Author Archives: nmpostelvinay

Debt forgiveness in the German mirror

The Economic Consequences of the 1953 London Debt Agreement

By Gregori Galofré-Vilà (Oxford), Martin McKee (London School of Hygiene and Tropical Medicine), Chris Meissner (UC Davis) and David Stuckler (Oxford)

Abstract: In 1953 the Western Allied powers implemented a radical debt-relief plan that would, in due course, eliminate half of West Germany’s external debt and create a series of favourable debt repayment conditions. The London Debt Agreement (LDA) correlated with West Germany experiencing the highest rate of economic growth recorded in Europe in the 1950s and 1960s. In this paper we examine the economic consequences of this historical episode. We use new data compiled from the monthly reports of the Deutsche Bundesbank from 1948 to the 1960s. These reports not only provide detailed statistics of the German finances, but also a narrative on the evolution of the German economy on a monthly basis. These sources also contain special issues on the LDA, highlighting contemporaries’ interest in the state of German public finances and public opinion on the debt negotiation. We find evidence that debt relief in the LDA spurred economic growth in three main ways: creating fiscal space for public investment; lowering costs of borrowing; and stabilising inflation. Using difference-in-differences regression models comparing pre- and post LDA years, we find that the LDA was associated with a substantial rise in real per capita social expenditure, in health, education, housing, and economic development, this rise being significantly over and above changes in other types of spending that include military expenditure. We further observe that benchmark yields on long-term debt, an indication of default risk, dropped substantially in West Germany when LDA negotiations began in 1951 and then stabilised at historically low rates after the LDA was ratified. The LDA coincided with new foreign borrowing and investment, which in turn helped promote economic growth. Finally, the German currency, the deutschmark, introduced in 1948, had been highly volatile until 1953, after which time we find it largely stabilised.

URL: http://EconPapers.repec.org/RePEc:nbr:nberwo:22557

Distributed by NEP-HIS on 2016-09-04

Review by Natacha Postel-Vinay (LSE)

The question of debt forgiveness is one that has drawn increased attention in recent years. Some have contended that the semi-permanent restructuring of Greece’s debt has been counterproductive and that what Greece needs is at least a partial cancellation of its debt. This, it is argued, would allow both faster growth and a higher likelihood of any remaining debt repayment. Any insistence on the part of creditors for Greece to pay back the full amount through austerity measures would be self-defeating.

One problem with this view is that we know very little about whether debt forgiveness can lead to faster growth. Reinhart and Trebesch (2016) test this assumption for 45 countries between 1920-1939 and 1978-2010, and do find a positive relationship. However they leave aside a particularly striking case: that of Germany in the 1950s, which benefited from one of the most generous write-offs in history while experiencing “miracle” growth of about 8% in subsequent years. This case has attracted much attention recently given German leaders’ own insistence on Greek debt repayments (see in particular Ritschl, 2011; 2012; Guinnane, 2015).

Eichengreen and Ritschl (2009), rejecting several popular theories of the German miracle, such as a reallocation of labour from agriculture to industry or the weakening of labour market rigidities, already hypothesized that such debt relief may have been an important factor in Germany’s super-fast and sustained post-war growth. Using new data from the monthly reports of the Deutsche Bundesbank from 1948 to the 1960s, Gregori Galofré-Vilà, Martin McKee, Chris Meissner and David Stuckler (2016) attempt to formally test this assumption, and are quite successful in doing so.

By the end of WWII Germany had accumulated debt to Europe worth nearly 40% of its 1938 GDP, a substantial amount of which consisted in reparation relics from WWI. Some already argued at the time that these reparations and creditors’ stubbornness had plagued the German economy, which in the early 1930s felt constrained to implement harsh austerity measures, thus contributing to the rise of the National Socialists to power. It was partly to avoid a repeat of these events that the US designed the Marshall Plan to help the economic reconstruction of Europe post-WWII.

versailles-hitler

 

Marshall aid to Europe between 1948 and 1951 was less substantial than is commonly thought, but it came with strings attached which may have indirectly contributed to German growth. In particular, one of the conditions France and the UK had to fulfil in order to become recipients of Marshall aid was acceptance that Germany would not pay back any of its debt until it reimbursed its own Marshall aid. Currency reform in 1948 and the setting up of the European Payments Union facilitated this process.

Then came the London Debt Agreement, in 1953, which stipulated generous conditions for the repayment of half the amount due from Germany. Notably, it completely froze the other half, or at least until reunification, which parties to the agreement expected would take decades to occur. There was no conference in 1990 to settle the remainder.

just-wanted-to-point-out

Galofré-Vilà et al. admit not being able to directly test the hypothesis that German debt relief led to faster growth. Instead, making use of simple graphs, they look at how the 1953 London Debt Agreement (LDA) led to lower borrowing costs and lower inflation, which comes out as obvious and quite sustained on both charts.

Perhaps more importantly, they measure the extent to which the LDA freed up space for social welfare investment. For this, they make use of the fact that Marshall aid had mainly been used for infrastructure building, so that the big difference with the LDA in terms of state expenditure should have been in terms of health, education, “economic development,” and housing. Then they compare the amount of spending on these four heads to spending in ten other categories before 1953, and check whether this difference gets any larger after the LDA. Perhaps unsurprisingly, it does, and significantly so.

This way of testing the hypothesis that the LDA helped the German economy may strike some as too indirect and therefore insufficient. This is without mentioning possible minor criticisms such as the fact that housing expenditure is included in the treatment, not control group (despite the 1950 Housing Act), or that the LDA is chosen as the key event despite the importance of the Marshall Plan’s early debt relief measures.

Nevertheless testing such a hypothesis is necessarily a very difficult task, and Galofré-Vilà et al.’s empirical design can be considered quite creative. They are of course aware that this cannot be the end of the story, and they are careful to caution readers against hasty extrapolations from the post-war German case to the current Spanish or Greek situation. Some of their arguments have somewhat unclear implications (for instance, that Germany at the time represented 15% of the Western population at the time, whereas the Greek population represents only 2%).

germany-wwii-debts

Perhaps a stronger argument would be that Germany’s post-war debt was of a different character than Greek’s current debt: some would even call it “excusable” because it was mainly war debt; it was not (at least arguably) a result of past spending excesses. For this reason, one may at least ask whether debt forgiveness in the Greek context would have the same — almost non-existent — moral hazard effects as in the German case. Interestingly, the authors point out that German debt repayment after the LDA was linked to Germany’s economic growth and exports (so that the debt service/export revenue ratio could not exceed 3%). This sort of conditionality is strangely somewhat of a rarity among today’s sovereign debt contracts. It could be seen as a possible solution to fears of moral hazard, thereby mitigating any differences in efficiency of debt relief emanating from differences in the nature of the debt contracted.

 

References

Eichengreen, B., & Ritschl, A. (2009). Understanding West German economic growth in the 1950s. Cliometrica, 3(3), 191-219.

Guinnane, T. W. (2015). Financial vergangenheitsbewältigung: the 1953 London debt agreement. Yale University Economic Growth Center Discussion Paper, (880).

Reinhart, C. M., & Trebesch, C. (2014). A distant mirror of debt, default, and relief (No. w20577). National Bureau of Economic Research.

Ritschl, A. (2011). “Germany owes Greece a debt.” in The Guardian. Tuesday 21 June 2011.

Ritschl, A. (2012). “Germany, Greece and the Marshall Plan.” In The Economist. Friday 15 June.

Advertisements

The USA’s First ‘Belle Époque’ (1841-1856)

America’s First Great Moderation

By Joseph Davis (NBER) and Marc Weidenmier (Claremont – McKenna University, marc.weidenmier@cmc.edu)

Abstract 

We identify America’s First Great Moderation, a recession-free 16-year period from 1841 until 1856, that represents the longest economic expansion in U.S. history. Occurring in the wake of the debt-deleveraging cycle of the late 1830s, this “take-off” period’s high rates of economic growth and relatively-low volatility enabled the U.S. economy to escape downturns despite the absence of a central bank. Using new high frequency data on industrial production, we show that America’s First Great Moderation was primarily driven by a boom in transportation-goods investment, attributable to both the wider adoption of steam railroads and river boats and the high expected returns for massive wooden clipper ships following the discovery of gold in California. We do not find evidence that agriculture (i.e., cotton), domestic textile production, or British economic conditions played any significant role in this moderation. The First Great Moderation ended with a sharp decline in transportation investment and bank credit during the downturn of 1857-8 and the coming American Civil War. Our empirical analyses indicate that the low-volatility states derived for both annual industrial production and monthly stock prices during the First Great Moderation are similar to those estimated for the Second Great Moderation (1984-2007).

URL: https://ideas.repec.org/p/nbr/nberwo/21856.html

Distributed by NEP-HIS on 2016-03-23

Review by Natacha Postel-Vinay (University of Warwick)

Those who like to study the causes of business fluctuations are often primarily interested in severe downturns, or sometimes wild upswings. They may be tempted to gloss over periods of relative calm where not much seems to be happening. Yet there is a good case for studying such phenomena: surely a long period of low volatility in output, prices and unemployment combined with relatively high sustained growth would make many policy makers happy. As such they deserve our attention.

The Great Moderation is usually thought of as one such period when, from the 1980s up to 2007, US economic growth became both more sustained and much less volatile. The causes of the Great Moderation are still being debated, and range from better monetary policy to major structural changes such as the development of information technologies to sheer luck (for example, an absence of oil shocks). Less well-known is the fact that the US economy experienced a similar phenomenon more than a century earlier, from the 1840s to the mid-1850s. In their paper, Davis and Weidenmier draw our attention to this period as it was, in their view, America’s First Great Moderation.  While much of the paper is spent demonstrating just that, they also look for its causes, and argue that important structural changes in the transportation industry were probably at the origin of this happy experience.

1234xNxNICNRRNewMartinsville.jpg.pagespeed.ic.7fm_WoZdB9

Despite being sometimes pointed out as America’s “take-off” period (Rostow, 1971), the idea that this was the US’s First Great Moderation is far from straightforward. This is partly because the period has been commonly known for its relative financial instability, with for instance financial panics in the late 1930s, 1940s and 1950s. In addition, the National Bureau of Economic Research (NBER)’s official business cycle data do not go further back than 1854, which itself results from the fact that most of the extant data on pre-1854 output is qualitative. Thorp’s Business Annals (1926), for example, are primarily based on anecdotal newspaper reports. Thorp identifies a recession in almost every other year, which Davis and Weidenmier think is a gross overestimation.

Instead, the authors use Davis’s (2004) index of industrial production (IP) and defend their choice by pointing out a number of things. First, this is a newer, high-frequency series which despite its industrial focus is much more precise than, for example, Gallman’s trend GDP data. It is based on 43 annual components in the manufacturing and mining industries which were consistently derived from 1790 to World War I. The series does not contain any explicit information on the agricultural sector, which produced more than half of US output in the antebellum era. However, Davis and Weidenmier argue that any large business fluctuations apparent in this sector would also be reflected in the IP index as the demand for industrial goods was very much tied to farm output. Conversely, the demand for say, lumber, could be intimately related to business conditions in the construction and railroad industries.

Simply looking at standard deviations makes clear that the 1841-1856 period was indeed one of especially low volatility and sustained growth in industrial production, with no absolute normal declines in output. From this data it is thus apparent that even the well-known 1837 financial panic was not followed by any protracted recession, thereby confirming Temin’s (1969) earlier suspicion. Testing more rigorously for breaks in the series, their Markov regime-switching model suggests that the probability of a low-volatility state indeed rises the most during this period as compared to the early and late 19th-century periods. Applying the same model up to the recent era, it even appears that the two Great Moderations were similar in magnitude – a remarkable result.

images

But what could explain this apparently unique phenomenon? To answer this question, Davis and Weidenmier first decompose industrial production into several sectors such as metal products, transportation machinery, lumber, food, textiles, printing, chemicals and leather. They then find that the probability of faster growth and lower volatility during the First Great Moderation is significantly higher for the transportation-goods industry. This corresponds to the general idea that the “transportation revolution” (especially in railroads and ships) was an important aspect of America’s take-off. However, increased production in transportation goods could be a result of increased demand in the economy as a whole. The authors then refute this possibility by showing that transportation production preceded all other industrial sector increases in this period, which would tend to confirm the importance of transportation investment spillover effects into other sectors.

Davis and Weidenmier therefore make a convincing case that the Great Moderation should in fact be called the Second Great Moderation, since a first one is clearly apparent from the 1840s to the mid-1850s. Interestingly, they emphasize that in both cases deep structural changes in the economy seem to have been at work, especially in the realm of general purpose technologies, with significant spillovers (transportation in one case, IT in the other).

69workmen

An important question left to answer is the extent to which an era of the “Great Moderation” type is to be desired, and on what grounds. Although aiming at a quiet yet prosperous era seems legitimate, it is important to remind ourselves that the 1850s ended with a severe financial and economic crisis which some argue had its roots in financial speculation and overindebtedness in preceding years. Likewise, we all know how the 2000s sadly ended. The “Second” Great Moderation also saw significant increases in inequality. Davis and Weidenmier acknowledge not being able to account for financial and banking developments during this era; perhaps this needs to be investigated further (or at least pondered upon). One may ask, indeed, whether such periods of prosperity may not bear in themselves the seeds of their own demise.

 

References

Davis, Joseph. (2004). “An Annual Index of US Industrial Production, 1790-1915.” The Quarterly Journal of Economics 119:4, pp. 1177-1215.

Gallman, Robert. (1966). “Gross National Product in the United States, 1834-1909” in Dorothy S. Brady (ed.) Output, Employment, and Productivity in the United States after 1800. New York, Columbia University Press.

Rostow, W. (1990). The Stages of Economic Growth: A Non-Communist Manifesto. 3rd Edition. Cambridge: Cambridge University Press.

Temin, Peter. (1969). The Jacksonian Economy. New York: W. W. Norton & Company, Inc.

Thorp, William. (1926). Business Annals. NBER.

 

Does Technological Progress Lead to more Human Capital Formation? Evidence from the French Industrial Revolution

The Complementarity between Technology and Human Capital in the Early Phase of Industrialization

By Raphael Franck (Bar-Ilan University and Brown University, raphael.franck@biu.ac.il) and Oded Galor (Brown University, Oded_Galor@brown.edu)

URL: http://d.repec.org/n?u=RePEc:bro:econwp:2015-3&r=his

Abstract

The research explores the effect of industrialization on human capital formation. Exploiting exogenous regional variations in the adoption of steam engines across France, the study establishes that in contrast to conventional wisdom that views early industrialization as a predominantly deskilling process, the industrial revolution was conducive for human capital formation, generating broad increases in literacy rates and education attainment.

Review by Natacha Postel-Vinay (University of Warwick)

While human capital is often thought to be at the root of any development process, early industrialization itself is often thought to be de-skilling. Images of children working long hours executing repetitive tasks usually come up when one thinks of the Industrial Revolution (Humphries, 2010). Yet there is also the idea that industrial and technical development might lead to a greater need for skilled labour to maintain, fix and adapt new machinery. In this case industrial development might lead to a greater supply of schooling and might result in significant human capital improvements. Focusing on early French industrialization in a recent working paper (distributed by NEP-HIS on 2015-05-02), Franck and Galor attempt to demonstrate just this.

Featured image

Steam engine from Lille (Nord departement)

Making use of data from the 1840s, the authors find a positive correlation across French departements between the number of steam engines and human capital indicators such as the share of literate conscripts, the share of pupils in the population, and the number of teachers (which would be more suggestive if also set relative to population). This correlation is best illustrated in a series of shaded maps (Figure 3), although the strikingly high levels schooling and literacy in the north-eastern part of France remain to be explained. Of course, correlation does not necessarily imply causation: it may be that other factors caused both the number of steam engines and the number of teachers to increase in certain areas, which could render any relationship between the two fortuitous.

Figure 3 in Franck and Oded Galor (2015).

Figure 3 in Franck and Oded Galor (2015).

To tackle this endogeneity problem, the authors make clever use of the fact that the first steam engine was introduced in 1735 in Fresnes-sur-Escaut in the Nord departement, near the northern tip of France. Since technology diffusion can be reasonably assumed to occur first around the region where the new technology was first introduced (which was indeed the case), it seems possible to use each departement’s distance from Fresnes-sur-Escaut as an instrument in the regression. In the first stage of the regression, they successfully show that the shorter a departement’s distance from the first steam engine location, the larger the number of steam engines in the departement, which seems quite reasonable.

To prove the exogeneity of the instrument, the authors have to show that human capital formation was not higher closer to the first steam engine location. This is trickier. To support their case, Franck and Galor investigate the relationship between distance from Nord and economic development indicators from around 1700, such as urban population, literacy rates and university location. They find that there is no correlation (although this may be surprising in light of Figure 1). More importantly, human capital may be quite imperfectly captured by these indicators in the pre-industrial era, when human capital may have developed in ways that are quite difficult to measure: through the transmission of skills from masters to apprentices, or learning-by-doing. It has often been shown that there was no clear relationship between technological progress and literacy rates in the early modern era (Mitch, 1999). Accordingly perhaps more detail should be provided in the paper as to why the steam engine was first introduced in this region and not elsewhere.

Figure 1 in Franck and Galor (2015)

Figure 1 in Franck and Galor (2015)

Which brings me to a broader point about the paper. Although its stated aim is to investigate the causal relationship running from technological progress to human capital formation, causality could run the other way around. Although endogeneity issues are explicitly addressed in the paper from (and confounding factors such as land suitability, rainfall, access to waterways, distance from Paris, and market integration duly controlled for), the specific problem of reverse causality is not explicitly dealt with in the text. Reassuringly the IV model should theoretically take care of reverse causality, but the authors could still discuss this possibility in more detail.

Featured image

Boys at school in Nord departement in the 19th c.

Overall though, Franck and Galor rather successfully tackle a very important and highly complex aspect of industrialization processes. By showing that technological improvement led to advances in human capital accumulation, these results in turn trigger a number of questions. Through which mechanism did industrialization lead to better schooling and literacy rates? Was the process demand-driven? Or did parents’ higher wages mean that children no longer had to work to help the family? Finally, could child labour abuse in factories have led to local initiatives to promote schooling? This latter hypothesis is discussed by Weissbach (1989), who emphasizes a particularly strong will to change the status quo in Alsatian and nearby regions — which could partly explain the greater spread of schooling in this part of France. Such inquiries could be the subject of fascinating future research.

Featured image

Children in a textile factory in 19th c. Provence

References

Humphries, Jane. 2010. Childhood and Child Labour in the British Industrial Revolution. Cambridge: Cambridge University Press.

Mitch, David. 1999. “The Role of Education and Skill in the British Industrial Revolution.” In Joel Mokyr, ed., The British Industrial Revolution: An Economic Perspective, 2 ed. Boulder, nd CO: Westview Press, pp. 241–79.

Weissbach, L. S. 1989. Child Labor Reform in Nineteenth Century France: Assuring the Future Harvest. Louisiana State University Press.

On #Trade, #Globalization, #Development and Steamships

The Wind of Change: Maritime Technology, Trade and Economic Development

by

Luigi Pascali (L.Pascali@warwick.ac.uk) University of Warwick (UK) and Pompeu Fabra University (Spain)

ABSTRACT

The 1870-1913 period marked the birth of the …first era of trade globalization. How did this tremendous increase in trade affect economic development? This work isolates a causality channel by exploiting the fact that the steamship produced an asymmetric change in trade distances among countries. Before the invention of the steamship, trade routes depended on wind patterns. The introduction of the steamship in the shipping industry reduced shipping costs and time in a disproportionate manner across countries and trade routes. Using this source of variation and a completely novel set of data on shipping times, trade, and development that spans the great majority of the world between 1850 and 1900, I …find that 1) the adoption of the steamship was the major reason for the …first wave of trade globalization, 2) only a small number of countries that were characterized by more inclusive institutions bene…fited from globalization, and 3) globalization exerted a negative effect on both urbanization rates and economic development in most other countries.

URL: http://econpapers.repec.org/paper/wrkwarwec/1049.htm

Review by Natacha Postel-Vinay

The 1870-1913 period saw the first significant wave of trade globalization, which introduced important economic and social changes throughout the world. Despite an abundant literature on the causes of globalization at the time, there are significant methodological issues with these studies. Even more surprisingly, very little has been said about the impact of globalization in this era on the economies of countries around the world. In particular, an essential question to ask seems to be whether the increase in trade witnessed at the time was conducive to greater economic development worldwide.  In a highly ambitious move, Luigi Pascali’s paper (distributed by NEP-HIS on 2014-07-13) tackles both issues at the same time, and in so doing contributes significantly to the larger debate on the causes and consequences of trade globalization.

The main challenge in answering these two questions is to deal in each case with an endogeneity problem. Start with the causes of the trade boom. In their attempts to determine whether the rise in international trade could be due to transportation costs, authors have often used freight rates as a proxy for these costs. The problem with this approach is that freight rates are the actual price of transportation. They may be affected by factors which are themselves related to the state of trade (such as demand for goods or economic activity). So causation may not actually run from freight rates to trade – but from other factors related to trade to freight rates.

A similar issue arises when looking at the causal relationship (if any) from trade to economic development. As economic activity may itself have a positive impact on trade – and not just the other way around – a researcher dealing with this question may find a positive correlation between the two but will eventually be faced with a potential endogeneity problem.

Pascali found a creative solution to these difficulties. He did so by making use of the fact that the steamship introduced asymmetric changes (ie. exogenous variation) in trade distances between countries.  Before the steamship, shipping times by sail were mainly determined by wind patterns. The steamship therefore introduced greater changes in shipping times between some countries than between others. Such changes were purely independent of other factors affecting trade, and only linked to such things as the direction of wind and water currents. It thus became possible for the author to examine the effect of a large change in shipping time on trade, independent of other factors linked to trade such as economic activity or market structure.

Clipper ship from the 1850s.

Clipper ship from the 1850s.

To compute such a variable, Pascali built an enormous dataset on sailing times (using such variables as velocity and direction of sea-surface winds) and calculated the likely effect of the adoption of the steamship on shipping times for 129 countries between 1850 and 1900. He also expanded available datasets to include more than 5,000 entries on imports and exports and data on urbanization for more than 5,000 different cities.

What he found was that the introduction of the steamship had a much larger (positive) impact on trade than was previously thought.

Pascali also found that he could use the steamship variable to search for causal links running from trade to greater income levels and development. As mentioned above he had isolated changes in shipping times including the influence of countries’ economic activity. But these changes were strongly related to trade itself. They were then used as instrumental variable in a two-stage least squares (2SLS). In other words, this variable effectively dealt with the endogeneity problem in the analysis of the effects of trade on development.

His results were somewhat surprising. Using this variable as an instrument, the regression of development (urbanization, population density and per-capita GDP) on trade yielded mostly significant but negative coefficients on the explanatory variable. It therefore appears that variation in the intensity of trade between two locations does not have a large impact on development – and may even have a negative one.

Even more interestingly, his findings suggest that whether an increase in trade has a positive impact on development depends on a country’s institutions:  only a few countries having a better established rule of law (as measured by “constraints on the executive” – taken from Acemoglu and Johnson (2005)) benefited from an increase in international trade in terms of development. This finding can be related to relatively recent literature (such as Krugman (1991) or Crafts and Venables (2007)) according to which a reduction in trade costs is only beneficial to a certain set of countries (in particular, those specializing in manufacturing).

A steamship from the 1900s.

A steamship from the 1900s.

Pascali’s paper thus contributes to questioning the positive effects of lowering trade barriers, which are too often taken for granted. He carefully suggests that trade may have a differential impact depending on countries’ institutions. Perhaps some elaboration and discussion on how exactly these relationships play out would have been welcome.  Nevertheless the author’s questions, creative methodology and findings all make for a fascinating read.

Additional References

Acemoglu, D. and S. Johnson (2005). “Unbundling institutions”. Journal of Political Economy 113(5): 949–995.

Crafts, N. and A. Venables (2007). Globalization in Historical Perspective. University of Chicago Press.

Krugman, P. (1991). “Increasing returns and economic geography”. Journal of Political Economy 99: 483-499.

Does Bank Competition Lead to Higher Growth?

Bank Deregulation, Competition and Economic Growth: The US Free Banking Experience

By Philipp Ager (University of Southern Denmark) and Fabrizio Spargoli (Erasmus University Rotterdam)

Abstract

We exploit the introduction of free banking laws in US states during the 1837-1863 period to examine the impact of removing barriers to bank entry on bank competition and economic
growth. As governments were not concerned about systemic stability in this period, we are
able to isolate the effects of bank competition from those of state implicit guarantees. We find
that the introduction of free banking laws stimulated the creation of new banks and led to
more bank failures. Our empirical evidence indicates that states adopting free banking laws
experienced an increase in output per capita compared to the states that retained state bank
chartering policies. We argue that the fiercer bank competition following the introduction of
free banking laws might have spurred economic growth by (1) increasing the money stock
and the availability of credit; (2) leading to efficiency gains in the banking market. Our
findings suggest that the more frequent bank failures occurring in a competitive banking
market do not harm long-run economic growth in a system without public safety nets.

URL: http://d.repec.org/n?u=RePEc:hes:wpaper:0050&r=his

Circulated by NEP-HIS on: 2013-12-29

Review by Natacha Postel-Vinay

In this paper, Philipp Ager (University of Southern Denmark) and Fabrizio Spargoli (Erasmus University Rotterdam) ask two very topical questions. Does increased bank competition lead to higher economic growth? And, if so, how? Following the recent crisis, many have wondered whether the alternative to “too-big-to-fail” — having many smaller banks competing with each other — would necessarily be a better one. Clouding the debate has been the difficulty of finding appropriate historical settings in which to test the hypothesis that more competition leads to greater growth. In their paper, Ager and Spargoli focus on what they consider the best instance of intense bank competition without any implicit government bail-out guarantee: the American free banking era.

Between 1837 and 1863 new laws were passed in a number of states allowing just about anyone to set up a bank, with very few requirements to fulfill. Until then, banks wanting to open needed a charter from their state, for which they had to meet relatively stringent criteria. As the authors show using a new quantitative analytical framework, the new laws greatly increased the creation of new banks in the states which passed them. As competition increased, however, a higher proportion of banks ended up failing. Could it still be the case that the introduction of free banking laws led to greater growth in those states?

A satire on Andrew Jackson's campaign to destroy the Bank of the United States and its support among state banks, 1836. It was partly to fill this gap that states allowed free banking.

A satire on Andrew Jackson’s campaign to destroy the Bank of the United States and its support among state banks, 1836. It was partly to fill this gap that some states allowed free banking.

The paper’s most important finding is that increasing competition among banks did lead to higher economic growth. Jaremski and Rousseau’s 2012 paper (previously reviewed in NEP-HIS here) found that a new “free” bank, as opposed to a charter bank, did not have a positive effect on the local economy. While this is an important finding in itself, it is also important to look at the effect of the introduction of free banking laws on aggregate bank behaviour, if only because the new entry of free banks may alter the willingness of charter banks to enter the market and their behaviour once in the market. Charter banks’ behaviour may in turn alter free banks’ behaviour, and so on. Interestingly, Ager and Spargoli’s study finds that in the aggregate, the acceleration in bank entry and resulting greater competition among all types of banks had a positive effect on economic growth.

To arrive at this conclusion, the authors are careful to include a number of controls. First, there is the possibility that growth opportunities led some states to adopt free banking laws, in which case the authors would face a reverse-causality problem. Hence they conduct a county-level analysis in which they include time-invariant county characteristics and state-specific linear output trends (although perhaps it would be nice to see these output trends going further back in time than 1830). Second, they also control for other laws that states might have introduced at the same time as the free banking ones, which could potentially bias the results. Finally, they control for unobserved heterogeneity between states by examining contiguous counties lying on the border of states that introduced free banking. Their results are robust to these different specifications.

Private Bank Note, Drover’s Bank, Salt Lake City, Utah, $3, 1856

Private Bank Note, Drover’s Bank, Salt Lake City, Utah, $3, 1856

Ager and Spargoli are of course also interested in where this growth came from. They find a positive relationship between the introduction of free banking laws and lending, and conclude that one of the main channels through which this increase in growth occurred was the increase in the availability of credit that greater competition fostered. This story is consistent with the finance-growth nexus literature, which argues that greater (and safer) access to credit is conducive to economic development.

Although this seems perfectly reasonable, it would perhaps have been interesting to see when most of the failures occurred. If, for instance, they mainly occurred towards the end of the period under study, say around the 1857 panic, then it is possible that the negative effects of such failures on subsequent growth would not have been picked up by this study, since it ends in 1860. This leaves open the possibility that the positive relationship between free banking and increased access to credit was not a beneficial one for the economy in the long run. Loan growth (and asset growth more generally) is not always a good thing, as the recent crisis has tended to show.

Private Bank Note, Mechanics Bank, Tennessee, $10, 1854

Private Bank Note, Mechanics Bank, Tennessee, $10, 1854

Overall however, Ager and Spargoli’s paper asks a very important question and offers a solid analysis. A natural next step would be to include output data on the periods preceding and following the free banking era, although the occurrence of the Civil War is an obvious obstacle to this study.

 

References

Jaremski, M., and P. L. Rousseau (2012): “Banks, Free Banks, and U.S. Economic Growth,” Economic Inquiry, 51(2), 1603–21.