资源描述
Full Terms b Assistant Professor of Economics, Department of Economics, North Central College, Naperville, IL, USA ABSTRACT Recent government attempts at reducing currency demand by promoting cashless payment technology, e.g., subsidies for point-of-sale terminals, have met with mixed success. In this paper, we argue that one reason is the shadow economy: as the size of the unrecorded economy grows, cashless payment technology has a weaker effect on the demand for cash. To test this hypothesis, we analyse a panel of 37 countries over the years 20042014 including data on point-of -sale terminals, financial cards in circulation, and the value of total card payments. Using the first principal component of these variables as a proxy for cashless payment technology, we find that a 1% increase in underground activity lowers the impact of cashless technology on the velocity of currency in circulation by 0.3%. This decrease is large enough that the average marginal effect of cashless technology on currency velocity is statistically insignificant for several high-shadow countries over the sample period. Overall, our results suggest that government subsidization of digital payment infrastructure will be ineffective when shadow markets are prevalent. KEYWORDS Currency velocity; money demand; payments; shadow economy; financial development JEL CLASSIFICATION C23; E26; E41; E42 I. Introduction A growing body of literature argues that the wide- spread use of paper currency is an impediment to economic growth, particularly in emerging mar- kets. The popularity of cash has been blamed for inhibiting financial inclusion (Bachas et al. 2018), acting as a regressive tax on the unbanked popula- tion (Suri 2017), facilitating theft (Chakravorti 2014), and hindering the effectiveness of monetary policy (Rogoff 2017). For these reasons, organiza- tions such as the United Nations and the G20s Global Partnership have advocated a transition away from cash towards digital payments. In an effort to accelerate this transition, governments have pursued policies including simple tax incen- tives (Greece, Japan, Morocco, Russia, South Korea), making public sector payments exclusively in the form of digital account deposits (Saudi Arabia and South Africa), and outright deployment of point-of-sale (POS) terminals at virtually no cost (Czech Republic, Hungary, India, and Poland). Yet despite the ever-increasing popularity of these initiatives, anecdotal evidence of their effectiveness has been mixed. For example, Polands Polska Bezgotwkowa programme, which included allowing companies to use free POS terminals for over a year, has coincided with a precipitous drop in the number of ATM withdrawls. 1 In contrast, after enacting a similar policy in 2016, Indian consumers still remain largely loyal to paper currency so much so that up to 30% of the new POS terminals installed at reduced costs (post-demonetization) were inactive and returned within a year. 2 The same is true in Mexico, where, despite introduc- tion of the government-backed electronic pay- ment platform CoDi, an estimated 90% of transactions are still done in cash. 3 Given this uneven track record, a natural ques- tion for policymakers is the conditions under which digital payment subsidies are more likely to be effective. In this paper, we suggest one contri- buting factor: the prevalence of the shadow CONTACT Brenden J. Mason bjmasonnoctrl.edu Assistant Professor of Economics, Department of Economics, North Central College, Naperville, IL 60540 1 2 transactions/articleshow/62934289.cms?from=mdr. 3 APPLIED ECONOMICS 2021, VOL. 53, NO. 6, 703718 doi/10.1080/00036846.2020.1813246 2020 Informa UK Limited, trading as Taylor Stix 2003; Bounie and Waelbroeck 2016), credit cards (Duca and Whitesell 1995; Yilmazkuday and Yazgan 2009), POS terminals (Snellman, Vesals, and Humphrey 2001) and ATM terminals (Daniels and Murphy 1994; Lippi and Secchi 2009). Several notable exceptions are Drehmann, Goodhard, and Krueger (2002), Amromin and Chakravorti (2009), and Bech et al. (2018), who find no evidence that electronic payments displace large denomination notes. Drehmann, Goodhard, and Krueger (2002) hypothesize that the discrepancy between small denomination bills and high denomination bills is that the latter are more extensively used for bad behaviour, which undermines the observed corre- lation between digital payments and cash use. Our dynamic regression results provide evidence for just that: the effect of cashless payment technology on currency velocity is weaker when underground economic activity is high. Our paper also contributes to the literature examining the relation between underground 4 In this paper, we take the term shadow economy to be synonymous with black market, grey market, informal economy and underground economy, i.e., all unrecorded economic activity. 5 For research on the effect of payment technology on the shadow economy and tax evasion, see Sung, Awasthi, and Lee (2017) and Immordino and Russo (2018). For the effect of financial development more broadly defined, see Straub (2005), Blackburn, Bose, and Capasso (2012), Capasso and Jappelli (2013) and Berdiev and Saunoris (2016). 704 P. MARMORA AND B. J. MASONactivity and alternative payment methods. Although it is generally accepted that the informal economy fosters greater demand for cash, there is surprisingly little empirical research connecting unreported income to the velocity of currency. Instead, a more common approach is to assume that the connection exists and exploit it to extra- polate shadow economy estimates from observed currency demand (Feige 1979; Tanzi 1983; Ardizzi et al. 2014). One notable exception is Onnis and Tirelli (2015), who find that an increase in the shadow economy measured by the MTE approach is associated with a long-run decrease in the velocity of currency in circulation. 6 However, Onnis and Tirelli (2015) do not examine cashless payment technology in their study. Since we also find that the shadow economy derived via the MTE approach reduces the velocity of currency in circulation, our analysis corroborates their gen- eral conclusion using a different empirical strategy and implies that the effect they document is stron- ger when digital payments are widely adopted. Our paper is structured as follows. Section 2 presents a simple inventory model providing intui- tion for why shadow markets reduce the effective- ness of combating currency demand with cashless payment technology. Section 3 describes our esti- mation of the shadow economy and our proxies for cashless technology. Section 4 contains the main dynamic panel regression results. Section 5 explores the robustness of our findings. Section 6 concludes, discusses policy implications, and offers suggestions for future research. II. Model To help motivate the discussion and provide intuition for the ensuing empirical analysis, we first present a simple Baumol (1952)-Tobin (1956) money inventory model incorporating cashless payment technology as in Akhand and Milbourne (1986), which we then extend by add- ing an unobserved underground sector. Assume that agents working in the observed official eco- nomic sector are paid a lump sum nominal income P t Y o t in their bank account at the beginning of each period and plan on spending it in full by the end of the period at a constant rate. Agents can keep their income in the form of interest-bearing bank accounts or withdraw cur- rency for transactions. The fixed cost of withdraw- ing cash from their account is P t w and the interest rate paid on the bank account is r. Next, assume that some cashless payment tech- nology exists which allows an agent to make pur- chases out of their bank account directly and forgo the cost of withdrawing cash. More specifically, assume that agents in the official economy know that available digital infrastructure allows them to conduct purchases in the amount of cP t Y o t straight from their bank account without needing to pay the transfer fee P t w. For example, the term c may represent the widespread availability of POS term- inals, higher acceptance of card payments, or avail- ability of mobile money accounts. Since agents have no incentive to hold currency other than for transactionary purposes, they will keep cP t Y o t in their account and initially divide the rest of their income 1 c P t Y o t between currency and their bank account. For simplicity, we normalize the length of each period to one so that average cash holdings in the official sector will be: M o t 1 c P t Y o t 2N (2:1) where N is the total number of withdrawals in one period. Since holding money incurs an opportunity cost of forgone interest rM o t , each agents problem is to minimize the total cost of money management: NP t w r 1 c P t Y o t 2N (2:2) Taking first-order conditions, solving for the opti- mal N, and substituting into Equation (2.1) yields average demand for real cash balances in the offi- cial sector: M o t P t w 1 c Y o t 2r r (2:3) 6 Schneider (2017) also documents a relationship between cash use and a cash-free version of his MIMIC shadow economy figures. Herwartz, Sarda, and Theilen (2016) use the cash-free MIMIC results to estimate a countrys total income for the purpose of arriving at accurate measurements of the income-elasticity of money demand. APPLIED ECONOMICS 705This is the same money demand equation pre- sented in Akhand and Milbourne (1986). For our next step, we depart from Akhand and Milbourne (1986) and assume there is also an unofficial sector of the economy where cashless payments are less likely to be accepted, effectively setting c 0 with- out loss of generality. If we define Y s t to be real income in the unofficial (i.e., shadow) economy, then repeating the previous steps yields real cash balances in the shadow sector: M s t P t wY s t 2r r (2:4) Combining Equation (2.3) and Equation (2.4) gives average currency demand for the entire economy: M o t P t M s t P t M t P t w 1 c Y o t 2r r wY s t 2r r (2:5) Finally, noting that available estimates of currency velocity are calculated by dividing nominal official income by total currency in circulation, the equiva- lent velocity measure in our model is equal to: V t P t Y o t M t 2rY o t w r 1 1 c p sh t p (2:6) where sh t is the size of the shadow sector relative to the official sector, Y s t Y o t . For our discussion, the velocity equation has three important features. The first feature is how the velocity of currency is affected by c. To see this more clearly, we can take a partial derivative: V t c rY o t 2w r 1 1 c p sh t p 2 1 c p 0 (2:7) which is strictly positive. Therefore, we conclude that the velocity of currency is strictly increasing in cashless payment technology. Intuitively, cashless payment infrastructure reduces the need to hold onto zero-interest bearing cur- rency, which lowers the total amount of cur- rency demanded relative to official income. Second, differentiating Equation (2.6) with respect to sh t yields a negative effect, which is consistent with conventional wisdom and supports the findings of Onnis and Tirelli (2015). Third, and what is particularly impor- tantly for our analysis, Equation (2.7) is also strictly decreasing in sh t , which we can see by taking the cross-partial derivative of Equation (2.7) with respect to sh t : 2 V t csh t rY o t 2w r 1 1 c p sh t p 5 1 c p sh t p 0 (2:8) In words, the positive effect of cashless payment technology on currency velocity is mitigated when the shadow sector is high relative to the official sector. The reason is that cashless payments are not used in the shadow sector, so that if the shadow sector represents a larger share of the total econ- omy, cashless payment adoption will have a smaller effect on the total economy and hence on total cash balances. In order to test this implication empiri- cally, we derive measures of sh t and c in the next section. III. Estimation of the shadow economy and measurement of cashless payment technology Shadow economy estimation: modified total electricity approach Debate over the exact meaning of the term shadow economy has a long history (Schneider 2005). For our purposes, we take the broad definition of Smith (1994) who defines the shadow economy as the market- based production of goods and services, whether legal or illegal, that escapes detection in the official estimates of GDP. By definition, underground activ- ity is difficult to measure. Two of the more popular approaches the currency demand approach and the multiple indicators multiple causes approach (MIMIC) use currency in circulation to extrapolate a measure of the underground economy. 7 This reli- ance on currency is problematic given our research question since using a shadow economy measure that was itself derived from currency will cause a bias in 7 For the currency demand approach, see Feige (1979), Tanzi (1983), and Ardizzi et al. (2014). For the MIMIC approach, see Schneider and Enste (2000), Schneider (2005), Schneider, Buehn, and Montenegro (2010), Schneider (2017), and Medina and Schneider (2018). Despite potentially biasing our results, we nevertheless consider the MIMIC-based shadow economy measures in our robustness section. 706 P. MARMORA AND B. J. MASONestimating the impact of the shadow economy on alternative payment methods. 8 To avoid this obvious bias in our shadow econ- omy measure, we estimate the shadow economy using the modified total electricity (MTE) approach of Eilat and Zinnes (2002), which builds on the work of Kaufmann and Kaliberda (1996). The MTE approach is based on the idea that elec- tric power consumption and economic activity have a near unit elastic relationship, so that an increase in electricity consumption not coinciding with an increase in official GDP is attributed to unreported economic activity. One criticism of this approach is that electricity consumption growth may not change in step with economic activity when there are energy-saving technological shifts or energy supply shocks. 9 To account for these influences, electricity consumption growth is first modified by extracting any impact of chan- ging electricity prices and size of the industrial sector. The MTE approach proceeds in three steps. First, the following equation is estimated: Electric it 0 1 ElectricPrice it 2 Industrial it v it (3:1) where i and t are country and year indexes, Electric is the annual percentage change in electric power consumption, ElectricPrice is the annual percentage change in electricity prices, and Industrial is the annual percentage change in industrial sector size. Once Equation (3.1) has been estimated, the influence of electricity prices and industrial sector size can be filtered out, leaving growth in total economic activity (TEA): TEA it Electric it 1 ElectricPrice it 2 Industrial it (3:2) Since TEA it measures changes in both unrec- orded and recorded economic acti
展开阅读全文