WORLD BANK WORKING PAPER The fiscal costs of monetary and exchange rate policy distortions in Zimbabwe Victor Steenbergen Carren Pindiriri Jimmy Psillos Marko Kwaramba © 2025 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions Attribution—Please cite the work as follows: Steenbergen, Victor; Pindiriri, Carren; Psillos, Jimmy; Kwaramba, Marko (2025). “The fiscal costs of monetary and exchange rate policy distortions in Zimbabwe”. The World Bank, Washington, DC: World Bank. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank, therefore, does not warrant that the use of any third-party- owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. 2 The fiscal costs of monetary and exchange rate policy distortions in Zimbabwe1 Victor Steenbergena, Carren Pindiriri*b, Jimmy Psillosc, and Marko Kwarambad a World Bank, vsteenbergen@worldbank.org; *bCorresponding author, University of Zimbabwe, Department of Economics and Development, pindiriric@gmail.com, https://orcid.org/0000-0002-3903- 2742; cConfederation of Zimbabwe Industries, jimmy.psillos@crystal.co.zw; dWorld Bank, mkwaramba@worldbank.org ABSTRACT This paper uses a multi-methodological approach to measure the fiscal costs of monetary and exchange rate policy distortions in Zimbabwe. It identifies three channels through which these policy distortions affect tax revenue: the Olivera-Tanzi Effect (inflation-related tax payment lags), an overvalued official exchange rate affecting customs duty on imports, and informalization undermining overall revenue collection. It then measures the fiscal costs of each of these three channels. The paper shows that although inflation tax increases government revenue through seigniorage, its negative impact on tax revenue outweighs its benefit. High inflation and exchange rate distortions have costed Zimbabwe Treasury over three billion US dollars between 2020 and 2023. Policies that remove exchange rate distortions and stabilize prices can substantially improve government revenue and help close the fiscal financing gap. Keywords: inflation tax; Olivera-Tanzi effect; informality; exchange rate distortions; money supply Jel classification: E31; E41; E42; E52; E58; E62; E63; E65 1 This paper was produced as a background paper for the World Bank’s “Zimbabwe Public Finance Review: Anchoring Macroeconomic Stability through Fiscal Policy” (2025). Special thanks to Abha Prasad (Practice Manager, EAEM1) for her helpful comments and support. 3 I. Introduction Zimbabwe continues to be burdened by macroeconomic instability. High inflation, exchange rate distortions, and a difficult business environment raise the cost of doing business, leading to underinvestment, a rise in informal activity (Van Nguyen et al., 2022), and erosion of the fiscal revenue base (Caballe & Panades, 2004; Tanzi, 1978). As such, macroeconomic instability significantly undermines the economy’s long-term growth prospects (Ahiadorme, 2022). Distortive monetary policy (through quasi- fiscal operations leading to high money supply growth) and exchange rate policy (through auction system manipulation and exchange rate controls) contributed to hyperinflation and rapid exchange rate depreciation (World Bank, 2024). In 2023 alone, reserve money increased by more than 1800 percent, ZWL inflation was at around 700 percent, and by March 2024 the official exchange rate had depreciated by more than 95 percent since December 2023, and the parallel market gap was over 50 percent (RBZ, 2023, 2024). These monetary and exchange rate distortions ultimately contributed to the demise of the Zimbabwean dollar in April 2024. The persistence of these macroeconomic distortions also raises questions about the fiscal costs of stabilization. Macroeconomic instability, driven by monetary and exchange rate policy distortions, is costly to domestic resource mobilization in developing countries and continues to widen the fiscal financing gap in these countries (Avellan et al., 2024). While the government may consider seigniorage revenue (inflation tax revenue) beneficial in the short term as it helps finance its expenditures, the resultant hyperinflation can have a significant negative impact on the inflation tax base and government tax revenue due to lags in tax payments (McIndoe-Calder, 2017; Tanzi, 1978; Cagan, 1956). Policymakers may see money printing as beneficial in the short run, but this may ultimately cause significant revenue losses in the medium- and long-term, which are often less visible (Cagan, Ibid). While Zimbabwe has not seen a major decline in tax revenue2, tax collection has been relatively stagnant in absolute terms (Figure 1A) and has seen a relative decline in 2023 (Figure 1B). It is thus likely that tax collection would have been higher in the absence of macroeconomic distortions. This raises the importance of identifying the magnitude of the fiscal losses brought about by monetary and exchange rate policy distortions. This paper aims to describe the monetary and exchange rate distortions channels affecting Zimbabwe and estimate the impact and size of each channel on government revenue. The paper focuses on three distinct channels, namely: loss in tax revenue from inflation-related payment lags (the Olivera-Tanzi Effect), loss in customs duty revenue due to an overvalued official exchange rate and loss in tax revenue due to informalization of the private sector. Although monetary and exchange distortions may significantly reduce government expenditures through a reduced wage bill and delayed supplier payments, in this paper we only consider the revenue impacts of monetary and exchange rate distortions for three main reasons. First, we assume that governments are not crooks who give a living wage to their workers using their right hand and then take it using their left hand. In other words, if improving public workers’ welfare is one of the government’s objectives, then inflation-driven wage bill reduction cannot be considered a fiscal benefit, but rather another form of a loss to the government’s intended outcomes. Second, suppliers are assumed to be rational. As such, they forward price their goods and services in an inflationary environment by accounting for inflation and payment lags in their pricing models. Hence, there are no significant gains in expenditure reduction for the government. Third, there are also some ethical reasons why we consider the revenue benefits of eliminating the distortions rather than the expenditure gains of the distortions. 2 This is likely because GoZ adopted a multi-currency system in 2020 that allows domestic transactions in foreign currency with tax paid in the currency of trade, providing the Treasury with growing foreign exchange tax revenues that limit inflation erosion. In addition, Zimbabwe also raised tax rates on VAT and CIT, and improving tax administration, to raise tax collection and outweigh any losses from macroeconomic distortions. 4 For instance, it is unethical for the government to deliberately devalue the currency and increase inflation for the objective of reducing its wage bill and suppliers’ bill. A. Tax revenue in USD B. Tax-to-GDP ratio 6.0 5.5 18 5.2 5.3 16.7 Tax collection (USD Billion) 15.7 15.3 Tax-to-GDP ratio (%) 16 4.0 3.1 14 13.3 2.0 12 0.0 10 2020 2021 2022 2023 2020 2021 2022 2023 Figure 1. Tax revenue and tax-to-GDP. Source: Authors’ illustration using MoF data. The rest of the paper is organized as follows. Section 2 provides an overview of monetary and exchange rate policy distortions in Zimbabwe. Section 3 describes methods and materials. Section 4 presents the main results. Lastly, section 5 concludes by proffering some policy implications. II. Overview of monetary and exchange rate policy distortions in Zimbabwe Zimbabwe’s biggest macroeconomic policy distortion has been the quasi-fiscal operations (QFOs) of the Reserve Bank of Zimbabwe (RBZ). QFOs have been the financing of the government, state-owned enterprises, and private sector companies by the Central Bank, commonly financed through money printing (IMF, 2022, 2023, 2024). This is common among non-independent central banks which operate within a broader political economy and frequently face governance disruptions (Mihailovici, 2015). The RBZ has historically been deeply involved in QFOs by printing money and using this to finance external debt servicing payments, supporting state-owned enterprises (SOEs), and making transfers to the real sector through agricultural subsidies and incentives for gold production (World Bank, 2023). The RBZ has engaged in QFOs to make payments in both ZWL (through direct ZWL loans) and in USD (by borrowing from international commercial lenders, and by exploiting export surrender requirements). The most straightforward type of QFOs came from the RBZ printing ZWL to lend directly to government (as was done between 2004 and 2007, which resulted in the 2008 period of hyperinflation). In some cases, RBZ wanted to issue USD-denominated loans or grants to government, SOEs and private sector, but it could not do so by printing money directly. Instead, RBZ borrowed from offshore commercial lenders, who were then repaid in ZWL. Yet, such money creation resulted in exchange rate depreciation, creating a vicious cycle where increasing amounts of ZWL must be printed to service USD-denominated loans. RBZ also maintains a surrender requirement that obliges exporters to exchange a share of their US dollar receipts to RBZ in exchange for local currency (ZWL between 2019 and April 2024). This is supposed to be resold in full via a domestic foreign currency auction. Yet, in many cases, RBZ only partially sold this USD to the auction while utilizing the rest to service its USD-denominated liabilities (see above). It paid for such USD-export surrenders through ZWL money creation, so again resulting in a vicious cycle where 5 increasing amounts of ZWL were needed to repay the USD-export surrenders. The conceptual overview of the Bank’s QFOs is presented in Figure 2. Figure 2. Conceptual overview of RBZ’s Quasi-Fiscal Operations (QFOs). Source: Authors’ illustration. The RBZ’s QFOs led to a significant increase in ZWL money supply, resulting in high inflation and rapid exchange rate deprecation. The overall growth in reserve money is closely correlated with Zimbabwe’s inflation, and big spikes in reserve money resulted in hyperinflation in the summer of 2020 (Figure 3A) and 2023 (Figure 3B). Close-to-perfect linear relationships exist between money supply and the consumer price index (Figure 4A). Yet, excess money supply also undermines the value of the local currency, leading to rapid depreciation. This, in turn, also hurts local prices as traders price-adjust to the new, lower exchange rate, thus importing inflation. As such, there is also a strong correlation between the consumer price index and the exchange rate (Figure 4B). A. 2019-2021 B. 2022-2023 1,000 500 1,200 3,500 Reserve Money growth (%) Reserve Money Growth (%) ZWL Inflation (%) ZWL inflation (%) 800 400 1,000 3,000 800 2,500 600 300 2,000 600 400 200 1,500 400 1,000 200 100 200 500 - - - - 2019M11 2020M11 2021M11 2022M11 2023M11 2019M5 2019M8 2020M2 2020M5 2020M8 2021M2 2021M5 2021M8 2022M1 2022M3 2022M5 2022M7 2022M9 2023M1 2023M3 2023M5 2023M7 2023M9 ZWL Inflation [LHS] Reserve Money [RHS] ZWL Inflation [LHS] Reserve Money [RHS] Figure 3. Money supply increases are closely correlated with high inflation. Source: Authors’ illustration using RBZ and CCZ data. A. Money supply and inflation B. Inflation and the exchange rate 6 160000.0 160000.0 Consumer price index Consumer price index 120000.0 120000.0 80000.0 80000.0 40000.0 40000.0 0.0 0.0 Money supply (ZWL billion) Parallel exchange rate Figure 4. Zimbabwe’s money supply, inflation and exchange rate depreciation are closely correlated. Source: Authors’ illustration using RBZ and Zimstat data. Zimbabwe’s long history of macroeconomic instability has been driven in large part by QFOs. A massive increase in the use of QFOs culminated in hyperinflation from 2004 to 2008 (Munoz, 2007). This behavior continued until early 2009, when Zimbabwe chose to shift to dollarization. Yet, even dollarization did not prevent the creation of excess liquidity. Instead, around 2014 the RBZ started exploiting the domestic US dollar clearing system used to facilitate local US dollar transactions - the Real Time Gross Settlement (RTGS) to create unbacked local (“nostro”) USD deposits and utilized this money-creation to resume its QFOs. This resulted in a parallel market for Zimbabwe’s “nostro” USD. The premium paid for “real” dollars grew steadily from around 5 percent in early 2016 to 330 percent in early 2019. With this premium now unstainable in 2019, the Zimbabwean Dollar (ZWL) had to be re-introduced at an exchange rate of ZW$2.5 to each US dollar and with all locally held US dollar balances converted into Zimbabwe dollars at this exchange rate (RBZ, 2019). Yet, RBZ’s QFOs did not stop there, so the ZWL-USD official exchange rate quickly depreciated from 2.5:1 at introduction in 2019 to 51:1 in 2020, 89:1 in 2021, 372:1 in 2022, 3509:1 in 2023, and ultimately 30,674:1 on April 5, 2024 (when GoZ announced the new ZiG currency). Zimbabwe also has several exchange rate policy distortions. Excess ZWL money supply growth resulted in pressure on inflation and rapid exchange rate depreciation (RBZ, 2023). Yet, because the resulting price rises are politically problematic, GoZ also adopted exchange rate policies to slow down depreciation, notably via influencing the auction system and official exchange rate controls. The auction system was affected by “moral suasion”, which prevented the official exchange rate from reflecting supply and demand forces and resulted in a high parallel market premium. The Reserve Bank of Zimbabwe managed the foreign exchange auction market since its inception in June 2020. Initially, the auction successfully narrowed the premium to below 25 percent in August 2020. Sadly, this success was short- lived, and the auction market has since been marred by backlogs, resulting in exchange rate losses for RBZ3. The auction did not follow the path of pure price discovery, as it became clear to the market that the Reserve Bank had an exchange rate target range and any bids outside that range were rejected. This resulted in the auction rate being artificially overvalued. Consequently, the auction failed to provide true price discovery for foreign exchange and the premium between the parallel and official exchange rates persisted (Figure 5A). In 2023, the rise in the parallel premium seemed to be derived mostly from RBZ’s excess money supply growth (Figure 5B). 3 The average backlog in 2021 was 7 weeks, 15 weeks in 2022, and 37 weeks up to the end of ZWL in April 2024. 7 A. 2019-2021 B. 2022-2023 250 500 12,000 4,000 Introduction of ZWL-USD Exchange Rate Reserve Money growth (%) Reserve Money growth (%) ZWL-USD Exchange Rate 200 forex auction 400 9,000 3,000 150 300 6,000 2,000 100 200 3,000 1,000 50 100 - - - - 2022M11 2023M5 2023M11 2022M1 2022M3 2022M5 2022M7 2022M9 2023M1 2023M3 2023M7 2023M9 Official exchange rate [LHS] Parallel exchange rate [LHS] Official exchange rate [LHS] Parallel exchange rate [LHS] Reserve Money [RHS] Reserve Money [RHS] Figure 5. Official and parallel exchange rate, and reserve money growth (y-o-y change). Source: Authors’ illustration using RBZ and WB data. Zimbabwe also adopted official exchange rate control by enforcing a 10 percent trading margin limit above the interbank rate, encouraging informalization. In 2021, GoZ strictly enforced all formal businesses to issue their pricing so that their USD and ZWL prices do not exceed the official USD-ZWL exchange rate plus a maximum 10 percent margin. Failure to comply has resulted in the RBZ’s Financial Intelligence Unit issuing fines and freezing companies’ bank accounts. This meant that the implied USD prices in the formal shops based on the official exchange rate became higher compared to US$ prices charged by the informal sector. Thus, customers would ordinarily convert the USD to local currency before purchase of goods and services in the formal sector. As a result, the persistence of a premium between the official and parallel market exchange rate has resulted in a situation where informal shops (which often only accept US dollars) charge lower USD prices than formal shops (that must adhere to the official exchange rate). This, in turn, has contributed to the increased informalization of the private sector. Because exchange rate controls are often adopted in response to concerns about rising prices, one can see a strong historical relationship between high increases in inflation (as a proxy for all monetary and exchange rate policy distortions) and informality as a rising share of Zimbabwe’s GDP (Figure 6). 66.0 100000 Informal Economy (% of GDP) Annual Inflation Rate 64.0 10000 (Log Scale) 62.0 1000 60.0 100 58.0 56.0 10 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 Informality Inflation Figure 6. Inflation and informality in Zimbabwe. Source: Authors’ illustration using RBZ and World Bank data. 8 The monetary and exchange rate policy distortions described in this section each have their distinct impact on government revenue. Three distinct direct channels are identified and presented in Figure 7. Note that there are many indirect channels in which monetary and exchange rate policy distortions affect government revenue. The biggest effect is from lower growth brought about by macroeconomic instability. An overvalued exchange rate also has important price effects –making imports cheaper and exports more expensive, affecting revenue and growth. Finally, a parallel market premium incentivizes trade distortions through the under-invoicing of exports and over-invoicing of imports. These are areas for further study. In this paper, we are however concerned with the three direct channels in Figure 7. First, monetary policy distortions result in high inflation, which undermines real tax revenue collection due to tax payment lags (Olivera-Tanzi effect). While seigniorage revenue (inflation tax revenue) may be beneficial in the short term by helping government finance its expenditures, the resultant inflation has a significant negative impact on the inflation tax base and government tax revenue due to lags in tax payments (McIndoe-Calder, 2017; Tanzi, 1978; Cagan, 1956). Most taxes are paid later after the economic activity has been done. Some taxes are paid monthly while others are bi-monthly, quarterly, semi-yearly, or yearly. In hyperinflation, the real value of tax revenue received with a lag will be reduced, and so would cause significant real money losses (Cagan, Ibid). Second, exchange rate distortions have led to an overvalued exchange rate, which leads to a loss in customs duty revenue. An overvalued official exchange rate may influence government revenue in many ways. The most obvious (and direct) method is the lower customs duty revenue on imports that the revenue authority receives in the presence of exchange rate distortions. Third, exchange rate controls together with a high parallel market premium may push businesses out of the formal sector and into the informal sector. The combination of exchange rate controls and a high parallel market premium has led to a situation where informal shops (not subject to exchange rate controls) can charge lower USD prices than formal shops (that must adhere to the official exchange rate). Such competitive pressures incentivize formal private sector businesses to shift into the informal sector. This would result in government losing formal tax revenue from such firms. Figure 7. Monetary, exchange rate policy distortions and tax revenue. Source: Authors’ illustration. 9 III. Methods and materials We quantified the magnitude of each channel in which monetary and exchange rate policy distortions impact government revenue. Thus, we considered to what extent is the government tax revenue influenced by the Olivera-Tanzi effect, how much tax revenue the government is losing from an overvalued exchange rate, and how informalization affects government tax revenue. An appropriate methodology was selected for each channel. We focused on the period 2020-2023 and applied monthly data in all the channels except the informality channel. Table 1 provides a list of data requirements and sources for the channels. Monthly data for Zimbabwe is from January 2018 up to December 2023. Panel data on informality and tax revenue is from 148 countries over 28 years from 1993 to 2020. Table 1. Variables and data. Variable Explanation and possible sources Monetary aggregates Monthly data for M0, M1 and M2 – RBZ Monetary policy rates All interest rates – RBZ CPI and Inflation Monthly and annual – ZIMSTAT, RBZ Monthly tax revenue in US$ Monthly – ZIMRA, RBZ, MoFED Monthly tax revenue in ZW$ Monthly – ZIMRA, RBZ, MoFED Time lag of revenue collection by tax type CZI and ZIMRA Exchange rates Monthly and annual – RBZ Auction data Monthly bids – RBZ Government expenditures Monthly expenditure – Ministry of Finance Informal sector size (% of GDP) Time series and panel data - World Bank Tax revenue Panel data – World Bank GDP Panel data – World Bank Channel 1: Olivera-Tanzi effect and tax revenue We applied the Olivera-Tanzi approach (Tanzi, 1978) to estimate the tax revenue loss from the inflation channel. Inflation generates revenue for the government through seigniorage (inflation tax), yet at the same time, it causes a loss in the real value of taxes (adjusted for inflation) due to lags in tax payments. We first estimated the size of revenue from inflation tax gains. Next, we considered the revenue losses from the lags in tax payments. The revenue impact is the net effect between these two elements. We focused only on all domestic taxes4 paid in ZWL. In this channel, customs duty was excluded as this is usually paid without any payment lag, but the second channel separately estimated the revenue impact on customs duty from exchange rate distortions. For this, we used a monthly breakdown of tax collection by tax type in USD versus ZWL. This is critical, as Zimbabwe has seen a significant dollarization that has also affected tax collection (Figure 8). While 89 percent of taxes were collected in ZWL in early 2020, by the last half of 2023, this dropped to only 37 percent (or 30 percent if using parallel rates). Such dollarization reduces the overall impact of ZWL inflation on tax revenue, both by limiting the potential gains from inflation tax and by reducing the potential revenue losses from tax payment lags. 4Customs duty is excluded as this is usually paid without any payment lag. The second channel separately estimates the revenue impact on customs duty from exchange rate distortions. 10 100% 89% 75% 63% 50% 37% 11% 25% 0% Jan Jan Jan Jan Jan Mar Mar Mar Mar Mar Mar May May May May May Sep Sep Sep Sep Sep Sep Jul Jul Jul Jul Jul Jul Nov Nov Nov Nov Nov Nov Jan May 2018 2019 2020 2021 2022 2023 Share USD Share ZWL Figure 8. Zimbabwe tax collection in ZWL versus USD. Source: Authors’ calculations using ZIMRA data. The time lag in tax payment was estimated as the weighted average of lags from various tax lines (Table 2). The weight is the percentage share of the corresponding tax line in total government tax revenue. These are averages of monthly data from January 2018 to December 2023. Because tax payment collection lags vary across tax types, we follow Tanzi (1978) and use a weighted average lag to account for this heterogeneity. Table 2. Lags in Zimbabwe’s tax payment collection. Tax line Weight Lag in days Unweighted Weighted lag Weighted lag lag in months in months in days Individuals 17.7% 28 1 0.177 5.0 Companies 13.7% 195 7 0.959 26.7 Other income taxes 2.2% 21 0.7 0.015 0.5 Carbon Tax 0.9% 0 0 0.000 0.0 VAT 31.7% 43 1.4 0.444 13.7 Excise duties 15.1% 31 1 0.151 4.7 Other Taxes 18.7% 18 0.6 0.112 3.4 Mean lag 100% 48.05 1.67 1.86 53.92 Source: Authors’ computations. We estimated the revenue from inflation tax ( ) as the rate of inflation () multiplied by the tax base which is the real money balance (), that is, the real inflation tax revenue is: = () (1) − where = () = 0 is the real money demand as defined by Cagan (1956) and denotes inflation tax. According to Cagan (1956), real money demand changes as the rate of inflation changes and is the coefficient of the sensitivity of the real demand for money to the anticipated rate of inflation. The real inflation tax revenue is, therefore, = 0 − (2) To find , we regress money balances on inflation. When inflation is zero or in period 0, = 0 . From equation (2), inflation has two opposing effects; it increases inflation tax revenue and at the same time 11 reduces inflation tax revenue through its effect on money demand (the tax base). The Cagan money demand function was estimated to obtain the inflation tax base and estimate the inflation tax revenue. Inflation tax causes losses in the real value of taxes. We define “normal tax revenue” as the tax revenue from all sources except seigniorage. The real value of normal tax revenue is: = (1+) 0 (3) where is the real value of normal tax revenue, 0 is the effective tax burden in the initial period, = 0, is the average tax collection lag in months, and denotes normal taxes. With a tax collection average delay of months, the government revenue loss from normal taxes ( ) is, = 0 0 − (1+) (4) Inflation tax is costly to the government if the magnitude of the revenue loss from normal taxes exceeds that of its revenue benefits from inflation tax, that is, 0 − 0 − (1+) > 0 (5) Total government revenue () is the sum of inflation tax revenue and tax revenue from normal sources, that is, = 0 − + (1+) 0 (6) To analyze the revenue implications of inflation tax, we applied equation (6) to estimate the baseline total government revenue. We then identified the period in which monthly inflation was lowest between 2018 and December 2023 and utilized the initial tax revenue as 0 , money balances as 0 and was estimated using regression. With the baseline estimate of total tax revenue, we simulated the impact of a change in inflation tax. Channel 2: Overvalued official exchange rate and tax revenue To estimate the historical loss in customs duty revenue due to an overvalued official exchange rate, we computed the potential customs duty revenue in the absence of exchange rate distortions less the observed customs duty revenue under exchange rate distortions. Alternatively, presented as = − (7) where is the potential revenue from customs duty in the absence of exchange rate distortions at time and is the observed customs duty revenue under exchange rate distortions at time . This loss was calculated by multiplying taxed imports by the exchange rate premium. Import duty paid in local currency using the official exchange rate results in a revenue loss equivalent to the total imports with duties payable in local currency multiplied by the exchange rate premium (parallel exchange rate less official exchange rate). We applied tax revenue data on import duties collected in local currency to compute this loss. Channel 3: Informalization and tax revenue The most difficult channel to estimate is the loss in tax revenue due to informalization of the private sector, brought about by monetary and exchange rate distortions. To estimate this effect, we used the inflation rate as a proxy for monetary and exchange rate policy distortions and exploited its historical relationship 12 with informality. In turn, a rise in the informal sector (as a share of Zimbabwe’s GDP) is then expected to reduce tax revenue collection. We estimated this in three steps. First, we predicted the impact of inflation on informality in Zimbabwe using a vector autocorrelation (VAR) model. We chose this model as it may account for the likely endogeneity of inflation in the informalization-inflation relationship (Pesaran & Shin, 1999). On one hand, hyperinflation may push businesses out of the formal sector into the informal sector especially where there are exchange rate controls. On the other hand, monetary policy may be rendered ineffective because most informal firms are excluded from the financial systems (Levine et al., 2010). The simple VAR model for the relationship between informality () and inflation () can be expressed as: = 0 + ∑=1 − + ∑=0 − + (8) = 0 + ∑ =1 − + ∑=0 − + (9) where , , , , , are constants and and are white noise processes with time-invariant positive definite covariance matrix and zero mean. We determined the lag length using the Akaike Information Criterion (AIC). To estimate this model, we used annual inflation rates from RBZ, together with data on Zimbabwe’s informality as a share of GDP from the World Bank’s Informal Economy database. This provides “multiple indicators multiple causes model-based” (MIMIC) estimates of informal output as a share of official GDP and is available from 1993 to 2020 for 157 countries. For details, see Elgin, Kose, Ohnsorge and Yu (2021). Because data on informality is only available until 2020, we applied equation (8) to forward-project the informal economy as a share of GDP for the period 2021 to 2023. Second, we examined the effect of informality on tax revenue using a cross-country approach. Using a cross-country estimate of the elasticity of informality-to-tax revenue allows us to avoid endogeneity issues from Zimbabwe data alone and enables us to distinguish the impact of informalization on tax collection above and beyond the Olivera-Tanzi effect, the impact of an overvalued official exchange rate, and endogenous policy responses. To estimate the revenue effect of informality growth, we again utilized the World Bank’s Informal Economy database and combined this with the GRD panel dataset from UNUWIDER on tax revenue as a share of GDP to create a panel database for 148 countries across 28 years. From this, we estimated a fixed effects regression model of the form: = + + + + + + (10) where is tax revenue of country at time , is the size of informality of country at time , is income (control variable) of country at time , is a regional dummy variable (also a control variable). is a regression intercept, , and are slope coefficients, and is an error term such that ~(0, 2 ). The coefficient measures the revenue effect of informalization. and are country- specific and time-specific effects, respectively. Appropriate tests for model (10) were done to check the suitability of a fixed effects model over a random effects model. Finally, we applied the estimates from the first and second steps to define a counterfactual scenario for Zimbabwe with low inflation for 2020-2023. We first considered the counterfactual size of the informal economy (as a share of GDP) in the absence of high inflation. Next, we applied a cross-country estimate of the elasticity of informality-to-tax revenue to estimate Zimbabwe’s counter-factual tax-to-GDP ratio. 13 IV. Results Olivera-Tanzi effect (cost of inflation) To estimate inflation tax revenue, we applied the tax base (demand for real money balances) and the response of money demand to inflation. The time series properties of the variables (money balances and inflation) used to estimate the response of money demand to inflation are presented in Figure 9 and Table 3. Only stationary series were used to estimate the relationship between inflation and money demand. 15000 .4 .3 10000 Monthly inflation CPI .2 5000 .1 0 0 2018m1 2019m7 2021m1 2022m7 2018m1 2019m7 2021m1 2022m7 Month Month .8 25 .6 Change in base money 24 Log of base money .4 23 .2 22 0 -.2 21 2018m1 2019m7 2021m1 2022m7 2018m1 2019m7 2021m1 2022m7 Month Month Figure 9. Monthly inflation and money balance trends. Source: Authors’ illustration. Table 3. Stationary results. Variable ADF test statistic p-value Conclusion Log of money 0.452 0.9833 nonstationary Log of CPI 0.113 0.9669 nonstationary Change in the log of money -8.663*** 0.0000 stationary Change in the log of CPI -3.444*** 0.0096 stationary *, **, and *** means statistically significant at the 10%, 5% and 1%, respectively. Source: Authors’ computations. The effect of inflation on money demand, , was estimated using a simple time series model with stationary series. We found this effect to be negative 0.55, that is = −0.55 (see Figure 10). This is closer to the one estimated for Zimbabwe by McIndoe-Calder (2017), −0.46. Cagan found it to be −1. Unlike Cagan’s findings, the response of money demand to inflation is not one-to-one in Zimbabwe. The regression of money demand on inflation provides a coefficient 0 of 24.4 billion. 14 .4 .2 0 -.2 -.4 0 .1 .2 .3 .4 Inflation growth dm Fitted values Figure 10. Relationship between inflation and money demand. Source: Authors’ illustration. Using the estimate from the Cagan approach, the estimated total tax revenue gained from inflation tax by the government from January 2020 to December 2023 is a cumulative total of US$1.4 billion (Figure 11A, Table 4). We checked for internal validity and established that this figure is close to the directly computed seigniorage revenue using a change in money supply approach. Computing inflation tax revenue assuming no effect on the tax base (real money balances) as in the accounting approach, the cumulative inflation tax revenue from January 2020 to December 2023 is US$1.57 billion, which is similar in magnitude to the Cagan approach. The central part of inflation tax estimation is the response of money demand to inflation. Hence, the measurement of the elasticity parameter (α) determines the quality of the estimates. Previous work on the demand function by McIndoe-Calder (2017), Munoz (2007) and Kovanen (2004) estimated the effect of inflation on money demand using OLS. This paper applied OLS and VAR estimates. They fall within the range of previous estimates. This is an indication of an internally valid elasticity parameter and its associated findings since the other part of the O-T revenue function is not parametric but mechanical. The authors also applied a mechanical method to check for the consistency of the estimates. The estimates still fall within the range of what the authors obtained using the mechanical approach. The analysis further suggests that from January 2020 to December 2023, Zimbabwe lost about US$2.8 billion because of hyperinflation-driven fall of value (Figure 11A, Table 4). To estimate the average monthly normal tax losses, the assumption is that the tax revenue received at time must have been received at time − 1.86 (the average lag). Therefore, the real value of the revenue received at time is given by (1+)1.86 and the loss caused by inflation is − (1+) 1.86 . The net fiscal costs from payment lag-driven losses in tax revenue significantly outweighs benefit from inflation tax (Figure 11B, Table 4). The analysis suggests US$1.4 billion in cumulative losses occurred between 2020 and 2023. As expected, most of the benefits accrue to the initial period (as inflation tax benefits often appear upfront, while the costs are experienced subsequently) and when dollarization is still limited – and so 2020 is the only year where inflation tax revenue was higher than the lag-induced tax revenue losses. Yet, every year thereafter the costs increased (in line with rising ZWL inflation rates). A. Inflation tax revenue benefits and lag- B. Net impact of Olivera-Tanzi effect on tax induced tax revenue losses (US$ million) revenue (US$ million) 15 400 200 Impact of inflation on tax revenue Net impact of inflation on tax 200 100 0 (USD million) (USD million) 0 revenue -100 -200 -200 -400 -300 2020M10 2021M10 2022M10 2023M10 2020M1 2020M4 2020M7 2021M1 2021M4 2021M7 2022M1 2022M4 2022M7 2023M1 2023M4 2023M7 -400 2020M10 2021M10 2022M10 2023M10 2020M1 2020M4 2020M7 2021M1 2021M4 2021M7 2022M1 2022M4 2022M7 2023M1 2023M4 2023M7 Inflation tax revenue benefit Lag-induced tax revenue loss Figure 11. Monthly benefits and costs of inflation on tax revenue. Source: Authors’ computations. Table 4. Net revenue effect of the Olivera-Tanzi effect. Year Inflation tax revenue Lag-induced tax Net revenue effect benefit (US$ million) revenue loss (US$ from inflation (US$ million) million) 2020 1,134 -573 561 2021 129 -362 -233 2022 123 -761 -638 2023 27 -1,099 -1,072 Cumulative sum 2020-2023 1,414 -2,796 -1,382 Share of GDP 2020-2023 (%) 1.14% -2.25% -1.11% Source: Authors’ computations. Overvalued official exchange rate The estimated monthly customs duty direct tax revenue loss ranged from US$3.36 million to US$27.49 million between January 2020 and December 2023. It averaged US$12.13 million with a standard deviation of US$0.96 million. Monthly customs duty revenue losses are illustrated in Figure 12. A. Customs revenue with and without B. Loss from exchange rate distortions (US$ exchange rate losses (US$ million) million) 50 30 Customs revenue 40 25 (US$ million) Customs revenue 30 (US$ million) 20 20 10 15 - 10 2020M10 2021M10 2022M10 2023M10 2020M1 2020M4 2020M7 2021M1 2021M4 2021M7 2022M1 2022M4 2022M7 2023M1 2023M4 2023M7 5 - 2020M10 2021M10 2022M10 2023M10 2020M4 2021M4 2020M1 2020M7 2021M1 2021M7 2022M1 2022M4 2022M7 2023M1 2023M4 2023M7 Customs revenue without exchange rate distortions Customs revenue with exchange rate distortions Figure 12. Customs duty revenue loss from exchange rate distortions. Source: Authors’ computations and illustration. 16 The analysis suggests that GoZ lost cumulative tax revenue from import duty of at least US$582.22 million between January 2020 and December 2023. In 2023, exchange rate distortions created a loss of US$130 million in customs duty revenue. Table 5 provides the conservative direct estimates of customs duty tax revenue losses resulting from exchange rate distortions. Table 5. Cumulative customs duty tax revenue losses (US$ million). Observed customs Potential customs duty at Net effect Year duty (US$ million) parallel rate (US$ million) (US$ million) 2020 161 250 -89 2021 256 428 -171 2022 262 454 -191 2023 270 400 -130 Cumulative sum 2020-2023 949 1,532 -582 Source: Authors’ computations. Informalization Inflation increases informality which in turn influences tax revenue. Both the Zimbabwean and global data show a positive association between inflation and the size of the informal sector (Figure 13). The time series properties of inflation and informality are presented in Table 6. Table 6. Stationary results. Variable ADF test statistic p-value Conclusion Log of inflation -2.329 0.1628 nonstationary Informality (% of GDP) -1.489 0.5388 nonstationary Annual inflation growth -4.943 *** 0.0000 stationary Informality growth -5.030*** 0.0000 stationary *, **, and *** means statistically significant at the 10%, 5% and 1%, respectively. Source: Authors’ computations. A. Zimbabwe B. Global 80 200 60 100 Informality (% of GDP) Informality growth 40 0 -100 20 -200 0 -100 0 100 200 300 -50.00 0.00 50.00 100.00 Inflation growth Annual inflation (%) Informaility growth Fitted values Informality_%GDP Fitted values Figure 13. Association between inflation and informality. Source: Authors’ computations. 17 Findings from the VAR model show that inflation influences informal sector growth with a lag in Zimbabwe. The inflation elasticity of the informal sector size is 0.0072. As such, a percentage increase in inflation in the current period is associated with an increase in informality of 0.0072 percent in the next period. Alternatively, a 100 percent increase in inflation this year will expand the size of the informal sector by 0.72 percent next year. This is a sizeable figure, especially for economies experiencing hyperinflation (for instance annual inflation increased by over 1000 percent in Zimbabwe between 2018 and 2023). A 7.2 percent growth of the already large informal sector is very substantial in 5 years. The results of the VAR model are presented in Table 7. The VAR diagnostic tests are in Annex A. Table 7. Impact of inflation on informality. Dependent variable Regressor Coefficient Standard t- p-value error statistic Log of informality Lag of informality 0.5552 0.0972 5.73 0.000*** size Lag of inflation log 0.0072 0.0014 5.24 0.000*** Intercept 1.7905 0.3941 4.54 0.000*** Log of inflation Lag of informality -7.8986 11.9304 -0.66 0.508 (non-negative) Lag of inflation 0.7018 0.1701 4.13 4.13*** Intercept 3.9249 48.5080 0.70 0.484 *, **, and *** means statistically significant at the 10%, 5% and 1%, respectively. Source: Authors’ computations. Next, we estimated the impact of informality on government’s tax revenue as a share of GDP. A scatter plot from data from 148 countries over 28 years shows a strong negative association between tax revenue and informality (Figure 14). Using the panel data model to estimate the relationship between informality and tax revenue, the findings reveal that a percentage increase in informality reduces tax revenue by 0.45 percent. The results of the panel data model and its tests are in Annex B. The negative relationship between informality and tax revenue is also consistent with previous studies by Ordóñez (2014) and Ihrig & Moe (2001). Furthermore, the authors conducted several diagnostic tests to check the fitness of the functional forms in parametric models. 60.00 Tax revenue (% of GDP) 40.00 20.00 0.00 0 20 40 60 80 Informality (% of GDP) Tax_%GDP Fitted values Figure 14. Tax revenue and informality association. Source: Authors’ illustration using global data. 18 The analysis suggests that informalization resulted in a loss of at least US$ 1.15 billion between 2020 and 2023. The first and second steps suggest that Zimbabwe’s high inflation had a significant impact on Zimbabwe’s informality and tax-to-GDP ratio with a 1-year lag (Table 8). Monetary and exchange rate distortions caused between 100 and 650 USD million per annum in lost tax revenue. This is a conservative estimate, as it assumes that informality adjusts on a year-by-year basis, while empirical estimates suggest that informality is an extremely persistent phenomenon and so while an initial shock can significantly increase the share of private sector operating in informality, it can take years to bring them back to the formal sector. Table 8. Estimated loss in tax revenue due to informalization of the private sector. Estimated lagged effect on Estimated lagged Total losses GDP (US$ ZWL annual informalization share of GDP effect on tax-to-GDP (US$ billion) Year inflation (%) (%) (%) million) 2019 255.3 n/a n/a n/a n/a 2020 557.2 1.84 0.83 23.6 195 2021 98.5 4.01 1.81 35.98 650 2022 193.4 0.71 0.32 31.05 99 2023 682.2 1.39 0.63 33.53 210 Average 2020-2023 1.99 0.90 289 Cumulative sum 2020-2023 1,154 Source: Authors’ computations. Note: ZWL inflation for 2023 is not available from RBZ, and so was proxied using data from CCZ. Aggregate cost of monetary and exchange rate distortions Overall, the analysis suggests that Zimbabwe’s treasury lost a cumulative US$3.12 billion from January 2020 to December 2023 due to monetary and exchange rate policy distortions. Due to the conservative assumptions used in the analysis, this is likely significantly to provide a minimum fiscal cost. A breakdown by year shows that the benefits from these distortions only outweighed their costs to government of Zimbabwe in 2020, while every other year distortions cost US$900 million-1.4 billion (Table 9). On average, this cost accounts for 2.5 percent of GDP, or 16 percent of tax revenue. The biggest loss comes from high inflation and the Olivera-Tanzi effect (US$ 1.38 billion), followed by informalization (US$1.15 billion) and customs duty foregone (US$580 million). These estimates suggest that in the absence of such monetary and exchange rate distortions, tax revenue in 2023 could have been as high as US6.7 billion (Figure 15A) or 20 percent of GDP (Figure 15B). Critically, monetary and exchange rate distortions are also a re- distribution from the Ministry of Finance (loss of tax revenue) towards the Reserve Bank of Zimbabwe (through gains from seigniorage). To account for the counterfactual tax-to-GDP ratio, we use only the tax revenue losses from inflation (rather than the net revenue effect that discounts seigniorage). Table 9. The aggregate cost of monetary and exchange rate policy distortions. CHANNEL (US$ million) Share of Share of Net effect of Overvalued official Informalization Total GDP Tax Year inflation exchange rate Revenue (Olivera-Tanzi Effect) 2020 561 -89 -195 277 1.2 8.8 19 2021 -233 -171 -650 -1,054 -2.9 -19.1 2022 -638 -191 -99 -928 -3.0 -17.9 2023 -1,072 -130 -210 -1,412 -4.2 -26.9 Cumulative -1,382 -582 -1,154 -3,118 -2.5 -16.3 2020-2023 Source: Authors’ computations. A. Tax revenue in USD B. Tax-to-GDP ratio 8.0 6.7 6.7 22.0 20.0 6.2 Tax collection (USD Billion) 5.5 5.3 20.0 6.0 5.2 4.0 18.0 15.7 4.0 3.1 16.0 14.0 2.0 12.0 0.0 10.0 2020 2021 2022 2023 2020 2021 2022 2023 Tax collection current Tax-to-GDP ratio current Tax collection without monetary and exchange rate Tax-to-GDP ratio without monetary and exchange rate distortions distortions Figure 15. Aggregate effect of monetary and exchange rate distortions on tax revenue and tax-to-GDP. Source: Authors’ computations. V. Conclusions and Policy Implications This analysis suggests that Zimbabwe’s monetary and exchange rate distortions may be detrimental to its revenue collection efforts through three distinct channels: 1) loss in real tax revenue from inflation-related payment lags (Olivera-Tanzi Effect). While governments gain from seignoirage (inflation tax), time lags in tax payments cause high inflation to erode the real value of tax revenue, 2) loss in customs duty revenue due to an overvalued official exchange rate. Exchange rate distortions can lead to an overvalued exchange rate, leading to lost customs duties on imports, and 3) loss in tax revenue due to informalization of the private sector. Monetary and exchange rate policy distortions increase informality thereby undermining revenue collection. This paper only considers channels linked to revenue mobilization. Yet, such distortions also significantly affect costs on the spending side (e.g. supplier price hedging, rising public wages). This is a potential area for further study. This paper aimed to estimate the size of each of the three channels of government revenue. The analysis suggests that Zimbabwe’s treasury lost a cumulative US$3.1 billion from January 2020 to December 2023 due to monetary and exchange rate policy distortions (a conservative estimate). On average, this cost accounts for 2.5 percent of GDP, or 16 percent of tax revenue. The biggest loss comes from high inflation and the Olivera-Tanzi effect (US$ 1.4 billion), followed by informalization (US$1.2 billion) and customs duty foregone (US$580 million). These estimates further suggest that in the absence of such monetary and exchange rate distortions, tax revenue in 2023 could have been as high as 19.9 percent of GDP. The main implication of the findings is that even over a relatively short time horizon of three years, maintaining fiscal discipline is a far superior strategy for dealing with chronic economic stress than 20 resorting to inflationary financing and exchange rate distortions. The paper highlights the shrinking real value of the domestic currency monetary base and how stabilization becomes increasingly crucial as inflation goes up. Inflation and exchange rate distortions cost the Treasury billions of dollars. Policies that remove exchange rate distortions and stabilize prices can substantially improve government revenue and help close the fiscal financing gap. Finally, it is important to note that removing monetary and exchange rate distortions would not immediately capture all the revenue benefits described in this paper. While the Olivera-Tanzi effect and increased customs duty could lead to swift benefits in tax collection, re-formalizing the private will likely take many years to show robust effects on tax collection. Acknowledgements This article is a product of a background paper for the World Bank’s Zimbabwe Public Finance Review 2018- 2023. We acknowledge the funding provided by the Bank to produce this paper. References Ahiadorme, J. W. 2022. “Monetary policy in search of macroeconomic stability and inclusive growth.” Research in Economics 76(4): 308-324. https://doi.org/10.1016/j.rie.2022.08.002. Avellan, L., Galindo, A., Lotti, G., and Rodriguez, J. P. 2024. “Bridging the Gap: Mobilization of Multilateral Development Banks in Infrastructure.” World Development 176. https://doi.org/10.1016/j.worlddev.2023.106498. Caballe, J. and Panades, J. 2004. “Inflation, Tax Evasion, and the Distribution of Consumption.” Journal of Macroeconomics 26(4): 567-595. https://doi.org/10.1016/j.jmacro.2003.06.001 Cagan, P. 1956. The Monetary Dynamics of Hyperinflation, In the Theory of Inflation, Michael Parkin (ed.), pp. 185-278. Elgin, C., M. A. Kose, F. Ohnsorge, and S. Yu. 2021. “Understanding Informality.” CERP Discussion Paper 16497, Centre for Economic Policy Research, London. Ihrig, J., and Moe, K. S. 2001. “Tax policies and informal employment: the Asian experience.” Asian Economic Journal 15(4): 369-383. IMF. 2024. “IMF Staff Completes 2024 Article IV Mission to Zimbabwe.” IMF Press Release No. 24/47. The International Monetary Fund, Washington DC. IMF. 2023. “Zimbabwe. IMF Country Report No. 23/361.” The International Monetary Fund, Washington DC. IMF. 2022. “Zimbabwe. IMF Country Report No. 22/112.” The International Monetary Fund, Washington DC. Kovanen, A. 2004. “Zimbabwe: A Quest for A Nominal Anchor.” Working Paper WP/04/130, International Monetary Fund, Washington D.C. Levine, P., Lotti, E., Batini, N. and Kim, Y. B. 2010. “Informal Labour and Credit Markets: A Survey.” IMF Working Paper,10/4. McIndoe-Calder, T. 2017. “Hyperinflation in Zimbabwe: money demand, seigniorage and aid shocks.” Applied Economics, DOI: 10.1080/00036846.2017.1371840. 21 Mihailovici, G. 2015. “Challenges for Central Banks Governance in the Context of a Deeper Economic and Monetary Union.” Procedia Economics and Finance 22: 442-451. Munoz, S. 2007. “Central Bank Quasi-fiscal Losses and High Inflation in Zimbabwe: A Note.” IMF Working Paper, WP/07/98. Ordóñez, J. C. L. 2014. “Tax collection, the Informal Sector, and Productivity.” Review of Economic Dynamics 17(2): 262-286. Pesaran, H., and Y. Shin. 1999. Econometrics and Economic Theory in the Twentieth Century: The Ragnar Frisch Centennial Symposium, 371–412. chapter An Autoregressive Distributed Lag Modelling Approach to Cointegration Analysis, Cambridge, UK: Cambridge University Press. RBZ. 2024. The 2024 Monetary Policy Statement. Reserve Bank of Zimbabwe: Harare. RBZ. 2024. Monetary and Financial Statistics. Reserve Bank of Zimbabwe: Harare. https://www.rbz.co.zw/index.php/research/markets/monetary-financial-statistics. RBZ. 2023. Mid-Term Monetary Policy Statement: Staying the Course to Price Stability. The Reserve Bank of Zimbabwe: Harare. RBZ. 2019. 2019 Mid-Term Monetary Policy Statement. The Reserve Bank of Zimbabwe: Harare. Tanzi, V. 1978. “Inflation, Real Tax Revenue, and the Case for Inflationary Finance: Theory with an Application to Argentina.” Staff Papers 25(3): 417-451. Van Nguyen, P., Vo, D. H., Tran, T. P. K., & Tran, N. P. 2022. “Public spending and the informal economy in Asian countries.” Cogent Economics & Finance 10(1). https://doi.org/10.1080/23322039.2022.2101220. World Bank. 2024. Country Engagement Note for the Republic of Zimbabwe (FY26-FY26). The World Bank: Washington, DC. World Bank. 2023. Zimbabwe Economic Update: Electrifying Growth through Reliable and Universal Access, Issue 4. International Bank for Reconstruction and Development / The World Bank: Washington, DC. ANNEX A. VAR Tests. Lag length test . varsoc Informality_GDP linflation Selection-order criteria Sample: 1997 - 2020 Number of obs = 24 lag LL LR df p FPE AIC HQIC SBIC 0 -94.5423 10.6924 8.04519 8.07123 8.14336 1 -68.8977 51.289* 4 0.000 1.76535* 6.24148* 6.31961* 6.53599* 2 -67.9945 1.8064 4 0.771 2.30751 6.49954 6.62977 6.9904 3 -67.5174 .95416 4 0.917 3.16529 6.79312 6.97543 7.48032 4 -65.0346 4.9656 4 0.291 3.74608 6.91955 7.15395 7.80309 Endogenous: Informality_GDP linflation Exogenous: _cons 22 The appropriate lag length of the VAR model is one according to all lag length tests. The VAR results presented in the following Table are based on these tests. The results show that the inflation tax in Zimbabwe significantly increases informality. The causal direction is one way: inflation drives informality but there is no evidence that informality drives inflation in Zimbabwe. The positive association between informality and inflation is also depicted by the scatter plot in Figure 4. . varsoc Selection-order criteria Sample: 1998 - 2020 Number of obs = 23 lag LL LR df p FPE AIC HQIC SBIC 0 -292.083 4.4e+08 25.5724 25.5973 25.6712 1 -285.67 12.826* 4 0.012 3.6e+08* 25.3626* 25.4371* 25.6589* 2 -283.761 3.8185 4 0.431 4.3e+08 25.5444 25.6686 26.0381 3 -279.484 8.555 4 0.073 4.3e+08 25.5203 25.6941 26.2115 4 -278.626 1.7157 4 0.788 6.0e+08 25.7935 26.017 26.6822 Endogenous: inforgr inflgr Exogenous: _cons 23 . var inforgr inflgr, lags(1/4) Vector autoregression Sample: 1998 - 2020 Number of obs = 23 Log likelihood = -278.6257 AIC = 25.79354 FPE = 5.96e+08 HQIC = 26.01704 Det(Sigma_ml) = 1.14e+08 SBIC = 26.68219 Equation Parms RMSE R-sq chi2 P>chi2 inforgr 9 85.5154 0.6120 36.28137 0.0000 inflgr 9 208.076 0.1779 4.978791 0.7598 Coef. Std. Err. z P>|z| [95% Conf. Interval] inforgr inforgr L1. -.3080405 .2080673 -1.48 0.139 -.7158449 .099764 L2. -.0313165 .1932115 -0.16 0.871 -.410004 .3473711 L3. .2183888 .159272 1.37 0.170 -.0937786 .5305563 L4. -.1374965 .1476775 -0.93 0.352 -.4269392 .1519462 inflgr L1. .4405567 .0864352 5.10 0.000 .2711468 .6099666 L2. .3375545 .1211171 2.79 0.005 .1001694 .5749396 L3. .2468799 .1296468 1.90 0.057 -.0072232 .500983 L4. .0611182 .1170296 0.52 0.601 -.1682555 .2904919 _cons 8.413579 14.07274 0.60 0.550 -19.16848 35.99564 inflgr inforgr L1. -.5507591 .5062692 -1.09 0.277 -1.543029 .4415104 L2. -.0075496 .470122 -0.02 0.987 -.9289718 .9138727 L3. .2406807 .3875406 0.62 0.535 -.5188849 1.000246 L4. .1223679 .3593289 0.34 0.733 -.5819037 .8266395 inflgr L1. -.0372068 .210314 -0.18 0.860 -.4494147 .3750012 L2. -.0721801 .294702 -0.24 0.807 -.6497853 .5054252 L3. -.026025 .3154565 -0.08 0.934 -.6443085 .5922584 L4. -.0070071 .2847562 -0.02 0.980 -.5651189 .5511048 _cons 20.42542 34.24177 0.60 0.551 -46.68721 87.53806 24 . var inforgr inflgr, lags(1/1) Vector autoregression Sample: 1995 - 2020 Number of obs = 26 Log likelihood = -322.2094 AIC = 25.24688 FPE = 3.17e+08 HQIC = 25.33048 Det(Sigma_ml) = 1.99e+08 SBIC = 25.53721 Equation Parms RMSE R-sq chi2 P>chi2 inforgr 3 94.5047 0.2881 10.52371 0.0052 inflgr 3 168.858 0.1121 3.281633 0.1938 Coef. Std. Err. z P>|z| [95% Conf. Interval] inforgr inforgr L1. -.0117252 .1621237 -0.07 0.942 -.3294818 .3060315 inflgr L1. .3363273 .1036825 3.24 0.001 .1331133 .5395412 _cons 10.06366 17.45673 0.58 0.564 -24.1509 44.27821 inflgr inforgr L1. -.5244545 .2896777 -1.81 0.070 -1.092212 .0433034 inflgr L1. -.0077535 .1852568 -0.04 0.967 -.37085 .3553431 _cons 13.39856 31.19115 0.43 0.668 -47.73497 74.5321 ANNEX B. Panel data model of the relationship between tax revenue and informality. Levin-Lin-Chu unit-root tests Table B1. Stationary results. Variable Adjusted t statistic p-value Conclusion Tax revenue (% of GDP) -6.3592*** 0.0000 stationary Informality (% of GDP) -5.2050*** 0.0000 stationary GDPPC (USD) -5.2900 *** 0.0000 stationary Annual inflation (%) -960*** 0.0000 stationary *, **, and *** means statistically significant at the 10%, 5% and 1%, respectively. Table B2. The Panel Findings. (PE) (FE) (RE) VARIABLES Tax revenue (% of Tax revenue (% of GDP) Tax revenue (% of GDP) GDP) Informality (% of GDP) -0.112*** -0.455*** -0.305*** (0.0139) (0.0500) (0.0362) GDPPC_USD 8.28e-05*** -6.58e-05*** (1.11e-05) (2.39e-05) _Iregion_co_2 -13.59*** 25 (0.961) _Iregion_co_3 -3.733*** (0.871) _Iregion_co_4 -7.711*** (1.030) _Iregion_co_5 -2.720*** (0.848) _Iregion_co_6 -11.01*** (0.885) _Iregion_co_7 -7.179*** (1.015) _Iregion_co_8 -2.977*** (0.767) _Iregion_co_9 -0.464 (0.753) _Iregion_co_10 -16.07*** (0.878) _Iregion_co_11 -4.433*** (1.022) _Iregion_co_12 -8.756*** (0.825) _Iregion_co_13 -10.89*** (1.472) _Iregion_co_14 -7.443*** (0.788) _Iregion_co_15 -11.35*** (0.846) _Iregion_co_16 -10.72*** (0.789) _Iregion_co_17 -12.93*** (1.159) _Iregion_co_18 -3.194*** (0.779) _Iregion_co_19 -11.22*** (0.770) _Iregion_co_20 -13.69*** (1.003) 2.region_code -11.61*** (4.316) 3.region_code -0.769 (3.868) 4.region_code -5.915 (4.647) 5.region_code -3.519 (3.891) 6.region_code -9.309** (3.946) 7.region_code -8.970* (4.605) 8.region_code -0.963 (3.427) 9.region_code 0.445 (3.334) 10.region_code -16.21*** (4.043) 11.region_code -3.586 (4.626) 12.region_code -7.710** (3.784) 13.region_code -12.35* (6.751) 14.region_code -6.051* (3.553) 15.region_code -11.05*** (3.796) 16.region_code -9.829*** (3.584) 17.region_code -15.30*** (5.221) 18.region_code -1.759 26 (3.466) 19.region_code -9.434*** (3.398) 20.region_code -13.81*** (4.604) Constant 26.25*** 32.54*** 32.52*** (0.816) (1.787) (3.124) Observations 4,144 4,144 4,144 R-squared 0.420 0.020 Number of Code 148 148 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Tests for the most appropriate model BP test rejects pooled in favour of random effects model. The F-test also rejects pooled effects in favour of FE. Finally, the Hausman test rejects random effects (RE) in favour of fixed effects. Hence, this paper considers the results of the FE model. Nevertheless, the findings from all the estimators demonstrate that informality is negatively associated with tax revenue. . xttest0 Breusch and Pagan Lagrangian multiplier test for random effects Tax_GDP[Code,t] = Xb + u[Code] + e[Code,t] Estimated results: Var sd = sqrt(Var) Tax_GDP 82.28185 9.070934 e 17.81876 4.221227 u 35.58798 5.965567 Test: Var(u) = 0 chibar2(01) = 22152.63 Prob > chibar2 = 0.0000 . hausman fixed Coefficients (b) (B) (b-B) sqrt(diag(V_b-V_B)) fixed . Difference S.E. Informalit~P -.4545665 -.3046664 -.1499001 .034517 b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(1) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 18.86 Prob>chi2 = 0.0000 27