“There is nothing scarier than ignorance put into action” — Goethe

A common mistake in economic analysis is to consider that any type of “desequilibrium” in an economy can be solved through a simple government intervention. However, this reasoning ignores that government actions can also be flawed. Government failures generally arise as a result of epistemic or organizational limitations within the State.

With regard to monetary policy, one justification that can be given for its implementation is that the monetary authority could conduct macroeconomic policies to stabilize fluctuations in economic indicators; such as GDP, inflation and exchange rate. However, the authority faces a constraint on the ability to conduct this policy optimally. To understand this restriction, it is necessary to understand the relationship between economic policy and output (GDP). It is possible to say that the effects of an economic policy on the output of an economy can be expressed as:

Y_t+1=Y_t+β

Where Yt+1 is the product in the presence of a policy, Yt is the product of the period prior to the implementation of that policy and β is the measure of the effects of the economic policy on the combined product at time t . Since we seek to know whether or not a given policy can stabilize an economy, the most appropriate measure will be the standard deviation of products around their means (Friedman 1953). The greater the deviation for a given term, the greater its instability. Since the monetary authority seeks a policy whose effects stabilize the income of an economy at an optimal level, then it is acceptable to say that it seeks a scenario where the deviation of Yt+1 is smaller than the variation of Yt given the measure of the β policy effects. Formalizing it is found that:

σY²_t+1 = σY²_t + σβ² + (2ωYβ) σY_t σβ

Where ω will indicate the correlation coefficient between Yt and β , so that it will determine if the policies taken in the order of magnitude σβ will generate a movement in the same direction or not in the variance of Yt. Considering that such a correlation occurs, it is possible to say that ωYβ will vary from +1 to -1. If it is -1, we have the ideal scenario where β is negatively correlated with the variance of Yt. As for the case of a perfect +1 correlation, we have the scenario of extreme destabilization, where β is highly correlated with Yt. In the case where ωYβ = 0, we have a random scenario without precise correlation and where the variables may or may not converge. Since only negative correlations will generate stabilization, policies are more likely to generate economic destabilization than stabilization. For this reason there is a natural restriction on the type of policies that can be adopted by a monetary authority.

This criticism, however, only gives a mechanical relationship between policy and product. It does not talk about the possibilities of economic policy failure that may arise due to variations in β. The previous model assumes that the effects of economic policies are given, but they may vary due to economic agents responding differently to the incentives created by these same policies.

Kydland and Prescott (1977) note that one of the mistakes of macroeconomic policies is to assume that agents form their expectations only in an ad hoc way based on past prices. In this scenario, a mechanical control of the macroeconomic aggregates is relatively easy, as the authority only needs to estimate the current variables and carry out a valuation of the effects of a given policy on their future values. However, if agents are able, with the information they have, to anticipate the objectives of economic policies, then any attempt to establish an optimal economic control will tend to the absurdity of fighting against the very variables that it seeks to control ^[1].

This restriction manifests itself above all in the way in which the monetary authority deals with the trade-off between unemployment and inflation. In a simple sticky price model, where firms set their prices based on the price of their competitors, the inflation rate will be determined by the percentage change in price adjustment of the firm relative to its competitors based on past demand. However, this past demand may or may not be high or low relative to the economy's potential output under full employment conditions. Thus, we can assume that firms do not price this information. Therefore, the inflation rate will behave according to the Phillips Curve, where:

π = f(Y_t-1 - Y_p / Y_p)

Where π is the inflation rate, Yt-1 is the effective aggregate output of the previous period and Yp is the potential output. However, in such a model agents price only based on changes in current prices, ignoring the fact that they can form their prices based on the rational expectation of a future price increase given certain present information. The concept of rational expectation employed here does not imply that they will make perfect predictions about the future, but rather that they, given some knowledge of their own, will form expectations about how the economy will behave given current information (Sargent and Wallace 1973). In this scenario, the inflation rate would no longer be determined by the Phillips Curve, but by the relationship of agents' expectations where:

π = π _e + f(Y_t-1 - Y_p / Y_p)

Where π e is the inflation rate expected by agents in period t+1. Thus, the inflation rate and, consequently, the trade-off between inflation and unemployment will not only depend on controllable variables, but also on the agents' expectations about the course of economic policy. Kydland and Prescott note that in this scenario, if the expected inflation rate equals the rate estimated by the monetary authority, then individuals will rationalize their assessments of the macroeconomic environment based on the portfolio of indexes used by the monetary authority in its estimates. This creates a delicate scenario for conducting monetary policy, as a deviation in the inflation rate can be interpreted as a future reduction in the purchasing power of money (real value).

Thus, a monetary authority would be limited in its possibilities of action by the expectations of economic agents, so that it would have to act in accordance with them. It could be argued that a policy could come closer to its ideal if the monetary authority used indicators of market expectations to guide the formulation of stabilizing policies. However, this argument ignores the flaws that can arise due to epistemic problems. Central banks are different organizations from those of a private monetary system (money markets) essentially because of two differences: (I) central banks are, by definition, a form of monopoly firm in the money supply and (II) they do not operate according to criteria of profit and loss, but according to discretionary political impulses or monetary policy rules.

The consequence of these two characteristics of central banks is that they operate in an environment exogenous to the market, in the sense that it is not involved in the same profit and loss process as other agents (Cachanosky and Salter 2020). Thus, unlike market agents who capture information from the market through the process of acting within it, central banks, in order to make their policies effective, have to learn about the market by gathering information about it and using this information as proxies for the type of knowledge that market agents already have. This knowledge constraint poses two problems for central banks: how to properly define what would be the appropriate monetary policy for a given scenario and what information to use to substitute market price signals to design such a policy?

This restriction remains even in the case of a central bank that tries to strictly follow market expectations expressed in its indicators or tries to transmit monetary policy via the banking market. Even in these cases, it can only do so by exogenously imposing its interest rate; given that for him the natural rate of interest is not tacitly known as it is for market agents. That is, the central bank never takes the money market equilibrium, it rationalizes such equilibrium by acting as the system's main agent.

This epistemic limitation becomes even more critical when considering a scenario where complex dynamics exist. Orphanides and Williams (2006) demonstrate that even a monetary policy based on rules and market information has limitations and possibilities of failure due to knowledge problems, especially when estimating interest and unemployment levels that should be considered natural. To carry out such policies, central bankers need to estimate real-time natural rates to calibrate the optimal β for a given policy. However, if such a process has a certain degree of uncertainty, the estimator will tend to be defective and the policy will be inadequate for stabilization purposes. This dynamic is even more complicated if one takes into account a scenario where economic agents form expectations through learning through finite data and where they take the time series data with certain subjective weights. Learning in a rational expectations model is associated with greater volatility of indicators and persistence of monetary policy errors.

Despite these epistemic limitations being considerable, it is interesting to note that even in the unlikely event that the monetary authority manages to overcome them, there would still be an important organizational limitation on its actions.

The first aspect of this organizational limitation concerns the size of the bank. Central banks are hierarchical organizations of large scale and scope that outnumber all agents within a money market. However, they cannot have a very large size due to coordination problems that would be generated in this case. Klein (1996) points out that one of the upper limits for the size of a firm is given by the transfer pricing problem. An organization that is structured as an integrated corporation composed of semi-autonomous administrative cores faces the problem of how to allocate resources from one unit to another; as in the case of a central bank allocating resources to its units spread across different regions or markets of a country. This problem can be expressed by a series of questions: how to value the resources that are being allocated? What is your relative price? In what proportion should it be allocated?

To resolve these issues, the organization's central administration can hardly rely on its internal prices, even if these are generated by bargaining between units, as these will imperfectly express the opportunity costs of alternative social uses of such resources compared to market prices. For this reason such an organization will need an external market price that it can use as a benchmark for its internal transfer price. This generates the restriction that no organization can be too big to the point of internalizing all markets. In the case of central banks, this translates into the fact that they can never adequately replace financial and banking markets and will always have to turn to them for information on money and credit allocations across markets.

Another organizational problem that monetary authorities may face is a special situation called the multiple principals problem (Oritani 2010). Unlike private banks, which essentially have to answer to two principals (depositors and shareholders), central banks have to answer to several principals: the public, the legislature, the executive, the financial market and other central banks ^[2]

In theory this would be good as it would prevent the central bank from being captured by the interests of a single principal to the detriment of others, but in reality such a problem means that the central bank will have to coordinate multiple conflicting interests and weigh which is the most appropriate while the private bank only needs to worry about maximizing its results. Such a question of weighing and influencing multiple interests in its decision-making can lead to monetary policy becoming inconsistent over time.

Finally, organizational problems, especially in the case of public governance, end up generating problems of a political nature. In the case of monetary authorities, the main political obstacle lies in the uses of the micro and macroeconomic functions of central banks for fiscal purposes. Due to the fact that there is a difficulty for the public to identify whether the causes of an inflationary process are real or monetary, a government that wishes to expand its discretionary consumption power via seigniorage may find a favorable scenario even in the case of the existence of a conservative central bank. Buchanan and Wagner (2000) note that members of the political body can act as free riders of price stability and incur inflationary deficits, as it will be the responsibility of the central bank and not the Treasury to maintain price stability. This raises the question that there may be incentives not to promote or to promote less price stability than would be appropriate given that stabilizing central bank actions will be offset by fiscal policy actions.

In addition, the government can use the regulatory powers of the monetary authorities as a source of tax revenue. The main form of this problem is when the government uses the banking regulations of the monetary authority as a way to leverage government bonds through the institution of mandatory bank reserves. Reinhart and Rogoff (2009) note that this form of financial repression is a common form of indirect taxation in historical evidence. This measure means that the citizens of a country are obliged to deposit their resources in a certain number of banks and causes the banks, in turn, to be forced by law to have minimum reserves in public debt bonds. This scheme allows the government to borrow funds from depositors at extremely low rates at the cost of linking the stability of the banking sector to the solvency of public accounts. These political failures end up justifying the existence of a certain independence of the monetary authorities in relation to government policy. However, the extent to which this separation is effected by efficient institutions is an open question.

Given all these questions raised so far, a prudent and rational citizen should at least be a little more skeptical about the easy speeches of populists that monetary policy is something simple to solve or even to understand.

^{^}
This is one of the reasons why Friedman advocated monetary policy being driven by a form of AI that follows automatic monetary policy rules.
^{^}
The other central banks are principals to the main national central banks because many of them offer custody services to each other, have obligations with each other and transact in exchange rate operations.