货币使用的意外回升（英文版）.pdf

The Surprising Recovery of Currency Usage Jonathan Ashworth a and Charles A.E. Goodhart b a Independent Economist b London School of Economics Currency usage began a long trend decline in the decades after World War II. This was expected to continue, and even accelerate, owing to payment technology innovations. Surpris- ingly, however, such usage as a percentage of GDP stopped falling and has increased quite sharply in recent years in most countries, with Sweden the major outlier. We examine to what extent this may have been due to increasing interest elasticity, nearing the zero lower bound, and also to rising tax evasion, as indirect taxes rise. We also show how currency holdings increased temporarily as the nancial crisis struck in 2008. JEL Codes: E40, E49, E63, H26, N10, N20. 1. Introduction Currency usage, as a percentage of gross domestic product (GDP), peaked in most developed countries toward the end of World War II, and then began a long trend decline. Data for the United States, the United Kingdom, the euro zone, and Japan are shown in gure 1. This decline has been ascribed mainly to innovations and improvements in payments technologies, e.g., a widening use of bank accounts and checks, followed by the availability of plas- tic (debit and credit) cards, and now electronic transactions (Inter- net/mobile phones). Insofar as anyone thought much about currency usage, it was generally viewed as an outdated relic, whose survival was in some large part a kindliness to the aged, unfamiliar with a The authors would like to thank Professor Harrison Hong, the editor, and an anonymous referee for very useful comments. They also thank participants at the Deutsche Bundesbank International Cash Conference 2017, “War on Cash: Is there a Future for Cash?” for suggestions and comments. Author contact: Ashworth: ashwo . Goodhart: Financial Markets Group, London School of Economics, Houghton Street, London WC2A 2AE, United Kingdom. E-mail: ; Tel: 44-20 7955 7555. 239240 International Journal of Central Banking June 2020 Figure 1. Currency-to-GDP Ratios (%) Sources: Ashworth and Goodhart, Friedman and Schwartz (1963), U.S. Bureau of Economic Analysis, U.S. Federal Reserve Board, Bank of England: Three cen- turies of macroeconomic dataversion 2.2, July 2015 (U.K. GDP data annual prior to 1955), U.K. Oce for National Statistics, European Central Bank, OEF, Bank of Japan, Cabinet Oce of Japan. Note: For Japan we use gross domestic expenditure until the end of 1979 and gross domestic product subsequently. world of swipe cards and PayPal; see Friedman (1999), King (1999), and more recently Wolman (2012). As the current generation of aged departed, so we would move toward a payment system where currencynotes and coins, but primarily notes 1 would have disap- peared. The opening page of Amromin and Chakravorti (2009) has the following quotes; also see Krueger (2016). 2 “Except for the smallest of transactions, money will no longer be a physical thing.” (Forbes 1967 as cited in Flannery 1996) “The use of cash and currency will drop drastically.” (Flannery and Jaee 1973) “Cash is dirty, inecient and obsolete.” (Gleick 1996) 1 Bank notes typically account for around 9597 percent of total notes and coins in circulation in developed countries. 2 But for a more considered, and contrary, point of view largely based on empir- ical studies, see Bagnall et al. (2014), Beer, Gnan, and Birchler (2016), Boeschoten (1992), Drehmann, Goodhart, and Krueger (2002), and Fischer, Kohler, and Seitz (2004Vol. 16 No. 3 The Surprising Recovery of Currency Usage 241 “The end of the cash era.” (The Economist 2007, cover page) Insofar as we did think along these linesi.e., that cash hold- ings would continue to decline, relative to GDPwe were wrong. In section 2 of this paper we show that, in many developed economies, currency usage (as a percentage of GDP) stopped falling in the mid-1980s/early 1990s, and in a number of countries then began to rise. Almost all of the alternative transaction technologies are bank mediated, i.e., check payments, debit/credit cards, etc. Banks can, and do, fail, whereas the liability of the government remains valid (often legal tender), e.g., for tax payment, as long as that form of government survives. 3 So, when a generalized concern for the credit- worthiness of banks as a group develops, a shift from bank deposits into cash ensues, as in the United States in the Great Depression, 192933. In a telling, but temporary, example of this, there was a spike in currency holdings in most, but not all, of the countries included in our analysis in 2008:Q4. This emergent panic was, however, soon halted by the aggressive, and often unconventional, monetary poli- cies of central banks; see Bernanke (2015) and Geithner (2014). As we show in section 3, calm in this sense was soon restored, and the shift out of bank deposits into currency, i.e., a rise in the currency- to-deposit (C/D) ratio, was neither as long-lasting nor nearly as extreme as in the United States during the Great Depression in 192933. In a companion short paper, we shall compare develop- ments in C/D ratios in 200809 with those that occurred in 192933, focusing primarily on the United States and the United Kingdom. Although the eect of the 200809 bank crisis had mostly dis- appeared from our time series by 2009:Q4, this did not check the continuing upward trend in currency usage. Indeed, in sev- eral countriese.g., the United States, Australia, South Korea, and Switzerlandthis upward trend has become stronger. 4 This has 3 Some governments also fail, owing to lost wars or revolutions, and then their paper notes become worthless, e.g., the Confederacy in the United States. 4 An outlying example of a developed country where currency usage has contin- ued to decline, even quite markedly, in recent years is Sweden. This is despite the fact that the central bank has introduced negative interest rates there. Sweden242 International Journal of Central Banking June 2020 practical policy relevance. The zero lower bound (ZLB) to (risk- less) interest rates is caused by the ability of agents holding nan- cial assets to switch them into zero-yielding currency notes. The recent sharp rise in currency usage, especially in high-denomination notes, is likely to be a symptom of this. We examine the developing relationship between interest rates and currency usage in section 4. Our particular focus is to examine whether the interest elasticity of demand for cash has increased as interest rates have fallen to the zero lower bound and beyond, into negative territory, i.e., whether such elasticity is potentially nonlinear. Besides its zero-yield feature, another key characteristic of cur- rency is that it is an anonymous, bearer, instrument. This makes it the payment mechanism of choice for those who do not want their transactions to be recorded, more widely known, and capable of being used in evidence against them. Since the black economy (ille- gal transactions) and the grey economy (transactions which would be legal if recorded and taxed, but are transacted anonymously by cash to evade taxation, plus some low-reputation activities) are not recorded by design, it is dicult to estimate how much of outstand- ing currency is held to facilitate such nefarious dealings. However, we do our best to explore this in section 5. The most vocal critic of this usage of currency, especially in the form of high-denomination notes, has been Rogo (1998, 2015, 2016), but also see Sands (2016). Cryp- tocurrencies such as Bitcoin are also becoming a medium of choice for the black economy (see Wolf 2019) and could increasingly be favored over cash in activities such as blackmail, kidnapping, sanc- tion busting, drug smuggling, etc., given that it avoids the need for physical pickup. Even though the speculative boom in Bitcoin is now past, its value has not gone down to zero, but has been quite stable over the last few months at around $3,800. One suspects this is because there is a regular clientele continuing to use it for black economy purposes. Since the characteristics of currency cause this instrument to become the lower barrier to nominal interest rates, at a time of slug- gish growth and low ination (i.e., ZLB), and to be the transaction provides an interesting case study, which again we shall review in a companion paperVol. 16 No. 3 The Surprising Recovery of Currency Usage 243 medium of choice both for criminality and tax evasion, it is not sur- prising that a cottage industry has grown up in this literature about how to abolish currency, or to change its characteristics in such a way as to mitigate these side eects. Much of this work was brought together in a 2015 conference titled “Removing the Zero Lower Bound on Interest Rates” at the Imperial College Business School where speakers included Kenneth Rogo, Willem Buiter, Marvin Goodfriend, David Humphrey, and Miles Kimball. Also see Agar- wall and Kimball (2015) and Buiter and Panigirtzoglou (2003). Our response to this was published in the Sveriges Riksbanks Economic Review; see Goodhart, Bartsch, and Ashworth (2016). Our contribution here to this literature is fourfold. First, we extend the description, and analysis, of aggregate currency usage in a cross-country study by several years. Second, we are, we believe, one of the rst to document the spike in currency demand in several countries caused by the nancial panic in the nal quarter of 2008 (post-Lehman). 5 Third, now that the zero lower bound has been reached, and in some countries transcended into negative interest rates, we revert to the question of the interest elasticity of demand for currency. Fourth, we explore further the relationship between the demand for currency and tax evasion. 2. The Recovery of Currency Usage After falling steadily in the decades after World War II, currency usage began to stabilize as a share of GDP in a number of major countriese.g., the United States and the United Kingdom in the mid-1980s/early 1990sand then began to rise gradually during the 1990s (see gure 1 and table 1). This stabilization and gradual rise came despite ongoing rapid developments in payment technologies, which should have reduced the use of notes and coins further. For example, the number of con- sumer transactions carried out by debit/credit cards soared during this period (see table 2). The number of card transactions has con- tinued to rise sharply in the post-nancial-crisis period, with data showing that card payments per capita increased from 62.5 and 5 Cusbert and Rohling (2013) did this exercise for Australia, and Bartzsch and Seitz (2015) did so for Germany244 International Journal of Central Banking June 2020 Table 1. Changes in Currency-to-GDP Ratios (%) United United Euro States Kingdom Zone Japan Australia Canada Sweden Switzerland Korea 1970:Q11980:Q1 13 38 13 2 15 2 0 1980:Q11990:Q1 0 36 10 13 3 10 18 33 29 1990:Q11999:Q4 33 4 2 46 18 20 2 4 19 1999:Q42007:Q4 2 7 30 26 8 6 21 6 4 2007:Q42016:Q1 43 28 52 25 21 23 54 93 110 Sources: Ashworth and Goodhart, Federal Reserve Board, U.S. Bureau of Economic Analysis, Bank of England, Oce for National Statistics, European Central Bank, OEF, Bank of Japan, Cabinet Oce of Japan, Reserve Bank of Australia, Australian Bureau of Statistics, Bank of Canada, Statistics Canada, Statistiska Centralbyran, Sveriges Riksbank, State Secretariat for Economic Aairs, Swiss National Bank, Bank of Korea. Note: For Switzerland, we used bank notes in circulation data for the periods 1970:Q11980:Q1 and 1980:Q11990:Q1Vol. 16 No. 3 The Surprising Recovery of Currency Usage 245 Table 2. Per Capita Use of New Payment Instruments Debit Cards Credit Cards 1987 1999 1987 1999 United States 0.42 27.5 29.55 68.94 United Kingdom 0.00 35.3 9.19 25.10 France 9.69 48.6 0.05 N/A Germany 0.01 5.2 0.62 4.00 Italy 0.00 4.3 0.22 3.80 Japan 0.01 N/A 2.88 6.51 Canada 0.01 54.3 27.16 37.50 Sweden 3.21 22.2 1.79 6.40 Source: Drehmann, Goodhart, and Krueger (2002). 140.3 in the euro zone and the United Kingdom in 2010, to 80.1 and 201.4, respectively, in 2014; see European Central Bank (2015). In the United States, card payments (credit plus debit) per capita increased from 191 in 2009 to 233 on the latest available data for 2012; see Board of Governors of the Federal Reserve System (2013). In most developed countries, currency usage declined or rose modestly in the pre-crisis decade, although the euro zone was the major outlier, perhaps due to the fact that the euros introduc- tion much increased its international attraction as a store of value. 6 Currency usage in Japan also continued to rise quite briskly, a con- tinuation of the trend (albeit at a slower pace) since the slowdown in the economy after the bursting of its real estate bubble in the early 1990s. Key drivers likely were the lower interest rate (which has been close to zero since late 1995), falling prices, and concerns about the stability of banks. There was a particularly sharp jump in its currency-to-GDP ratio in the early part of that decade amid the ongoing banking crisis, although the currency-to-GDP ratio broadly atlined in the immediate years before the Great Financial Crisis (GFC). 6 There were particularly large gains in currency in circulation in Finland, France, and Germany, although it is not clear why this was the case, as a euro printed in one country is treated identically to one printed in another246 International Journal of Central Banking June 2020 Subsequently, currency usage has been more generally rising, indeed quite sharply and consistently in most countries since the onset of the GFC. Our broader data set consists of 36 countries (OECD plus the BRIC countries), representing over three-quarters of global GDP when measured at purchasing power parity, and runs to 2016:Q1. We plan to revisit this analysis in a couple of years time when we have an additional ve years of data to utilize. Strikingly, 40 percent of countries still had declining currency-to-GDP ratios in the pre-crisis period between 1999:Q4 and 2007:Q4. The largest fall was recorded by Norway at 37 percent, followed by Spain, China, and Sweden at 36 percent, 26 percent, and 21 percent, respectively. Nevertheless, the median increase in the currency-to-GDP ratio of all countries during this period was 8.9 percent, with lower and upper quartiles of 5.9 percent and 31.1 percent, respectively. In the post- crisis period between 2007:Q4 and 2016:Q1, just ve countries (14 percent of our sample) had a declining currency-to-GDP ratio. These were Sweden, Norway, China, South Africa, and Brazil, with all but the last previously having declining currency-to-GDP ratios in the pre-crisis period. Sweden registered the largest fall at 54 percent, followed by Norway and China at 12 percent and 11 percent, respectively. The other major Scandinavian country, Denmark, was also initially on course to have a declining currency-to-GDP ratio in the