisation, and the public policies that shape them.

Research cited in this episode

Crowding in, not crowding out. Moretti, Steinwender and Van Reenen tracked industries across twenty-three economies over several decades and found that higher defence R&D spending raised private R&D rather than displacing it, with knock-on gains for productivity growth in the following decades.

The SBIR Open Topics reform. The US Air Force Small Business Innovation Research programme traditionally ran "conventional" competitions specifying the technology wanted; from 2018 it added "open" competitions inviting firms to propose any idea useful to the Air Force. Howell, Rathje, Van Reenen and Wong compared near-winners with near-losers and found the open awards produced new military technology, more original patents, and civilian spillovers such as venture capital funding; the conventional awards mostly produced lock-in.

Spin-offs from military research. Nuclear power, GPS and the internet each began as military projects before becoming civilian technologies; Van Reenen reaches back further to the claw of Archimedes, built to fend off the Roman fleet at Syracuse, as an early example of defence invention finding a wider use.

The Draghi report. Van Reenen worked with Mario Draghi on his 2024 report on European competitiveness; he draws on it to argue that fragmented standards and duplicated procurement across Europe waste money, and that common standards and joint procurement would let countries specialise where they hold a comparative advantage.

More VoxTalks Economics episodes

In January, Tim spoke to Moritz Schularick of the Kiel Institute for the World Economy about whether Europe can convert its industrial base into credible deterrence. Listen to Can Europe Defend Itself?

In January 1860 the New York Times gave its blessing to a new machine: the sewing machine. These "iron needle-women", it wrote, were the only invention that could be claimed “chiefly for women's benefit”. Sewing was women's work in the nineteenth century, rich or poor, and a machine could now do it in a fraction of the time. So did it set women free?

Philipp Ager and Davide Coluccia have traced the adoption of the sewing machine in Massachusetts between 1850 and 1900, using census records and digitised business directories to work out who was exposed to it, in the factory and in the home. For poorer women the machine meant work, in garment factories and in boot and shoe production; they married later, had fewer children, and many never married at all. For wealthier women, who had few acceptable jobs open to them, the hours it saved went into earlier marriage and earlier motherhood. Philipp tells Tim Phillips the story of a machine that had very different impacts in different social classes.

The research behind this episode:

Ager, Philipp, and Davide Coluccia. 2026. "Liberation Technology? The Impact of the Sewing Machine on Women." CEPR Discussion Paper No. 21496. CEPR Press, Paris and London. CEPR Discussion Papers are gated; CEPR members and subscribing institutions can download the paper at the link.

To cite this episode:

Phillips, Tim, and Philipp Ager. 2026. "Did the Sewing Machine Liberate Women?" VoxTalks Economics (podcast). 

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guests

Philipp Ager is professor of economics at the University of Mannheim, a Research Fellow of the Centre for Economic Policy Research, and an editorial board member at Explorations in Economic History. His research spans the economic history of the United States, technological change, and the long-run effects of crises and disasters; his work on the Great Fire of London of 1666 featured in an earlier episode of VoxTalks Economics.

Research and sources cited in this episode

The Song of the Shirt. Thomas Hood's poem about a destitute seamstress was first published anonymously in Punch in December 1843. Hood based it on the case of Mrs Biddell, a London widow prosecuted after pawning clothes she had been given to sew. 

Godey's Lady's Book. The most widely read women's magazine in the US at the time crowned the sewing machine "the queen of inventions" in 1860, having calculated that a man's shirt took 20,620 stitches and 14 hours to sew by hand, against an hour and a quarter by machine. 

Singer and the Sewing Machine: A Capitalist Romance. Ruth Brandon's 1977 biography of Isaac Singer (Google Books) is the source for both Singer quotations read in this episode. .

How the Other Half Lives. Jacob Riis, a Danish-born police reporter in New York, published his account of tenement and sweatshop life in 1890 (free at Project Gutenberg). The shirtmaker's testimony read in this episode was given to the State Board of Arbitration during the shirtmakers' strike and reported by Riis in his chapter on the working girls of New York.

The household appliance revolution. Philipp contrasts the sewing machine with the washing machines and vacuum cleaners that arrived two generations later, which economists have credited with freeing women to join the workforce; "Engines of Liberation" by Jeremy Greenwood, Ananth Seshadri and Mehmet Yorukoglu, Review of Economic Studies, 2005, covers this topic. The sewing machine saved time in the same way, but in the 1860s far fewer acceptable jobs awaited the women whose time it saved.

More VoxTalks Economics episodes

The economic effect of the Great Fire of London. Philipp Ager's previous visit to VoxTalks Economics, with Paul Sharp, on what contemporary records reveal about London's uneven recovery after 1666.

Related reading on VoxEU

Gender norms and the labour market, a VoxEU column on how norms, both internalised and enforced by peers, constrain women's labour market outcomes; the modern counterpart of the stigma that kept married women in Massachusetts out of paid work.

Every day, billions of transactions settle between strangers who have no idea which bank the other uses. That lack of friction is not automatic. Nine-tenths of the money in daily circulation has been created by commercial banks, but it stays trustworthy only because central banks stand behind it, and keep the system in balance.

In this week’s episode Tim Phillips talks to Stephen Cecchetti (Brandeis University, CEPR) about what happens when new forms of digital money test that architecture. Cecchetti is one of the authors of the eighth Barcelona Report in The Future of Banking series, part of the Banking Initiative at IESE Business School, just published by CEPR as a free download.

Will retail central bank digital currencies, tokenised deposits, and stablecoins upset the delicate balance of system that has been running for decades? Stablecoins, for example, do not create money, but they claim the status of money without the institutional guarantee that makes money trustworthy. Three jurisdictions — the US, the EU, and the UK — are each resolving the same underlying contradiction in different ways. None has fully resolved it.

The research behind this episode:

Niepelt, Dirk, Stephen G. Cecchetti, Hélène Rey, and Xavier Vives. 2026. Digital Money: The Future of Banking 8. London: CEPR Press. Available as a free download from CEPR.

To cite this episode:

Phillips, Tim, and Stephen G. Cecchetti. 2026. “The digital money supply.” VoxTalks Economics (podcast). 

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Stephen Cecchetti is the Rosen Family Chair in International Finance at Brandeis University, a Research Fellow of the Centre for Economic Policy Research (CEPR), and a Research Associate at the NBER. He was previously Economic Adviser and Head of the Monetary and Economic Department at the Bank for International Settlements, and Director of Research at the Federal Reserve Bank of New York. His research spanning monetary policy, financial stability, and banking regulation has shaped both academic and policy debate over three decades. He blogs at moneyandbanking.com.

Research cited in this episode

Walter Bagehot's lender of last resort doctrine. In Lombard Street: A Description of the Money Market (1873), Bagehot argued that a central bank under stress should lend freely against good collateral at a penalty rate. The prescription remains the intellectual foundation for how central banks manage runs and systemic crises. Cecchetti invokes it to make the point that no private substitute for a central bank backstop has ever proved durable, and that the doctrine is now, one hundred and fifty years on, being tested by instruments its author could not have imagined.

Monetary uniformity, mobility, and elasticity. The three institutional conditions underpinning general acceptance of money, developed in analysis by the Bank for International Settlements and discussed extensively in the report. Uniformity means a pound is a pound regardless of which bank holds it. Mobility means claims move between users and institutions at low cost and settle with finality. Elasticity means the supply of money can expand when it is under stress. Together they explain why we accept a deposit at face value without doing any analysis of the bank that issued it; and together they identify exactly where new forms of digital money create institutional gaps.

Silicon Valley Bank failure, March 2023. SVB's collapse illustrates both the lender of last resort functioning and the limits of no-bailout commitments. Cecchetti notes that SVB's liabilities were still trading at par on the Thursday before its Friday failure because the Federal Reserve stood behind them. He also notes that Circle, the issuer of USDC, held $3.3 billion of its reserves at SVB and was effectively bailed out in the resolution. The episode is one of two occasions in the past twenty years where money market fund-like instruments have been backstopped by the Federal Reserve under stress.

Genius Act (United States). Principle-based stablecoin regulation expected to come into effect in the US around 2027. Under its provisions, only stablecoins issued by bank-affiliated issuers will have access to the Federal Reserve; only those will therefore have the institutional backing needed to function as money. Stablecoins issued by non-bank entities will not.

Markets in Crypto Assets Regulation (MiCA), European Union. The EU framework for crypto assets, which entered into force in 2024. For stablecoins, MiCA requires issuers to hold 30 to 60% of their reserves in bank deposits, with no provision for central bank backing. The stated rationale is to keep deposits within the banking system; Cecchetti notes this creates a different category of vulnerability and leaves the question of what happens under stress unresolved.

Bank of England stablecoin proposal (United Kingdom). The Bank of England's approach differs from both US and EU frameworks by explicitly requiring large stablecoin issuers to hold significant reserve deposits at the Bank of England, making them in effect narrow banks with a direct central bank backstop. Cecchetti regards this as the most coherent of the three approaches in terms of institutional logic, though the same fundamental question applies: whether holding to that design under stress would be politically sustainable.

Tether and the jurisdictional challenge. Tether, the largest stablecoin issuer, is registered in El Salvador having previously operated out of the British Virgin Islands. Its tokens are held by users in multiple countries, traded on exchanges in multiple jurisdictions, and backed by US Treasury securities. Cecchetti uses this to illustrate why local regulation, however well-designed, is necessary but not sufficient; effective oversight of instruments that are genuinely global requires international standards and coordination.

Fractional reserve banking and the goldsmith model. The institutional structure described in the episode has roots in mid-seventeenth century England, when goldsmiths began issuing more paper receipts than they had gold in their vaults. The goldsmiths became bankers; the paper became money; the vulnerability to runs became a structural feature of private money creation that persists today. Cecchetti uses the history to make the point that while technology changes how we store and transmit information, the underlying architecture of trust in private money is as old as Newtonian physics.

More VoxTalks Economics episodes

Making banking safe, Stephen Cecchetti and Kermit Schoenholtz. Our financial system is supposed to be more resilient than before the global financial crisis, but that didn’t save Silicon Valley Bank, Signature Bank or First Republic. So what went wrong?

Related reading on VoxEU

New coins on the block: Digital currencies and the financial system. The authors of the Barcelona Report warn that “Digital money will be reliable only where sound institutions and robust technology come together.”

Someone once held a patent on the swing. A piece of wood. Two ropes. The US Patent Office granted it. How often does that actually happen, and what does it cost when the system gets it wrong? Or, how often is a valid patent claim rejected?

Until now, no one knew. Tim Phillips talks to Mark Schankerman of LSE and CEPR, who with co-authors William Matcham spent eight years building the tools to find out. Using natural language processing across a dataset of around one million patent applications, twenty million claims, and fifty-five million examiner decisions, they measure how similar each incoming claim is to the hundred million claims that preceded it, going back to 1976. They find that 81% of initial patent claims fall below the patentability threshold; examiners must negotiate that figure down round by round. And they do a pretty good job. But around a third of all abandoned applications contain at least one valid claim the system failed to protect. You don’t see patents that aren’t awarded, so those errors have, until now, been invisible.

The research behind this episode:

Matcham, William, and Mark Schankerman. Forthcoming. "Screening Property Rights for Innovation." Econometrica. Available as CEPR Discussion Paper DP18334 (gated). Current version dated January 2026.

To cite this episode:

Phillips, Tim, and Mark Schankerman. 2026. “How “well does patent screening work? VoxTalks Economics (podcast). 

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Mark Schankerman is Professor of Economics at the London School of Economics, where his research spans innovation, intellectual property, and the economics of technology. His work has examined how patent rights shape R&D incentives, the market for technology, and the behaviour of innovative firms, with particular attention to the institutions that govern how property rights are allocated and enforced.

Research cited in this episode

Prior art. In patent law, prior art is any publicly available knowledge that predates a patent application. Examiners are required to search prior art and reject claims insufficiently distinct from it. The concept defines the outer boundary of what can be granted protection; the closer a claim is to prior art, the weaker the case for granting it.

Type I and Type II errors in patent screening. A Type I error occurs when an examiner grants a claim that should have been rejected, typically because it is too similar to prior art. This allows the holder to charge royalties and, in the US context especially, to bring litigation. A Type II error occurs when a valid claim is refused or abandoned, depriving the applicant of protection they deserve and reducing future incentives to innovate. Schankerman argues that Type II error is systematically under-discussed in public debate: you can point to a patent that should not have been granted; you cannot point to the invention that was never protected.

Structural model. The paper uses a dynamic structural model, meaning it models the actual institutional rules, incentives, and decision sequences that govern patent prosecution at the USPTO. Structural models allow researchers to run counterfactual experiments, asking what would happen if specific rules or incentives were changed, without running those experiments for real. This is the methodological basis for the paper's policy analysis.

Patent distance measure. The paper's key methodological innovation is a quantitative measure of how similar a patent claim is to existing claims, constructed using natural language processing. The algorithm is trained on existing patent documents and compares the textual content of each incoming claim against all prior claims, covering roughly a hundred million filings going back to 1976. This produces a scalar distance figure that can be compared against an estimated patentability threshold.

Deadweight loss. The standard economic term for the welfare cost created when prices are raised above competitive levels. In the patent context, a wrongly granted claim allows its holder to charge higher licensing fees than the market would otherwise bear, generating a cost for users without a corresponding social benefit.

Request for Continued Examination (RCE). A procedural mechanism in the US patent system that allows applicants to re-open a finally rejected application in exchange for a fee. Unlike the European Patent Office or China's patent system, the USPTO places no hard limit on how many times an applicant can return. Schankerman's counterfactual analysis finds that restricting rounds to one substantially reduces screening costs and discourages strategic padding of claims.

Unified Patent Court (UPC). A specialised European court that began operating in June 2023. Its remit covers the enforcement of patent rights across participating EU member states; it does not conduct patentability examinations. Schankerman argues that by reducing the cost of enforcement, the UPC raises the stakes of the upstream screening process: a wrongly granted patent becomes cheaper and easier to assert.

Amazon one-click patent. Amazon received a US patent on the one-click online purchasing process. Schankerman uses the case to illustrate the core economic argument: the relevant question is not whether an invention is valuable, but whether patent protection was necessary to induce its development. If the invention would have occurred regardless, the grant creates costs without providing the intended innovation incentive.

Intrinsic motivation. The tendency for individuals to pursue a task for its own sake rather than for external rewards. Schankerman's model estimates that USPTO examiners exhibit substantial intrinsic motivation and that this is the primary driver of screening quality. In counterfactual simulations, removing intrinsic motivation causes outcomes to deteriorate markedly; removing the credit-based extrinsic incentive system has a much smaller effect.

Padding. Schankerman's term for the strategic behaviour in which patent applicants include claims that are broader than what is strictly novel, hoping some will survive examiner scrutiny and expand the scope of their eventual property right. The paper measures the extent of padding directly from the distance data and confirms it is widespread.

More VoxTalks Economics episodes

Patent pools for generic drugs, Mark Schankerman talks about how diffusion of new drugs is painfully slow in low-income countries. Do patent pools accelerate the process, and how we could still do a better job of licensing life-saving medicines?

Related reading on VoxEU

Patent screening, innovation, and welfare, Florian Schuett and Mark Schankerman, 6 Nov 2020. Critics of the patent system claim that patent rights are becoming an impediment to innovation, and an instrument to extract rents through patent litigation. This column develops a framework to quantitatively assess the effectiveness of the current US patent system and the welfare impact of reforms.

The fiat money system has survived the Great Inflation, the global financial crisis, and a pandemic. But can it survive digital currencies?

Bitcoin and the blockchain solved a genuine problem in computer science: how to stop people spending the same money twice. Forty years of successful inflation control means central bank money is stable; that is the stability in stablecoins, attempting to solve the volatility problem. What's next? What if the unit of account itself were indexed to consumer prices? Digitalisation might finally make that approach viable at scale. Price stability, by design.

Will we still need cash? Maybe not now, But if you never use it, it may not be there if the blackout comes.

The research behind this episode:

Stracca, Livio. 2025. Redefining the Monetary Standard in the Digital Age: Digital Innovations and the Future of Monetary Policy. Springer Nature.

To cite this episode:

Phillips, Tim, and Livio Stracca. 2026. "Redefining the monetary standard." VoxTalks Economics (podcast). 

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Livio Stracca is Deputy Director General for International and European Relations at the European Central Bank, where he has worked for more than two decades. His research spans monetary economics, international finance, and the implications of digitalisation for central banking, with extensive work on exchange rates, capital flows, and the architecture of the international monetary system. 

Research cited in this episode

The double-spend problem. The fundamental challenge in any decentralised digital payment system: how to prevent a participant from spending the same unit of money twice when there is no trusted central authority to verify transactions. Bitcoin's 2008 white paper offered an innovative solution by making the transaction ledger public, cumulative, and computationally expensive to rewrite. The trade-off is that transparency sacrifices privacy; every transaction is visible to all participants in the network.

The blockchain. A distributed ledger in which transactions are grouped into sequential blocks, each cryptographically linked to the one before. Reversing any transaction requires rewriting every subsequent block, which demands enormous computational effort. This design solves the double-spend problem in a decentralised network but makes the system slow and costly to operate at scale.

The payment trilemma. A framework discussed in the episode and in Stracca's book: any digital payment system can optimise for at most two of three properties simultaneously (universal access, security against fraudulent transactions, and privacy). Cash is the only instrument that escapes the trilemma; digital systems must accept a trade-off among the three, and the choice is often made implicitly by the designer of the system rather than through democratic deliberation.

Hayek, Friedrich A. 1976. Denationalisation of Money. London: Institute of Economic Affairs. The classic argument for currency competition: let currencies compete freely and the one providing the most stable prices will win. Economists, including Milton Friedman, largely rejected the proposal on the grounds that money exhibits strong network externalities; the more people use a currency, the more attractive it becomes to the next user, producing a natural tendency towards monopoly. A formal modern revisitation, finding similar conclusions, is Fernández-Villaverde, Jesús, and Daniel Sanches. 2019. "Can Currency Competition Work?" Journal of Political Economy 127 (3): 1017 to 1058.

Irving Fisher's compensated dollar. A proposal published in Fisher, Irving. 1913. "A Compensated Dollar." Quarterly Journal of Economics 27 (2): 213–235 (the same year the Federal Reserve was created). Fisher argued for a dollar whose purchasing power was held constant by adjusting its gold content in line with prices. The mechanical details of his proposal are no longer relevant, but its animating idea (indexing the unit of account to a price level) has gained new plausibility in a digital context.

The Unidad de Fomento. Chile's inflation-indexed unit of account, in operation since 1967 and updated daily against the consumer price index. It is used widely in long-term contracts, including mortgages, and functions as a security that can be traded. Stracca cites it as evidence that an indexed monetary standard is operationally feasible, and as a prototype for what a digital equivalent might look like at larger scale.

The Great Moderation. The period of low and stable inflation in advanced economies running roughly from the mid-1980s until the inflation episode of 2021 to 2023. Economists attribute it to improved monetary policy frameworks, particularly central bank independence, inflation targeting, and (crucially, in Stracca's account) the introduction of interest on reserves, which gave central banks precise control over the short-term interest rate without draining liquidity. Stracca treats the Great Moderation as the benchmark against which any proposed reform of the monetary standard must be judged.

Programmable money. A form of digital money in which payment is conditional on an independently verifiable event, potentially confirmed by a machine rather than a human intermediary. Example: a payment that executes automatically when a delivery is confirmed by a sensor. Decentralised ledgers make such conditional payments technically straightforward; traditional banking systems can approximate them but with far greater friction. Stracca notes significant enthusiasm for programmable money but also real scepticism about whether the benefits outweigh the complexity in practice.

More VoxTalks Economics episodes

Stablecoins and Global Imbalances, Gilles Moëc explains why we can think of stablecoins as a radical macroeconomic experiment that has arrived at exactly the moment the US external position is showing signs of stress.

Can blockchain decentralise money, contracts, and finance? Bruno Biais on blockchain’s potential, its flaws, and its future.

Do stablecoins threaten financial stability? Richard Portes thinks so.

Europe's NATO members have pledged 3.5% of GDP to rearmament. The political argument is already about which social programmes will be sacrificed to pay for this, when the government chooses guns instead of butter. What does history tell us about what politicians will do?

Christoph Trebesch and Johannes Marzian spent four years assembling the Global Budget Database: 150 years of primary government budget documents from 20 countries, with 116 identified military spending booms in peace and war. They find that governments almost never cut social spending when they rearm; they expand both military and welfare budgets simultaneously. The bill arrives later, as higher taxes. Top income rates typically rise by 10 to 15 percentage points in the decade following a military boom, funded mainly through broad-based income and value-added taxes. With rearmament underway, will history repeat itself?

The research behind this episode:

Marzian, Johannes, and Christoph Trebesch. 2026. "Guns and Butter: The Fiscal Consequences of Rearmament and War." CEPR Discussion Paper 21193. [Gated]

To cite this episode:

Phillips, Tim, and Christoph Trebesch. 2026. "Guns and Butter." VoxTalks Economics (podcast). 

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Christoph Trebesch is Director of the Research Center on International Finance at the Kiel Institute for the World Economy and Professor of Macroeconomics at Kiel University. His research spans sovereign debt, financial crises, China's role in global finance, the economics of populism, and the long-run fiscal history of military spending. He is a Research Fellow of the Centre for Economic Policy Research (CEPR). In 2024 he received the Hermann Heinrich Gossen Award, Germany's leading economics prize for economists under 45.

Research cited in this episode

The Global Budget Database is the primary dataset introduced in this paper. Marzian and Trebesch constructed it from primary archival sources, including national parliamentary budget documents, for 20 countries from 1870 to 2022. Unlike existing datasets that rely on planned rather than realised expenditures, it records what governments actually spent, broken down by ministry and purpose. The Switzerland case illustrates the stakes: standard sources record Swiss military spending at around 2% of GDP during the World Wars. The archival record shows actual spending reached 10% once off-budget items are included; five times the apparent figure.

The Correlates of War (COW) Military Expenditures Dataset is one of the most widely used secondary-source datasets for historical military spending, maintained by the Correlates of War Project. Trebesch uses the Swiss case to illustrate the limitations of secondary-source data: the COW series misses off-budget military items that primary archival documents capture, producing a significantly distorted picture of wartime mobilisation in a number of countries.

Credit booms methodology provided the template for identifying military spending booms. Trebesch and Marzian define a boom as an increase of at least 6.5 percentage points of military spending as a share of GDP over two consecutive years, ending when spending growth falls to zero. This approach, adapted from the literature on financial credit expansions and their economic consequences, allows systematic cross-country and cross-period identification without relying on retrospective classification alone. Each algorithmically flagged episode was then verified against historical sources.

Local projections are the main statistical technique used to trace the long-run fiscal path following military booms. The method estimates how a variable (here, tax revenues and top income rates) evolves over time following an identified shock. It is well suited to the protracted dynamics Trebesch and Marzian observe: tax rates rising over a decade or more after a military buildup and, critically, not returning to pre-boom levels once the spending episode ends.

Exogenous military shocks are the basis of the paper's causal identification strategy. To separate the fiscal effects of military spending from broader economic conditions, the authors distinguish episodes triggered by external geopolitical events from those driven by domestic factors. France's rearmament in the mid-1930s, forced by Nazi Germany's military expansion regardless of French domestic politics, is used as an example of an exogenous peacetime boom. Germany's own rearmament in the same period would not qualify as exogenous, since Germany initiated the shock. The same logic applies to wars: a country attacked faces an exogenous event; the attacker does not.

More VoxTalks Economics episodes

In Can Europe Defend Itself?, featuring Moritz Schularick, Christoph’s colleague from the Kiel Institute, we examine whether Europe has the industrial and strategic capacity to convert its rearmament commitment into credible deterrence, and what European rearmament could mean in practice. 

Related reading on VoxEU

Defence spending: no free lunch, a VoxEU column arguing that increased military expenditure adds modestly to near-term economic activity while adding to fiscal pressure; lasting economic benefits from rearmament are far from guaranteed.

Macroeconomic impacts of defence spending, a VoxEU column modelling the EU-wide effects of raising NATO members' defence spending to 5% of GDP by 2035; projected GDP gains are modest and come at the cost of higher debt-to-GDP ratios.

Converging military spending and its fiscal consequences, a VoxEU column examining long-run trends in military expenditure across countries and the fiscal footprint they leave behind.

The economic effects of military support for Ukraine: evidence from fiscal multipliers in donor countries, a VoxEU column finding that spending multipliers for military expenditure can exceed those for other categories of public spending.

More than one in eight people living in the EU today was born in another country. In fourteen of the bloc's largest economies, it is closer to one in six. For ten years, the same team of researchers has asked what happens to those people next: do they find work, close the gap with their native-born neighbours, and build a settled life? The tenth Migration Observatory report is about to be published, and the decade-long picture it paints is not what the political debate might lead you to expect.

Tommaso Frattini of the University of Milan, one of the report's editors, joins Tim Phillips to examine what a decade of consistent, comparable data actually reveals about immigrant integration across Europe. Who are Europe's immigrants, and has that changed? Is the employment gap between migrants and natives closing, stable, or widening? And does it matter whether a migrant arrives from inside the EU or out? The politics of migration is often poisonous, but the data tells a different story.

The research behind this episode

Frattini, Tommaso, and Anissa Bouchlaghem.  2026. "Immigrant Integration in Europe." Migration Observatory Annual Report, 10th edition. Collegio Carlo Alberto / LdA / CEPR Press. Free download from CEPR Press, forthcoming on 18 May.

To cite this episode

Phillips, Tim, and Tommaso Frattini. 2026. "Immigration and integration in Europe." VoxTalks Economics (podcast).

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Tommaso Frattini is Professor of Economics at the University of Milan and a member of the CEPR Research Policy Network on the Political Economy of Migration. His research spans labour markets, immigration economics, and the long-run integration of migrant populations in Europe. He is one of the founding editors of the Migration Observatory Annual Report series, now in its tenth year, and a co-author of the Collegio Carlo Alberto / LdA reports that underpin this episode.

Research cited in this episode

European Union Labour Force Survey (EU-LFS). Eurostat, collected annually by national statistical offices and harmonised across EU member states. The EU-LFS is the primary source for the Migration Observatory's comparative analysis of employment outcomes across countries and over time. The figures cited in this episode are drawn from the 2024 edition, the most recent available at the time of publication.

The employment gap. A measure of labour market integration defined as the percentage-point difference in the probability of being employed between migrants and native-born residents of the same country. A gap of zero would indicate full employment parity. The Migration Observatory computes the gap both raw and adjusted for observable characteristics such as age, education, and gender; the adjusted figure isolates the portion of the gap that cannot be explained by differences in workforce composition between the two groups.

Migration Observatory Annual Report series. Published annually since 2016 by the Collegio Carlo Alberto and the LdA (Laboratorio di Economia Applicata), in partnership with CEPR. Each edition uses the EU-LFS to benchmark migrant labour market outcomes against those of natives across EU member states. The tenth edition, published in 2026, is the first to offer a consistent decade-long comparison across the full series.

The EU Pact on Migration and Asylum. Agreed by EU member states in 2024, the Pact is the EU's most significant attempt to harmonise migration and asylum policy across member states. Frattini describes it as a step forward on harmonisation; he also notes that European policy continues to prioritise border control over integration, a balance he argues the data does not support.

More VoxTalks Economics episodes

Immigration and Public Goods (June 2023). Do immigrants put pressure on local schools, hospitals, and public finances? Research from the United States tests the most common fears directly. The findings have only become more relevant since the episode aired.

Content note: this episode discusses assisted dying, end-of-life choices, and suicide. Some listeners may find the content distressing.

In April 2024, Daniel Kahneman — one of the most influential psychologists of the twentieth century — emailed his close friends to say goodbye. He was 90 years old, his kidneys were failing, his mental lapses were increasing, and he had decided it was time to go. He flew to Switzerland to end his life at an assisted dying clinic there, because New York, where he lived, did not permit it. Thirteen American states currently allow medical assistance in dying; most require a terminal diagnosis with death expected within six months. Canada, Belgium, and Switzerland allow it on broader terms. The UK introduced a bill to parliament, but it failed to pass. The debate on whether we have the right to end our own lives has not been resolved. 

This week Tim Phillips talks to Al Roth of Stanford University about how economics can contribute to the debate on medical aid in dying (MAID). Roth, a Nobel Prize laureate, has written a new book that argues this, and similar debates, often miss the key insight: the binary choice of “allow” versus “ban” rarely reflects reality. For example, in the United States, he explains that physicians in jurisdictions where assisted dying is illegal are familiar with the practice of administering doses of drugs that will relieve pain, but also end life. 

Roth's argument is not that assisted dying is always right. It is that a moral position that ignores the costs of a ban is not more ethical — it is less honest. Economists, he says, bring one specific thing to this debate: the insistence that trade-offs be made explicit.

The book discussed in this episode:

Roth, Alvin E. 2026. Moral Economics: What Controversial Transactions Reveal about How Markets Work. Basic Books. Published 21 May 2026.

To cite this episode:

Phillips, Tim, and Alvin Roth. 2026. “The right to choose to die." VoxTalks Economics (podcast).

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About the guest

Alvin Roth is the Craig and Susan McCaw Professor of Economics at Stanford University. He was awarded the Nobel Prize in Economics in 2012, shared with Lloyd Shapley, for the theory of stable allocations and the practice of market design. He is one of the architects of modern matching market design, having redesigned the systems used in the United States to match medical residents to hospitals and students to schools. A previous book, Who Gets What — and Why, was published in 2014. 

Research cited in this episode

Repugnant transactions is Alvin Roth's term for a class of transactions that are controversial not because no one wants to engage in them — that would be disgust — but because some people do want to engage in them and others believe they should not be allowed to, typically on moral or religious grounds. The key feature is that the objectors suffer no direct externality from the transaction; their objection is to the thing happening at all, regardless of whether it affects them. Roth's examples across the book include medical aid in dying, kidney sales, paid blood plasma donation, surrogacy, and access to certain drugs. The policy implication is that repugnant transactions, unlike ordinary market failures, cannot be resolved by standard economic tools; they require explicit engagement with the moral contest and careful mechanism design to decide what is permitted, to whom, under what conditions.

Oregon's Death with Dignity Act (1997) was the first US state law permitting physician-assisted dying. It requires a terminal diagnosis with death expected within six months, confirmation from two physicians, a waiting period, and self-administration of the medication by the patient. According to the 2024 report of the Oregon Health Authority, assisted dying accounts for roughly 0.9% of all deaths in Oregon; many patients who obtain a prescription never use it. Oregon's 27 years of data make it the most-studied model for the policy, and its take-up rates and population demographics have informed both advocates and critics in other jurisdictions.

Ezekiel Emanuel and vulnerable populations: A 2016 paper by physician and bioethicist Ezekiel Emanuel and co-authors examined the demographics of patients who access assisted dying in jurisdictions where it is legal and found no evidence that vulnerable populations — defined by disability, age, mental illness, or socioeconomic status — accessed it at higher rates than the broader population of dying patients. Roth cites this as evidence against the argument that legalisation creates pressure on the vulnerable to choose death, while noting that this population-level finding does not rule out individual cases of pressure.

The Hippocratic Oath is the earliest recorded professional commitment by physicians not to participate in assisted dying. Roth notes that Hippocrates formulated the oath in the fifth century CE, and that the very inclusion of a prohibition on helping patients die implies the practice was already occurring — physicians were being asked to do it. The religious objection — that decisions about life and death belong to God — and the medical objection — that a physician's role is to save life, not end it — have both been consistent features of opposition to assisted dying across more than two millennia.

The Canadian Supreme Court decision (Carter v. Canada, 2015) struck down Canada's criminal prohibition on physician-assisted dying on the grounds that it infringed Canadians' constitutional rights to life and to security of the person. The court's reasoning included the counterintuitive argument that denying access to assisted dying could cause people to end their lives earlier and less safely — while still capable of doing so — out of fear of being unable to later. The Canadian framework that followed is more permissive than US state laws: it does not require a terminal diagnosis but instead an irremediable condition causing intolerable suffering. Canada has since debated, and repeatedly delayed, extending the framework to mental illness as a sole underlying condition.

Mechanism design is the field of economics concerned with designing rules, institutions, and processes to achieve desired outcomes, particularly in settings where participants have private information or conflicting interests. Roth is one of its leading practitioners. In the context of assisted dying, mechanism design asks: who can apply, through what process, verified by whom, with what waiting periods, and with what safeguards against coercion or mistaken diagnosis? The differences between Oregon's model (terminal diagnosis, self-administration, annual reporting), Canada's model (irremediable suffering, physician or nurse practitioner administration permitted), and Switzerland's model (available to non-residents) are, in Roth's framing, different mechanism designs with measurably different outcomes.

More VoxTalks Economics episodes

In February, Tim spoke to Martin Ellison and Julian Ashwin about what decisions seniors will take about their later years and whether policy can accommodate both their abilities and their needs. Listen to The Economic Consequences of Living Longer. 

The standard story of American innovation features Silicon Valley, venture capital, and the heroic startup founder.When you trace the history of the internet, GPS, mass-produced penicillin, or the COVID vaccine, the starting point is not a term sheet but a government grant. How much does this matter,  and can we measure it?

Tim Phillips speaks to Paolo Surico of London Business School and CEPR who, working with Andrea Gazzani, Joseba Martinez, and Filippo Natoli, has built the first systematic empirical account of how government-funded innovation has shaped US productivity since the Second World War. The headline result: government-funded patents account for roughly 2% of all patents filed in the post-war period, but explain around 20% of medium-term fluctuations in total factor productivity and GDP growth. The return on every dollar of public R&D is more than double the return on every dollar of private R&D. The key mechanism is not that government crowds out private investment; it crowds it in. For every dollar of public research, roughly another dollar of private investment follows, as talent from universities and research institutes moves into startups that commercialise what the public sector seeded. The logic is high-risk, high-reward: the government takes on the uncertainty and fixed costs that the private sector will not bear, accepting a large number of failures in order to find the breakthroughs that private capital would never have funded. 

The model is now under pressure: 2025 brought the largest cuts to US federal science funding in the post-war period. AI adds a further complication: for the first time, a general-purpose technology is being driven primarily by private capital, and that capital is now pulling the best scientific talent out of research institutes and universities and into industry. If that shift becomes permanent, the direction of innovation will be shaped by profitability rather than by broad productivity and living standards. 

The paper discussed in this episode:

Gazzani, Andrea, Joseba Martinez, Filippo Natoli, and Paolo Surico. 2026. "The Public Origins of American Innovation." CEPR Discussion Paper DP20788. Centre for Economic Policy Research. [gated]

To cite this episode:

Assign this as extra viewing. The citation above is formatted and ready for a reading list or VLE.

About the guest

Paolo Surico is Professor of Economics at London Business School and a Research Fellow of CEPR. [verify URL before publishing] His research focuses on macroeconomics, monetary policy, and the economics of innovation and growth. He has advised central banks and governments on macroeconomic policy and is one of the leading empirical macroeconomists working on the aggregate effects of technology and public investment.

Research cited in this episode

Science: The Endless Frontier (Vannevar Bush, 1945) is the report commissioned by President Roosevelt as the Second World War was ending. Bush, Roosevelt's chief scientific advisor, was asked to distil what the wartime mobilisation of research had taught, and how it could be translated into a peacetime innovation ecosystem. The report identified three pillars: government, to set the direction of innovation by funding areas of strategic importance; research institutes and universities, to push the frontier of knowledge without the constraint of commercial goals; and the private sector, to transform new knowledge into new products. The framework became the organisational blueprint for post-war American science and, Surico argues, is the institutional foundation of American technological and economic leadership. The report is in the public domain and available online.

The NIH and NSF are the two federal agencies whose funded innovations show the strongest subsequent links to productivity growth in the paper's results. The NIH (National Institutes of Health) funds health and biomedical research; the NSF (National Science Foundation) funds basic research across science and engineering. Both are predominantly funders of university and research-institute work — which is, Surico argues, precisely why their output generates larger productivity gains than defence-funded innovation. The result is not that health research is inherently more productive than defence research; it is that both the NIH and NSF fund more basic, frontier-pushing work, and that basic research generates the largest spillovers regardless of the department that pays for it.

Crowding in versus crowding out is the central empirical question in the public R&D literature. Crowding out would mean that government spending on research displaces private spending that would have happened anyway, leaving total innovation roughly unchanged. Crowding in means the opposite: public research creates opportunities and trains talent that then attracts additional private investment. The paper finds consistent evidence of crowding in, particularly when government funds flow to universities and research institutes. For every dollar of public R&D, roughly another dollar of private investment follows, typically as researchers from publicly funded institutions move into startups to commercialise what they developed. This is why the aggregate return on public R&D is more than double the return on private R&D, even though government-funded patents are only two percent of the total.

The Solyndra and Tesla parallel is used to illustrate why anecdote-based arguments about public R&D are unreliable. Solyndra — a solar energy company that received a US government loan guarantee and then failed spectacularly — is a frequently cited example of government waste in innovation funding. Tesla received a loan guarantee in the same round of funding and became one of the most valuable companies in history. Surico's broader point is that the government's logic for innovation investment is high-risk, high-reward: it should expect and accept a large number of failures, because the gains from the successes — when they are large enough — more than compensate for the losses. Evaluating public R&D by its failures misses this; evaluating it by its headline successes also misses it. Systematic analysis across the whole portfolio is required.

Philippe Aghion's Nobel Prize lecture is cited by Surico on the relationship between innovation, competition, and market structure. Aghion, who shared the Nobel Prize in Economics in 2018, developed Schumpeterian growth theory — the idea that economic growth is driven by creative destruction, with new entrants displacing incumbents through innovation. The key implication Surico draws on is that incumbents have a structural incentive not to innovate disruptively, because doing so would destroy the market position they already hold. Startups, which have no existing position to protect, are the natural vehicle for disruptive innovation. This is why the paper finds that government-funded startups generate larger macroeconomic impacts than government-funded incumbents: startups have both the mandate from public funding and the commercial incentive to take market share.

DARPA (the Defense Advanced Research Projects Agency) is the US defence department's high-risk research arm, responsible for funding some of the most consequential technologies of the post-war era, including early internet infrastructure. Surico mentions a less celebrated DARPA project — an attempt to embed microchips into bags for tracking, before drone technology made the approach obsolete — as an example of a genuine failure. It illustrates the high failure rate that comes with high-risk public R&D, and the importance of evaluating the portfolio rather than individual projects.

The Draghi report on European competitiveness is cited by Surico as a potential catalyst for a different model of European public investment in innovation. Europe's problem, in his analysis, is not the level of public spending but its composition: too much goes to procurement and too little to basic research and later-stage startup support. Europe has the talent, the research institutes, and the early-stage startups. What it consistently lacks is the capacity to fund the scaling-up phase, which causes European innovations and innovators to be commercialised in the United States. A reallocation of spending toward public R&D that acts as a venture catalyst for later-stage startups — analogous to what Vannevar Bush's framework did for the US after 1945 — is what Surico believes the Draghi report could enable, if acted on.

In 2003, Premier Wen Jiabao warned that China's growth model was unbalanced between supply and demand, over-reliant on investment and exports. More than 20 years later, the imbalance is smaller — but China is vastly larger. What its economy produces and exports now moves global markets. The argument about China's external surplus is no longer just a spat between Beijing and Washington.

Yiping Huang, Dean of the National School of Development at Peking University, has written a chapter in the fourth Paris Report, published jointly by CEPR and Bruegel, examining China's structural imbalances from the inside. His argument: the same policies that powered 45 years of growth also suppressed household income and consumption. Factor market distortions, especially artificially low interest rates, kept the cost of capital down and subsidised state-owned enterprises; decentralised GDP-target competition pushed local governments toward investment and industrial expansion rather than services and household support.

The result was a powerful supply side with a persistently weak domestic demand side. When you produce more than you can sell at home and you are a small economy, you export the rest. When you are the world's second largest economy, the world notices. 

China's consumption share of GDP rose from around 50% in 2010 to 57% in 2024, still well below the mid-seventies average of comparable economies, and two fresh crises complicate the path. 

The property market has been contracting since mid-2021 and it is now a drag on local government finances, household wealth, and bank balance sheets. Local government subsidies have created overcapacity in new industries such as electric vehicles and batteries. Huang's conclusion is that rebalancing is necessary and achievable, but it requires the government stepping back from direct resource allocation, the private sector and market taking on larger roles in innovation, and a significant strengthening of social protection to give households both the income and the confidence to spend.

The report discussed in this series of episodes:

Rey, Hélène, Beatrice Weder di Mauro, and Jeromin Zettelmeyer (eds). 2026. The New Global Imbalances. Paris Report 4. CEPR Press and Bruegel. Free to download at cepr.org.

The chapter discussed in this episode:

Huang, Yiping. 2026. "Rebalancing of the Chinese economy: Challenges and policy options." In Rey, Weder di Mauro, and Zettelmeyer (eds), The New Global Imbalances. Paris Report 4. CEPR Press and Bruegel. 

To cite this episode:

Phillips, Tim, and Yiping Huang. 2026. “Rebalancing the Chinese Economy”. VoxTalks Economics (podcast).

Assign this as extra listening. The citation above is formatted and ready for a reading list or VLE.

About Paris Report 4

The fourth Paris Report, The New Global Imbalances, is a joint publication of CEPR and Bruegel. It was edited by Hélène Rey (London Business School and CEPR), Beatrice Weder di Mauro (Geneva Graduate Institute and CEPR, and President of CEPR), and Jeromin Zettelmeyer (Bruegel and CEPR). The report examines how, in a high-debt and fragmented world, excess savings, rising surpluses, and rising deficits pose a risk to stability and undermine the global trading system. It is free to download at cepr.org.

About the guest

Yiping Huang is Dean of the National School of Development at Peking University. [verify URL before publishing] He is one of China's leading macroeconomists, with research spanning China's economic transition, financial reform, and the political economy of development. He has advised Chinese policymakers and international institutions including the IMF and the Asian Development Bank on issues of growth, financial reform, and structural change.

Research cited in this episode

Asymmetric liberalization is Yiping Huang's term for the approach China took when reforming its economy from the 1980s onward. Rather than the shock therapy adopted by former Soviet economies — privatising state-owned enterprises overnight and hoping markets would fill the gap — China used a dual-track approach. It opened the economy to private firms and foreign investors while maintaining state-owned enterprises in parallel, accepting some inefficiency in exchange for stability in output, employment, and growth. To subsidise the SOEs without direct fiscal transfers, the government kept factor markets, particularly financial markets, partially distorted: deposit and lending rates were held below market-clearing levels, reducing funding costs and effectively transferring income from savers and households to producers. The result was a very strong supply side and a structurally weak domestic demand side, which Huang identifies as the root cause of China's persistent external surpluses.

Involution (Chinese: 内卷, nèijuǎn) is a term in wide use in China to describe a particular form of competitive overextension: effort that intensifies without producing proportional gains in quality, efficiency, or welfare. In the economic policy context Huang uses it, involution refers to the overcapacity problem in China's newer industries, including electric vehicles, batteries, and solar panels. Local governments, motivated by GDP targets and decentralised competition, have subsidised capacity expansion in these sectors without requiring corresponding advances in technology or product quality. The result is high-volume, low-margin competition that can suppress prices globally while leaving firms unable to earn sustainable returns domestically. Huang distinguishes this from the property market crisis, which has a different structure and cause.

New quality productive forces is the term used in China's 15th Five-Year Plan (2026 to 2030) to describe the supply-side transformation the government is aiming for: a shift away from labour-intensive, low-value-added manufacturing toward high-technology, innovation-driven sectors. It reflects the recognition that the industries China dominated in its first decades of reform — low-cost assembly, commodity manufacturing — are no longer competitive given rising domestic wages and costs, and that the next stage of growth has to be driven by productivity and technology rather than factor accumulation.

The 15th Five-Year Plan (2026 to 2030) is China's current medium-term planning document. Huang identifies two key anchors: the development of new quality productive forces on the supply side, and a shift toward domestic demand — particularly private consumption — on the demand side. The plan signals a different role for government, more focused on providing social infrastructure, basic research, and protection for households, and less focused on direct resource allocation and industrial project selection. Huang describes the two anchors as a circuit: if supply-side innovation and demand-side consumption can be connected efficiently, the Chinese economy can sustain growth for much longer without relying on external demand.

The Japan comparison is used by Huang to set expectations for China's consumption rebalancing. Japan's private consumption share of GDP was at its lowest in 1970 and did not reach the average of comparable advanced economies — around the mid-seventies — until around 2010: a process of roughly forty years. China's consumption share is currently around fifty-seven percent, still well below that average. Huang acknowledges the parallel but expresses hope that China can close the gap faster than Japan did; the point of the comparison is that raising household consumption is a structural, decades-long process, not a policy lever that can be pulled in a single plan cycle. It requires sustained growth in household income and improvement in the social safety net to reduce precautionary saving.

China's current account surplus peaked at 9.8% of GDP in 2007, immediately before the global financial crisis. Huang notes that significant adjustment has already taken place: the average surplus between 2018 and the mid-2020s was below two percent of GDP, and the investment share of GDP fell from a peak of forty-seven percent in 2011 to forty-one percent in 2024. The surplus rose to 3.7% of GDP in 2024 partly as a result of weak domestic demand following the property market correction. Huang's argument is that the external imbalance and the internal consumption shortfall are the same problem viewed from different angles; fixing one requires fixing the other.

More VoxTalks Economics episodes

This is the third episode in our series on Paris Report 4. In the first episode, Maurice Obstfeld of the Peterson Institute for International Economics examines the history of global imbalances and what previous episodes can teach today's policymakers. In the second episode, Gilles Moëc, Chief Economist at AXA, explains why the US government is so keen to promote stablecoins and the risks they may pose to the financial system.

For an interview with two of the report's editors, Beatrice Weder di Mauro and Jeromin Zettelmeyer, on the problem of global imbalances, listen to The Sound of Economics, Bruegel's podcast. Available at bruegel.org.