Financing the Sustainable Development Goals

This paper contends that carving out pathways to finance the SDG agenda entails to reconsider tacit assumptions regarding the functioning of financial systems. We first use a history of economic thought perspective to demonstrate the flaws of the loanable fund theory, which has come to underlie SDG finance strategies. We then introduce the alternative endogenous money theory using a consistent theoretical and accounting framework. This allows us to identify and discuss a set of financing mechanisms, which would permit to bridge the SDG budget gap. These mechanisms include the issuing of sovereign green bonds, the modification of the European Central Bank’s collateral framework, changes in capital adequacy ratios, a market of SDG lending certificates and the introduction of rediscounting policies. We back up the discussion with examples from economic history.


Introduction
The sustainable development goal agenda (SDGs) 2 , adopted by all 193 UN Member States on September 25, 2015, will have very significant resource implications across the world. According to current estimates, global annual capital expenditure required to achieve the SDGs is comprised between 5 and 7 trillion US dollars, which would necessitate an annual incremental increase of about 2-3 trillion US dollars, compared to current levels (SDSN, 2018). In the field of energy transition alone, the global additional investment required to keep the temperature increase below 2 °C from both supply and demand sides are estimated to about USD 800 billion/year according to McCollum et al. (2014).
In 2018, the European Commission High-Level Expert Group on Sustainable Finance (HLEG) underlined that "reaching our Paris agreement goals requires no less than a transformation of the entire financial system, its culture and its incentives" (HLEG, 2018). Academic research therefore has a role to play in identifying new financing pathways in order to deliver timely and adequate flows of 'missionoriented' (Mazzucato, 2018) public and private investment. The ultimate objective of this research agenda should be to inform action needed by policy makers, financial actors, and public and private investors in order to upscale SDG financial flows in response to sustainability demands. As argued by several authors (Lagoarde-Segot and Paranque, 2018;Muniesa, 2015;MacKenzie and Milo, 2003;Chambost and Lenglet, 2018) the behavior of actors and the legal/technical system within which they operate is shaped by underlying theoretical representations of the nature of social reality and economic knowledge. This requires to clarify the meta-theoretical assumptions underlying existing discourse and proposals.
In line with previous work (Lagoarde-Segot, 2015, Paranque and Lagoarde-Segot, 2018, we contend that bridging the SDG finance gap requires excavating -and critically assessing -a set of tacit beliefs, which seemingly underlie the current framing of finance policies 3 . In particular, we put forth that one specific conception of money and finance -the 'loanable fund theory'has come to dominate academic and policy discourse in the past decades. Paradoxically, several major economists (Schumpeter, Wicksell, Keynes) and central bankers around the world (such as the Bank of England (2014)), have repeatedly demonstrated the logical flaws of the loanable fund theory. In particular, it has been pointed out that it does not account for the mechanisms of credit creation, and provides, overall, an inaccurate portrayal of the relationship between finance, savings, interest rates and investment in the real world. We then offer an in-depth discussion of the alternative endogenous money theory championed by Post-Keynesian authors, and we show how it provides an adequate conceptual framework within which to carve out financing pathways for SDG-oriented investment.
Our central thesis is that the EDG financing gap is primarily the result of an optical illusion created by looking at sustainable finance through the prism of the loanable fund theory. The biggest obstacle to financing the EGD may not be the scarcity of money, or the unavailability of policy options, but, rather, our economic zeitgeist. We back up our argument using relevant historical examples, illustrating that the rules and conventions that government write and self-impose regarding the creation of temporary credit money can in fact be modified, in the face of demanding circumstances. In particular, we mention the financing of the American 'Arsenal for Democracy ' between 1942 and 1945, the Banque de France's collateral and rediscounting policies in the aftermath of WWII, and the Indian experience with priority sector lending since 1974.
The paper proceeds as follows. The second section documents and discusses the magnitude of the global and European 'sustainable finance gap', with a specific focus on the European Green Deal announced in 2019. The third section discusses the loanable fund fallacy from a history of economic thought perspective, and highlights its pervasive impact on current discussion on SDG finance. The fourth section introduces the endogenous money theory view of the economy. In so doing, we introduce a new consistent social accounting matrix transaction flows matrix, in the spirit of Godley and Lavoie (2012). The fifth section uses endogenous money theory as a grid to discuss a set of simple financial mechanisms, which would allow for a quick release of the amount of finance required for SDG-related transformative investments. These mechanismswhich are by no means exhaustive -include the issuing of sovereign green bonds, and the greening of money creation by banks through a modification of the European Central Bank's collateral framework, changes in capital adequacy ratios, SDG lending certificates and rediscounting policies. The sixth section brings together our conclusions.

The sustainable finance gap in the European context
The announcement of a European Green Deal (EGD) in 2019 reset the EU's commitment to tackling climate and environmental-related challenges. The EGD is an ambitious strategy, which aims to transform the EU into a "fair and prosperous society" where there are no net emissions of greenhouse gases in 2050 and where "economic growth is decoupled from resource use" (European Commission, 2019).
The EGD, however, appears to encounter a significant financing gap. According to the European Commission's estimates and projections, achieving the current 2030 climate and energy targets will require €260 billion of additional annual investment 4 . The EGD investment plan, however aims to mobilize 1 trillion euros during the next decade. This amount only represents about 38% of the investment required to achieve climate and energy targets. The funding may thus come short of the investment required to attain SDG objectives in a timely manner. Figure 1 plots the estimated cost of climate adaptation in the European Union together with the value of the components of monetary aggregates and long-term liabilities in the Eurozone. The figure shows that the cost of achieving the EU's climate and energy targets exceeds the value of debt securities of a maturity of over 2 years, the value of deposits redeemable with at notice of up to 3 months, and the value of deposits with an agreed maturity of up to 2 years. The financing gap (calculated as the difference between the European Green Deal and the global cost of reaching the climate and energy targets), in turn, exceeds the value of currency in circulation as well as the value of short-term deposits (maturity of up to 2 years). These figures show that rising to the sustainability challenge will involve a disruption of the structure and the volume of assets and liabilities in the European financial system.
In the current context, bridging the SDG financing gap appears to involve difficult policy trade-offs. The main financing arm of the EDG is indeed the EU budget, which relies mainly on VAT and GNIbased national contributions, and its modification entails distributional consequences. For instance, the introduction of levies on households' electricity and fuel consumption to fund increased contribution to the EU's budget disproportionately affect poor households, and might thus lead to an increase in inequality (Zachmann et al, 2018;Hall et.al, 2018). Another option might be to reallocate the existing budget across policy priorities. Indeed, the European Green Deal will mobilize 25% of the EU's budget (European Commission, 2019). This, however, also entails significant distributional consequences across sectors and economies, as evidenced by ongoing discussions around the future of the Common Agricultural Policy. Finally, the climate strategy may have hidden costs and contradict other SDGs and economic convergence in the EU. For instance, according to the Polish trade union Solidarnosc (2018) about 800,000 jobs in the coal industry are under threat because of the new European energy policies. Eastern European economies might be hurt disproportionately given their dependency on those industries.
In this context, a contradiction appears to emerge between the legitimacy of economic policy and the imperative of funding initiatives to achieve climate goals. This contradiction may prove destabilizing both at the European level and at the national level -as demonstrated by the 2016 Brexit campaign (which promised to return the UK's net contribution to the EU's budget to the taxpayer) and the gilets jaunes movement in France (which started out as a protest against an ecological tax on diesel).
Research is therefore needed in order carve out new mechanisms in order to to upscale SDG-related financial flows in response to sustainability demands, in a manner that is consistent with the prevailing set of political and financial constraints prevailing in the EU. In what follows, we use a history of economic thought perspective to show that the "loanable fund theory" (hereafter LFT) -which has come to underlie academic and policy discourse in finance -is a powerful hindrance to the adoption of SDG financing strategies, by virtue of the representation that it generates. We will demonstrate that the aforementioned policy trade-offs disappear when one adopts the more realistic endogenous money theory. This allows us to identify financing pathways to achieve SDG targets in a timely manner.

Figure 1 Climate adaptation, monetary aggregates and longer-term liabilities in the Eurozone
Note: data is labelled in billion euros. The first panel of the figure plots cost of climate data and is taken from the European Commission (2019). The second and third panels of the figure show the value of long-term liabilities and monetary aggregates in the Eurozone as of January 2020, respectively. Data is taken from the ECB's Statistical Data Warehouse.

The premises of the theory
The central tenet of the LFT is that investment levels are constrained by the size of a preliminary pool of savings. 5 . This idea is very intuitive: given that savings are a withdrawal from the income stream, while investment is an injection, it seems only reasonable to suppose that at the macro-economic level, savings supply the funds required for investment. This idea is also confirmed by the facts of individual experience (Fletcher, 1987, p.96). At an analytical level however, the LFT relies on a set of specific hypotheses and adjustment mechanisms.  Electronic copy available at: https://ssrn.com/abstract=3552572 The LFT views the financial system as a market where savings are exchanged for new capital assets. Money is a neutral veil (a numéraire) on real market exchanges, and the underlying stock of money is exogenous (i.e. provided by the Central Bank). In this representation, money is neutral: equilibrium in the money market (and by extension, in financial markets) merely reflects the real market equilibrium (and not the other way around). This entails a conception of financial institutions (pension, insurance and wealth funds, commercial banks, development banks, crowdfunding, venture capital etc.) as mere intermediaries, transforming money capital into fixed assets seeking a return.
The logical consistency of the LFT rests on a classical market-based adjustment mechanism, in which the interest rate -which is akin to the price of financial resources -adjusts in order to bring savings into equality with the demand for investment. In this context, any shift in the demand curve for capital or the curve relating savings to interest rate, will determine the new rate of interest, and a new loanable funds market equilibrium, corresponding to the intersection of the new positions of these curves. This can be illustrated with the following diagram:

Figure 2 The fund market view of savings, investment and interest rates
If, for example, savings increase, the supply of loanable funds increases as the S curve shifts to the right from S 0 to S 1 . Assuming that the number of profitable projects remains unchanged, this leads to an excess supply of loanable funds. The interest rate then drops from ri 2 to ri 1 which brings out new profitable investment projects and brings the fund market back into equilibrium. Unless thwarted by rigidities, the funds market equilibrium is characterized by 'such a rate of interest that savings flow into the market at precisely the same time-rate or speed as they flow into investment producing the same net rate of return as that which is paid savers for their use' (Knight, 1934).
The core idea of the LFT is that 'savings cause investment'. This idea is highly intuitive, and many observers adhere to it by default as it corresponds to the constraints governing individual budget, spending and debt.
The policy implications of the LFT are far-reaching. Its key corollary is the primacy of savings: one implication of figure 2 is that increasing corporate profits and capital income will increase the pool of savings. This, in turn, will decrease the cost of borrowed funds, and ultimately stimulate investment by increasing the marginal return on capital. Another policy corollary is the ineffectiveness of financial repression policies (McKinnon, 1973). As represented in figure 3, the LFT predicts that policies seeking to control financial markets (such as ceiling on interest rates, which would fix the interest rate to rir, below its equilibrium value ri1) would entail the apparition of credit rationing (the unsatisfied component of the demand for investment being represented in figure 3 by the gap between Dr and Sr).

Figure 3 Financial repression policies in the loanable fund theory framework
Overall, the widespread acceptance of the LFT has offered a powerful justification for liberalization, disintermediation and deregulation policies in the financial sector. It has provided a theoretical background for policies such as the decrease of taxes on capital gains, corporate profits, high salaries, and, more generally, for the 'trickle-down' economic strategies adopted since the 1980s in many OECD countries.

LFT and the framing of climate finance policies
As seen through the lenses of the LFT, the SDG finance gap is primarily an asset allocation problem. Indeed, if one is willing to accept that legitimate (public and private) investment requires preliminary savings, then the only way to bridge the sustainable finance gap is to nudge markets to allocate a greater shares of savings funds towards a new 'sustainable' segment of ecosystem-backed securities.
At the policy level, this involves prioritizing an increase in the volume of sustainable finance assets, as well as their appropriate branding (such as ESG, SRI, and impact investing…), in order to allow them to compete for funds with 'standard' assets. The comparative advantage of 'sustainable finance assets' in financial markets is founded on a double promise: first, a promise to make finance flows consistent with low emission and climate-resilient development; and second, a promise of maintaining a vigil for 'financial efficiency' (the latter being, again defined with reference to the concept of asset market 'equilibrium'). Gradually then, the SDG finance discussion has morphed into a supply-side discussion founded on the apparatus of modern finance theory, in which an additional dimension ('sustainability') is added to the array of investor decision criteria (along with price, expected rate of return, portfolio diversification, equilibrium return, risk premium…) (e.g. Baker et.al, 2018).
From a financing perspective, the fundamental problem with this approach is that the characteristics of the sustainable financial assets may depart from the fundamental value of underlying real ecosystem. The trading of ecosystem-backed securities indeed necessitates the securitization of ecosystems (Kemp-Benedict and Kartha, 2019). The latter is a process by which one assigns an economic value to the ecosystem service, establishes of a fungible proxy (commodification) offering a flow of payments and a tradable right to these payments. Given the complexity of ecosystem services, the traded assets cannot adequately represent the underlying value of nature. Instead, it tends to create a financial hyper-reality in which the underlying systemecosystemsis shaped to the image of the indicators supposed to represent it (Scott 1998, Schinckus, 2008, Robertson, 2006. For instance, applying financial logic to sustainability issues inevitably leads to maximizing the provision of the ecosystem services that yield the largest and most stable payments. The securitization strategy may thus come at the expense of other, un-commodified or un-commodifiable ecosystem flows. In addition, the market price of financial assets depends on its trades. These are determined by a wide variety of non-ecological, market-based trends and behaviors (such as trading strategies, exposure to systemic risk, etc..). Empirical research has shown the pricing of sustainable asset portfolios does not depart from that of other portfolios traded in the same market (Erragragui & Lagoarde-Segot, 2016). It follows that the price of ecosystem-backed securities has a natural tendency to depart from the fundamental value of the underlying natural asset it is supposed to represent.
Finally, the ability of the securitization policy to deliver the amount of finance at an appropriate scale and speed can be questioned. As highlighted in Lawson (2019) or Lagoarde-Segot (2019), economic and financial reality is a non-ergodic and open system. The aggregate response of financial agents to 'nudges' and labels, is therefore hard to predict given the complexity of the financial system, the emerging macroeconomic dynamics, and the resulting change in opportunities for banks and financial operators competing for profits in a globalized and technological environment.
Overall, the ecosystem-backed securitization of the environment strategy is not up to the task of providing timely and adequate financing to the SDG agenda. We contend that these problems are not a matter of labelling, incentives or calibration (i.e. what we may call 'puzzle-solving' issues (Ardalan, 2008). Rather, they are the manifestation of the tacit prevalence of the LFT as the main underlying discourse in academic and policy spheres.

The LFT as a "nonsense theory"
Ironically, the LFT has been the object of much controversy in the history of economic thought. It was first criticized by authors such as Patinkin, Wicksell, and ultimately crumbled some 90 years, ago under the attacks of John Maynard Keynes. Keynes, who unambiguously labelled the LFT a "nonsense theory" (1936, p.155) underlined two major logical flaws of this theory.
Keynes began by acknowledging that the LFT had it right in equalizing savings with investment. Defining income (Y) as the sum of consumption (C) and investment (I) (Y=C+I) and savings (S) as difference between income (Y) and consumption (C) indeed yields the following accounting equations: The fundamental mistake of the LFT, however, is to take equation (3) in isolation and to infer that savings cause investment (through an adjustment of the interest rate). While this is certainly true at the individual level (one can freely decide to increase one's savings, and thereby increase one's financial wealth and investment capacity), at the macroeconomic level, however, savings and investment are codetermined through their relationship with income (equation 1 and 2). This renders the LFT's core adjustment mechanism: as shown in figure 2, a shift of the investment curve from I0 to I1 which affects income (equation 1), could then trigger several possible movement of the savings curve S1, S2 or S3 (equation 2), leading to indeterminacy regarding the future level of interest rates. In other words, the loanable fund model fails due to the endogeneity of income.
The only way to rescue the LFT, Keynes argued, would be to assume constant income. Given equation 1, this would require us to assume that the movements of the savings and investment curves balance each other out perfectly, so that, for instance, a given drop in consumption (i.e. an increase in savings) would be perfectly offset by a corresponding increase in investment. Nevertheless, such perfectly balanced movements would leave the equilibrium interest rate permanently unchanged in figure 1making, again, the LFT collapse.
The second shortcoming of the LFT identified by Keynes has to do with the determination of interest rates. The LFT indeed assumes that the savings decision is a one-step decision based on the timepreference of agents. In reality, however, savings is always (at least) a two-step decision: once the decision on how much to save out of her current income has been made, one needs to decide on what form to hold one's savings. The most basic portfolio decision consists in determining what proportion of one's savings will be held as money (which is liquid but does not bear interest) as opposed to other less liquid, but interest-bearing assets (such as bonds or stocks). Given this fact, Keynes argued, the interest rate cannot be determined as the price which 'which brings into equilibrium the demand for resources to invest with the readiness to abstain from present consumption' but rather, as the price which 'equilibrates the desire to hold wealth in the form of cash, with the available quantity of cash' (Keynes, 1936, p.144). It follows that the savings and investment curves shown in figure 1 do not determine the level of the interest rate. Rather, they tell us what income (and therefore the aggregate volume of employment) will be, if we know, from some other source, what the interest rate is (say, for instance, ri1) (figure 3) 6 .
Rejecting the LFT has far-reaching policy implications. The most fundamental one is that savings are no longer considered to be a factor which will increase investment, but, rather, as a societal scourge, which decreases income and employment. Recent econometric work validates this view, by showing that the development of the financial sector is actually detrimental to economic growth, wages and investment in OECD countries, especially since the 2008 crisis (Gimet et.al, 2019). Policy recommendations that follow typically include the curtailing of financial markets, the taxation of multinational corporate profits, the adoption of stricter regulation of speculative activities -including secrecy jurisdictions -or the adoption of higher marginal tax rates.
But if one is willing to accept that investment and interest cannot be solely explained by savings, then one needs to frame sustainable finance policies from a different perspective than the LFT-based issuing of ecosystem-backed securities. As we shall see in the next section, the Post-Keynesian endogenous theory of money provides a more realistic theoretical account of the relationship uniting savings, investment, and the economic system, and permits to identify fresh strategies to bridge the sustainable finance gap. 6 Of course, contemporary economic and financial systems are much more complex than those prevailing in Keynes' days. Recent theories of interest rates underline a wide array of explanatory factors including banklending practices, the demand for money, the monetary policies of the Central Bank, prudential regulations, and the shape of the yield curve (Lavoie, 2015). Keynes' core message, nonetheless, remains valid: the determination of interest rates obeys economic forces more complex than the time-preference of savers and the 'supply' of capital assets.

A transaction flows matrix
Following the seminal work of Godley and Lavoie (2012), we use a consistent theoretical and accounting framework, to present an up to date explanation of the endogenous theory of money. We begin by realistically integrating the real and financial sector side of a simplified economy using a transaction flows matrix (table 1). Each economic transaction in the matrix is double-sided, i.e. one's income corresponds to another's spending, and one's asset is always another's liability. Changes to the stock of assets and liabilities by each sector (in columns) result from its budget constraint, and the latter depends on flows of income and spending (in rows).
The coherence of the matrix, in which a (+) sign indicates a source of funds and a (-) sign indicates a use of funds, is built on double entry bookkeeping logic: each row and each column must sum to zero. At first glance, the matrix reveals that savings and investment appear, not as the determinants of the economic system, but rather as their twin determinates. For instance, positive household savings is a surplus of income over spending which materializes as assets appearing in the balance sheet of households, and as liabilities in the balance sheet of the issuing sector. Whether a given sector issues liabilities (or accumulates assets) at the end of each period, however, depends on a complex set of interrelated macroeconomic, behavioral, institutional factors, which determine the flow of transactions, in a manner, which is always consistent with accounting rules. The LFT conception of money as a 'given stock', which has no counterpart in the rest of the economy, hence appears meaningless.
The matrix comprises four sectors: the household sector, the production firm sector, the public sector (made of a government and its Central Bank) as well as a relatively sophisticated financial sector based on recent development by Bouguelli (2019). The financial sector comprises banks and a 'shadowbanking sector' comprising money market mutual funds, broker-dealers and special purpose vehicles). Macroeconomic income appears a memo, in line with equation (1): the excess of income over consumption (i.e. savings) cannot differ from the addition to investment. Column 1 shows the budget constraint of households. Households receive various flows of income such as wages (W) dividend payments (from equity holdings in production firms ( ), banks ( ), and money market mutual funds ( )), and interests on their banking deposits ( (−1) ℎ(−1) ). They spend their income on consumption (C), interest payment on their loans ( (−1) h(−1) ) and tax payments net of transfers ( ℎ ). Any surplus of income over spending adds to their stock of assets. The latter comprises cash (∆ ℎ ), bank deposits (∆ ℎ ), and equities, which are issued by banks (∆ ), production firms (∆ ) and money market mutual funds (∆ ). Any excessive spending relative to income implies a decrease in assets or an increase in household debt (∆ ℎ ).
Column 2 shows the receipts and outlays of production firms in their current account. Firms receive payment flows on their sales of final goods (C) and capital goods (I), pay wages (W), taxes ( ) and interest on their bank loans ( (−1) (−1) ). Column 3 shows that firms' capital expenditure are financed via retained earnings ( ), new loans (∆ ℎ ) or equity issues (∆ ).
Columns 4 to 8 describe a financial sector comprising banks and a simplified 'shadow banking' system with money market mutual funds, broker dealers and special purpose vehicles.
Column 4 and 5 show the current account and the capital account of banks. Banks receive interest payments on loans ( (−1) (−1) ), on their T-bill holdings ( (−1) (−1) ), and pay interest on their clients' deposits ( (−1) (−1) ). The difference constitutes their profit ( ), which is split between dividend payments ( ) and retained earnings ( ). These retained earnings, along with the new deposits (∆ ) and equity issues (∆ ) appear on the liability side of banks' balance sheet. They are the counterpart of the assets owned by banks, which include granted loans (∆ ), high-powered money (cash and reserves) (∆ ) and T-bills (∆ ).
Column 6 describe the money market mutual funds. MMMF issue equities (∆ ), in exchange for banking deposits ( ∆ ℎ ). They also hold Treasury bills (∆ ), which they purchase from brokerdealers in reverse repos operations. They distribute all their profits to households ( ).
Column 7 and 8 describe broker-dealer (BD) entities and special purpose vehicles (SPV), which are both subsidiaries of commercial banks. Broker-dealers hold T-bills (∆ ), which they sell to MMMF in repos operations in exchange for banking deposits (+∆ ℎ ), and hold MBS shares issued by special purpose vehicles (∆ ) . Special purpose vehicles issue mortgage-backed securities (MBS) (∆ ), and use the proceeds to purchase loans from the banks (∆ ) . Profits of brokerdealers and SPVs ( and ) are distributed back to the banking sector.
Column 9 describes the budget constraint of the government. It shows that any investment expenditure ( ) that is not financed by taxes ( ) (or by Central Bank dividends ) must be financed by an issue of T-bills (∆ ). Following Godley and Lavoie (2012), in our matrix the interest on Tbill paid by the government to the Central Bank are returned to the government, so that net interest disbursement are : Column 10 and 11 represent the current account and the capital account of the Central Bank. The Central Bank owns T Bills (∆ ) and its main liability is the high powered money (cash and reserve currency) which it issues (∆ ). Any addition to the bond portfolio of the Central Bank must be accompanied by an increase in the amount of high-powered money. Central Bank profits (equal to T bill rate) are transferred to the government (Fcb).

Credit, finance and savings: the endogenous money perspective
The transactions flow matrix shown in table 1 is a useful device with which to analyze the financing of the economy, from an endogenous money perspective. This starts by acknowledging that private banks do not lend preexisting funds, but instead create new credit money every time they grant a loan (Bank of England, 2014). Examining the actual functioning of a banking system shows that the causality relationship does not run from savings to investment (as indicated by the LFT) but, in fact, in the opposite direction.
We may illustrate this through the simplest case, where a bank extends a loan to a firm that wishes to extend its production capacity. Table 2 accordingly focuses on the relevant entries in the transaction matrix. It shows that credit creation entails four balance sheet entries: a new deposit enters the bank's balance sheet on the liability side and the borrower's balance sheet as an asset; and a new loan, which enters the bank's balance sheet as an asset and the borrower's balance sheet as a liability 7 . As opposed to popular belief, there is no ex-ante intermediation when a bank grants a loan, since both sides of the balance sheet (the loan/the deposit) involve the same client (Jakab and Kumhof 2015).
As soon as the firm undertakes investment by spending its new deposits on wages (W), macroeconomic income increases (in line with equation 1). New deposits transferred to the bank account of the households providing the goods and services to the firm 8 . These deposits are income not spentwhich means, at an accounting level, that savings go up.
As households provide labor in exchange for the salaries, firms accumulate inventories, which appear on their capital account (I) as the ex-post counterpart of the bank loan. Finally, households spend a part of their income on consumption (C). This allows firms to retrieve their money balances and to repay the initial bank loan. At this point, the credit money, which allowed production in the first place is destroyed, as the bank's balance sheet shrinks on both sides. This mechanism is fully consistent with equations (2) and (3): following an increase in investment, income will increase to a level, which makes the change in saving equal to the change in investment. At the macroeconomic level, net accumulation of savings necessitates a preliminary increase in debt-financed investment (and not the other way around) 9 .

Changes in deposits
−∆ (−∆ ) +∆ 7 There are, of course limits to the power of banks to create money. Banks must meet profitability constraints in order to remain competitive and must ensure that the firm has a capacity to repay. They must also abide by prudential regulations, in particular capital requirements. Finally, banks must be able to acquire reserves at a low cost if the firm wants to spend its deposits. Therefore, even though banks can create unlimited amounts of deposits, they have no incentive to do so because it may expose them to both insolvency and illiquidity risks. 8 As highlighted in Bouguelli (2019), the loan's counterpart is the saver's deposit but there should be no confusion regarding causality: the loan is funded by the deposit and financed by an ex-nihilo money creation (Lavoie 2016, 65). The bank can be thought of as an 'ex-post' intermediary (Lavoie 2016;Unger 2016). 9 Denis (1999) offers a simple and elegant demonstration. Letting W represent the wage bill, S represents household savings, and I represents firms' net investment in a given year, the economy's sale of final goods is equal to household consumption (W-S) plus net investment (I). Firm profits π are, in turn, are equal to the difference between sales (W-S+I) and the wage bill W, i.e. π =I-S. Profits thus correspond to the share of firms' net investment which is not financed by savings.
In the Post-Keynesian view, finance and savings are therefore two very different things. Finance is a flow of credit determined on the basis of entrepreneurial expectations, while savings are a residual stock caused in the first place by macroeconomic dynamics. This has practical implications exactly opposed to those implied by the LFT. In particular, increase in savings by the household sector (a 'thrift campaign') does not result in increased investment. Instead, it implies a direct loss of income for production firms (as consumption spending (C) decreases). This loss is unlikely to trigger new investment plans. It is more likely to lead to the formation of pessimistic entrepreneurial expectations, which reduces the investment rate, and, ultimately, the accumulation of financial wealth -i.e. savingsat the next period 10 .

Incorporating 'shadow banking' into the analysis
The development of the so-called 'shadow banking system' (also called 'non-bank finance' or 'market based' finance) is an important feature of the modern economy (Lagoarde-Segot, 2015). In the endogenous money perspective, shadow banking does not provide additional financing to the economy, but adds layers of debts within the financial sector, based on a pre-existing stock of banking deposits. As indicated in Bouguelli (2019, p.17): "The raw materials used by the shadow banking system to produce securities are created by traditional banks, so the former cannot expand if the latter does not grow in the first place".
Assume for instance that households decide to rebalance their portfolio and buy shares issued by the money market mutual fund (MMMF). A fraction of banking deposits ∆ (which were created, in the first place, through production credit, as in table 2), are then transferred from the household's account to the MMMF's account in exchange for shares (∆ ). These deposits still appear on the liability side of the banking sector's balance sheet.

Changes in T-bills Changes in MBS shares Changes in loans −∆
The MMMF then enters into a reverse repurchase agreement with a broker dealer. The broker dealer sells a T-bill to the MMMF (∆ ), agreeing to repurchase it at a future date. The banking deposits now appear on the asset side of the broker-dealer's balance sheet. This situation does not last very long, as the broker-dealer uses these acquired deposits to purchases MBS issued by a special purpose vehicle:

Changes in loans −∆
The SPV, in turn, uses the deposits to buy packages of loans from the commercial bank. As a result, the bank moves the loans off its balance sheet, which destroys a corresponding amount of credit money.

Changes in loans −∆
In summary, the shadow banking system entails the issuing of a chain of liabilities (ownership rights and a portfolio of pre-existing loans (MBS)) against an initial flow of banking deposits. This scheme permits banks to move loans out of their balance sheet, and gives them incentive to finance riskier projects. In the endogenous money perspective, the shadow banking system adds layering, complexity, and fragility to the financial system, without altering the fundamental macroeconomic causality running from credit to deposits and from investment to savings. Paraphrasing Keynes, "the investment market can become congested through a shortage of cash. It can never become congested through a shortage of savings." (1973, p.222). The next section attempts to carve out policy responses to the sustainable finance gap using endogenous money theory as a conceptual framework.

The case for SGB issues
Sovereign green bond (SGBs) issues would probably be the most direct way to make available the money needed to tackle the climate crisis, whilst avoiding pitfalls related to the securitization of ecosystems. The direct counterpart of SGBs would be vigorous action geared towards energy transition (such as renovation and energy efficient buildings, renewable energy generation and transmission, and low-carbon transportation, to name a few), or climate resilience investment (to counteract flooding, heatwaves, drought, cyclones, wildfires, and other extreme climate events).
SGBs would permit to bridge the SDG finance gap without raising new taxes, and hence avoid the distributional trade-offs discussed in section 2 of this paper. SGBs -a form of T-bills -are the corner stone of financial markets. SGBs would therefore represent a new category of 'risk-free assets' that could be purchased by banks and pension funds, which have strong regulatory incentives to do so (the ponderation is 0% under pillar I of Basel's solvability ratios) 11 .
One should note that sovereign currency issuers have often issued T-bill in order to tackle tough circumstances. One classical historical example is the financing of the US 'Arsenal for Democracy' during WWII. In contrast with the European belligerents' WWI finance policies -which relied mostly on taxes and foreign borrowings -US involvement in WWII was, to a very large extent, financed through close collaboration between the US Treasury and the Federal Reserve. The US Treasury issued as many T-bills as was necessary to finance the war-related purchases. In turn, the Federal Reserve supplied banks with adequate reserves to purchase these T-bills, of which it directly purchased large amounts itself. This kept the cost of deficit financing as low as possible, while pegging the interest rate on shortterm and long-term Treasury bills at 3/8 and 2.5 percent, respectively (Federal Reserve, 1942). These unusual policies enabled a rapid and automatic increase in American military expenditure, which rose from a few hundred million a year before the war to $85 billion in 1943 and $91 billion in 1944.
Beyond financial issues, SGBs would involve a strong policy stance at the European level. As discussed in Mazzucato (2015), State funding has often provided the initial push, early state, high-risk funding and institutional environment that were behind most technological revolutions (for instance in the pharmaceutical sector, communication and IT sector, or the green energy sector) (Mazzucato (2015) 12 .
In the current climate crisis, the "entrepreneurial courage of the State" (Mazzucato, 2015) may thus contribute to a structural transformation of economies towards a circular model.
One legitimate area of concern, however, is the applicability of this strategy given the prevailing institutional rules in the Eurozone. Indeed, the Stability and Growth Pact places a ceiling of 3% for budget deficit and 60% for public debt (relative to GDP). One should note however that existing treaties provide a legal basis for sustainability targeted sovereign bond issues. Indeed, article 119 of the Treatise on the Functioning of the European Union stipulates that the activities of the Member State shall include the adoption of an economic policy for the purposes set out in Article 3 of the Treaty on the European Union. This article, in turn, establishes sustainable development as a guiding principle of member countries. There is hence a legal case for Eurozone governments to subtract sovereign green bond issues from the calculation of deficit ratios (Grandjean et.al, 2018). More fundamentally, the Stability and Growth Pact can be changed by decision of the European Council. It was indeed amended in 2005 to add exemptions to the deficit rule (taking into account behavior of the cyclically adjusted budget, the level of debt, the duration of the slow growth period, and the possibility that the deficit is related to productivity-enhancing procedures).

SGB issues in the Eurozone
In the Eurozoneas opposed to other monetary systems such as the US, Sweden, the UK or Japanthe central bank refrains, except under exceptional circumstances ('outright monetary transactions'), to directly purchase Treasury securities from the government (as per article 123 of the Lisbon Treaty). In what follows, we will therefore examine the impact of the Sovereign Green Bond issue on the various parties involved, in this particular institutional setting. One should nonetheless bear in mind that governments have the power to change the rules that they write and self-impose regarding the creation of credit money.
We will use the analytical framework provided in Enhts (2017) and we will assume that the government is spending in the private sector to undertake its climate policy (note, however that the results would be unchanged were spending made in the public sector).
The SGB process begins when the government's Department of Finance and Department for Sustainability identify and budget a set of priority investment that would be required to tackle climate crisis. Ideally, external stakeholders (such as banks, investors, insurance companies) would be involved early in structuring and pricing process of the SGB through appropriate platforms, such as a Public-Private Green Bond Advisory Council. The Treasury is then in charge of issuing a fresh Green bond in order to finance this climate strategy.
In the first step, a bank borrows reserves from the Central Bank (against collateral) in order to purchase the SGB 13 . This operation results in four accounting entries: the new loan enters in the asset side of the Central Bank's balance sheet, and on the liability side of the bank's balance sheet; the new reserves enter in the asset side of the bank's balance sheet and on the liability side of the Central Bank's balance sheet. At this stage, the balance sheets of the Treasury and the private sector are unchanged. Banks then transfer the reserves to the account of the Treasury at the Central Bank in order to purchase the new SGB. In this step, the balance sheet of the Central Bank is unchanged, however the reserves now belong to the Treasury. In the Treasury's balance sheet, this new asset is matched by a new liability, which itself appears as an asset in the bank's balance sheet.

−∆ −∆
The government's Department for Sustainability now spends the proceeds of the SGB issue to implement its climate policy. New banking deposits thus appear on the asset side of the production firms', balance sheet, and on the liability side of the bank's balance sheet. A corresponding increase in reserves, drawn from the Treasury's account at the Central Bank, also enters the asset side of bank's balance sheet.

+∆ −∆
The bank is now in a position to repay its reserves loan to the Central Bank. The final positions of the balance sheets are the following:  (2) Capital (3) Current (4) Capital (5) Changes in deposits

Changes in loans Changes in high-powered money
As can be seen in table 10, this policy leads to a net increase in private sector wealth. Indeed, the private sector now holds new banking deposits, and banks hold a new interest-bearing SGB. In the process, real resources (labor, technologies) have been allocated to the pursuit of SDGs.

Greening macro-prudential policy
Endogenous money theory also provides insight into the mechanisms by which credit creation and banking behavior can be aligned with SDG priorities. Assuming that a 'sustainability taxonomy' of assets has been established 14 , table 8 now divides the production sector into the 'non SDG sector' and the 'SDG sector'. It also divides bank loansand depositsinto loans made to the non-SDG sector ( ∆ ) and to the SDG sector (∆ , ). Our zero-sum rule ensures that the flow of lending is equal to loan assets held by the banking sector (∆ + ∆ , = ∆ ). Banks now also hold both regular T-bills and SGBs in their portfolio of assets 15 . In what follows, we use this simple setting to discuss a set of prudential policies that would entice banks to issue a particular portion of their credit to specific sectors of the economy, and to rebalance their portfolio of assets towards SGBs.

−∆ +∆
In particular, table 12 shows two indicators which the European Central Banks could use as part of its prudential policy. The first indicator tracks credit to the SDG sector as a proportion of total credit.
Reaching the target value of this indicator a given year (so that, for instance, ∆ , ∆ = % ) would imply that x% of the newly created credit money (∆ ℎ ) circulating in the economy was issued in response to SDG investment demands that year. The second indicator measures the share of SDG-contributing assets in banks' balance sheets. Reaching the target value of this indicator for a given year would keep the financing costs low for SDG-related asset issuers (in particular SGBs) by maintaining adequate demand. Increases in SDG sector assets holdings will also increase the financing costs for non-SDG sector assets -and provide incentives to align production structures in the economy with sustainability objectives. 14 In the words of the HLEG (2018) this taxonomy should allow us to identify "under which conditions or criteria any given investment or financial product will contribute to the EU's sustainability objectives" (HLEG, 2018, p.15). 15 Of course, banks hold many other assets in their balance sheets but adding them into the matrix would add unnecessary complexity.
These policies entail a redefinition of the Eurosystem's mandate. This redefinition appears aligned with the Treaty on the European Union 16 . It may also contribute to the European Central Bank's core objective of price stabilization by diminishing the impact of climate change on the European economy. It is indeed recognized that physical risks will turn into financial risks and affect the prices of assets and macroeconomic dynamics (NGFS, 2019). Finally, one should also note that the Eurosystem has been able to show remarkable flexibility and resilience when faced with a crisis context. The sovereign debt crisis of 2010-2013 led to vigorous responses, such as the interdiction of naked CDS operations and strict supervision of short selling in November 2011, and the announcement of 'outright monetary transactions' in August 2012. A redefinition of the Eurosystem's mandate to achieve EDG objectives therefore appears within reach. In a letter to the European Parliament, former ECB President Mario Draghi announced that the Eurosystem "should support the sustainable development of Europe", while underlining that "it falls to the political authorities to define and decide on the appropriate measures to achieve the objectives of the Paris agreement" (Draghi, 2017).
In what follows we discuss policy tools that would upscale the banking sector's contribution to the SDG agenda. These policies include for instance the modification of the ECB's collateral framework, amendments to capital adequacy ratios, the development of a lending certificates trading platform, and the introduction of rediscounting policies for SDG loans.

Modifying the ECB collateral framework
By law, Eurozone banks need to hold reserve deposits with the European Central Bank of 1% of the value of their deposits with a maturity of up to two years 17 . The ECB, in turn, has to provide reserve on demand, at its base interest rate, up to the value of the collateral pledged by the bank requesting reserves.
To determine the bank's borrowing capacity, the European Central Bank applies a haircut to the value of the collateral. For instance, if the bank provides a collateral of a value of X and the haircut rate is 5%; the ECB will lend (X/1+5%) euros in reserves. This loan, backed up by collateral, is the Central Bank's asset, as shown in table 13. The ECB currently determines the haircut category of an asset by combining asset type and issuer group criteria (Bindseil et.al, 2017). Supervisory authorities regularly review these haircut categories from a risk perspective. Adjusting the haircut applied to different assets and issuers would be an effective way to maintain banks' appetites for SDG assets (such as SGBs).
The ECB may decide, for instance, to decrease its haircut for SDG assets, and increase it for 'brown assets'. This policy, which would amount to a modification of the asset-type classification system, would sustain the demand for SGBs in financial markets. The Central Bank could also modify its haircut rate for banks that meet or fail to mett the indicators in table 9. Banks that extend excessive loans (or hold assets) to the non-SDG sector, would receive a higher haircut. This would amount to a modification of the issuer-group classification system. > %). In order to achieve targeted lending to priority sectors, banks can sell and purchase Priority Sector Lending Certificates (PSLCs) through a trading platform managed by the Central Bank. The banks with surplus sell fulfilment of priority sector obligation and the buyer bank buy the obligation with no transfer of risk or loan assets. The fee is determined through market-based mechanism. This strategy incentivizes surplus banks to sell their excess achievement and encourages banks to lend more to the SDG sector.

Amending capital requirements
Under Basel III, the minimum capital adequacy ratio that banks must maintain (including the capital conservation buffer) is 10.5%. The capital adequacy ratio measures a bank's capital in relation to its risk-weighted assets. Changing risk-weights to entice banks to hold greener assets is an option currently championed by several banks. However, as pointed out by Finance Watch (2018), sector-specific riskweight adjustments as high as 25% have been experienced in the EU for SME lending, with little or no impact on the volume of loans (European Banking Authority, 2016). Generating sufficient volume of credit would thus necessitate large changes to risk weights, and may come at the cost of higher systemic risk (allowing banks to increase their leverage). Another option would be to raise risk weights for brown assets. This would strengthen financial stability while discouraging lending for fossil fuel activities.

Rediscounting policies
Another policy option, which the ECB could use to incentivize banks, would be to rediscount SDGector loans. Under such policies, banks directly obtain loans and advances from the Central Bank, using their loans as collaterals. The impact on the balance sheet of the bank and the ECB would be similar to the one showed in table 10 (to the difference that the amount obtained would cover the value of the loan, rather than reserve requirements). Rediscounting has fallen out of favor in most OECD countries, but was mainstream until the 1990s especially in developing countries. In a 1985 report on lending to small enterprises, the World Bank stated that "the most successful arrangement in inducing commercial banks to become intermediaries is that of having an agency to rediscount the loans made by lending institutions" (Levitsky, 1985, p.22). Rediscounting was also used successfully in France where, in the aftermath of WWII, the rediscounting of loans by the Banque de France was allowed for a maturity of up to 5 years. These operations permitted to raise 50 billion French francs in July 1947 and to finance large-scale investment programs in nationalized utilities (coal and electricity) and automobile enterprises (Feiertag, 1995). Similarly, during WWII, where the US government established a program of guaranteed loans to speed up industrial expansion in the war: loans under $100,000 (i.e., more than half of all loans) were directly handled on the spot by Reserve banks and branches (Fed, 2019). Many Central Banks in developing countries (e.g. Philippines, Nigeria, or India) currently use rediscounting policies to provide sufficient financing to priority sectors. More recently, the Green New Deal for Europe platform has championed this approach by calling for the European Investment Bank to issue long-term green bonds, which would then be repurchased by the European Central Bank as part of its asset-purchasing program (GNDE, 2019).

Conclusion
This paper has sought to inform action needed by policy makers, financial actors, and public and private investors in order to upscale SDG financial flows in response to sustainability demands. We first documented the magnitude of the SDG finance gap in the EU and discussed some of the trade-offs faced by European policy makers. We then argued that such trade-offs are largely the result of the hegemony of the 'loanable fund theory' (LFT) which has come to dominate academic and policy discourse in the past decades. Using a history of economic thought perspective, we have highlighted the logical flaws of the LFT and called for its replacement by the more realistic endogenous money theory put forth by Post-Keynesian authors. We discussed recent developments in the endogenous money theory using a consistent social accounting matrix transaction flows matrix, in the spirit of Godley and Lavoie (2012). Finally, we used it as a grid to introduce a set of simple financial mechanisms, which would allow the quick release of the amount of finance required for SDG-related transformative investments. These mechanisms -which are by no means exhaustive -include the issuing of sovereign green bonds, and the greening of money creation by banks through a modification of the European Central Bank's collateral framework, changes in capital adequacy ratios, SDG lending certificates and rediscounting policies.