Schmelzing : In virtually every macro conference in the decade after the financial crisis , you saw these charts that started in 1975 and show global real interest rates and global nominal rates decline continuously for a few decades . The big underlying assumption in all of these contributions and all of these discussions was that we had a “ normal ” level of interest rates during the 1960s and that level got shocked by the oil crisis in the ’ 70s and ’ 80s . [ The assumption was that ] we just have to find the right policy tool , the right fiscal tool , to get back to that normal level of interest , which has always existed and which we want to go back to , and which is the ideal steady state interest rate level . That is also a big idea that is formalized in a lot of the finance models and a lot of the underlying economic models / growth models . It has a really foundational role in a lot of the literature — the idea that there is a natural level of interest rates . I would say you can at least trace it back to Irving Fisher and his 1908 book , The Rate of Interest , where he says something very similar . He doesn ’ t call it the steady state interest rate , but he says there is a natural level of interest rates which should be related to the consumption growth rate .
I found that whole view very interesting , but also very deficient in terms of the evidence . Because if you just take a chart that shows you three or four decades and you build this whole foundation view of capital markets and interest rates around a four-decade chart , for a financial historian , that ’ s a very problematic kind of approach . And in that sense , my view was that working on interest rates is not just potentially highly relevant from the policy side , but also from a more fundamental side , a more foundational side . It should be very interesting to look at what the long-run evidence says about the idea that there is a natural rate of interest rates . Where is the evidence for that ? I didn ’ t see any . So that was the other big factor that motivated me .
Pak : Your data is so diverse and profound . What were the challenges in collecting and analyzing this data ?
Schmelzing : I was very fortunate to rely on many printed sources , so it ’ s really a combination of a lot of different types of sources that I ’ m using . Of course , I would be remiss not to mention people who have worked on interest rates . I ’ m not the first one to target Venice or Florence or these early economies and look at what ’ s going on in capital markets . However , a very patchwork type of a body of work exists so far with hundreds of years of gaps for individual countries . And it ’ s a very patchwork situation that we have on interest rates . It ’ s usually not calculating real interest rates . It ’ s often not distinguishing maturities or issuance , so the bottom line is that we cannot really use what is out there for our present discussions because no finance person , no economist would accept the empirical basis that is out there as being remotely useful .
The first step is really to look at the printed material , obviously , to go through all these journal articles in the Economic History Review , in the Journal of Economic History , the standard journals , existing books ( such as Dick Sylla ’ s classic , A History of Interest Rates ), and look through everything that has been done on the printed side . But very quickly , you reach the limit there . And you realize that to really build robust , high-frequency new data that is useful for all present discussions , you need to get your hands dirty , so to speak , and collect new primary data yourself . And the two big sources that give you new primary data are so-called printed primary sources , as the historian calls it , and archival sources .
I am German and a lot of this stuff has been printed in German in the high phase of the so-called historicism period in German science . A ton of archival material was simply printed in volumes in the late phase of the 19th century and in the period of historicism . It ’ s simply printed out of archives , uncommented in big volumes . The most valuable class of sources on that front are so-called historical registries . That means beginning with the Renaissance , virtually every political entity started to make political and economic diaries is the best way to describe it , where every meaningful event on the political front , economic front and financial front was documented by the town scribe or by the regional scribe of the duke or whatever . It was written down [ and ] it was archived back in the day , and a lot of these files have fortunately survived . To the extent that they are printed , we can find them in these 19th-century registries . So that ’ s where a meaningful share of the data originates . About half or so of the other data points , they come from archival sources . For a lot of geographies , these printed primary sources do not exist . And in these cases , if you want to reconstruct advanced economies as a whole with global real rates , as I call them , there ’ s no way to evade the archives and to really go through these dusty files and be very patient .
Pak : So what was your goal ? What did you want to end up with so that you could analyze the data ? And how long did it take you to actually collect it ?
Schmelzing : The goal was really to build the most comprehensive , most robust , most long-term overview of capital market trends that we can find . Not just sovereign interest rates , but also virtually everything that we can find : private interest rates and implied rates embedded in office sales , for instance . Whatever type of financial asset that I could find , I included , to include every other feature of that data point and build the most robust overview of capital market trends for the past 700 years . For every data point , I don ’ t just want the interest rate per se because finance people and economists want to have much more information that is attached to that . What is the maturity of that contract ? What is the collateral that this type of person had to provide ? All this kind of secondary information was highly relevant as well , because when you do that , you can for the first time distinguish series with particular maturity attributes and particular collateral attributes , and create a wide range of series that feature just certain segments of the market over time . On that kind of level , with that kind of nuance , it has never been done before as I see it . And that kind of reconstructing on that kind of level I felt was the most promising approach to really say something about our current low interest rate environment . Through this approach , you can really qualify the whole post-1970s inflection point debate on the most comprehensive basis .
Pak : This dataset that you created , how many countries did it include ? What was the beginning and what was the endpoint , in terms of time ?
Schmelzing : The starting point is the year 1311 , which has at least two reasons . One is that to calculate real interest rates , you
www . MoAF . org | Summer 2022 | FINANCIAL HISTORY 37