“Introduction to Post-Keynesian Economics”, part 2: Demand, Employment, and Boxes

Today, we’ll cover chapters 3 and 4 of Marc Lavoie’s “Introduction to Post-Keynesian Economics”. You can find the previous two chapters covered here.

Chapter 3: A Macroeconomic Monetary Circuit

“The principle of effective demand … — that is, the causality that runs from investment to saving — is best understood within the context of a macroeconomic explanation of the monetary circuit.” Post-Keynesians reject monetarism and the quantity theory of money, instead following endogenous money theory. They believe that “the banking system fixes a rate (or a set of rates) for the money market and then lends however much borrowers ask for, provided that they can offer satisfactory collaterals”. In other words, banks will make a loan as long as the person taking on the loan is a credible borrower.

Contrast this to the doctrine of loanable funds, still accepted by most new Keynesians. This theory says that the role of the central bank is to ensure the real interest rate equals a natural interest rate. When the former is less than the latter, inflation inevitably occurs. Post-Keynesians reject that such a rate exist.

3.1 Main characteristics of post-Keynesian monetary analysis

A key feature of this approach is that loans create deposits, which reverses the causation of mainstream economics. How does this work, exactly? Say the bank finds a credible borrower. To make the loan, they simply create and deposit money in the borrower’s account. Put another way: bank reserves are not required to make loans. They are only kept to follow rules set by the central bank, or to allow folks to withdraw cash if they want.

Additionally, “inflation is not caused by an excessive rate of growth in the money supply. If anything, the causality is also reversed. The growth rate of prices and output instead cause the stock of money to increase [through higher demand for loans]. The inflation rate is explained through other causes.”

We should take care now to separate out two types of banking systems. Asset-based systems are the norm in Anglo-Saxon countries, while Europe and Asia have overdraft economies. As we’ll explore, it’s less obvious that money is endogenous in the former systems, but post-Keynesians believe money is endogenous regardless of institutional setup.

Financial markets have many assets, each with different rates. At least one of these rates is set exogenously by the central bank of a country. We’ll call this the benchmark rate. Also known as the key or official rate, it serves as a reference point for all other rates. In most countries today, the benchmark rate is the overnight or interbank rate: “the rate at which commercial banks lend and borrow funds to and from one another in the overnight market, also called the federal funds rate in the USA.”

3.2 The relationship between commercial banks and the central bank

High-powered money (HPM) consists of currency supplied by the central bank, as well as bank deposits (reserves) held by commercial banks at the central bank. Reserves are a liability to the central bank, and an asset to private banks, as private banks can withdraw from them at any time. Similarly, your deposits at at bank are a liability to the bank, but an asset for you. And just as loans by banks create deposits, the central bank can create reserves by buying up assets.

The balance sheet above is also helpful in differentiating overdraft and asset-based economies. In the former, banks hold few government securities (such as T-bills). So to obtain reserves, they must borrow them from other banks, or from the central bank as a lender of last resort. In the latter, the central banks hold “open-market operations” that buy and sell T-bills to influence bank reserves.

One thing we haven’t mentioned on the balance sheet yet are foreign reserves. In the mainstream Mundell-Fleming model with fixed currency exchange rates, a balance of payments surplus (meaning more money is flowing into a country than out of it) leads to an increase in bank reserves, which then increases lending and the money supply. But post-Keynesians instead believe in the compensation principle: that all surplus is absorbed by the central bank through selling T-bills or decreasing debt owed to the central bank by private banks. Either way, the overall money supply is not changed.

We can summarize by saying that the central bank does not directly control the supply of money in general, nor HPM specifically. Instead, they supply whatever the banking system demands at a certain price. What they do control is this price, the benchmark rate. It’s an administered price: just like the administered prices of last chapter, it’s re-evaluated and changed periodically. While the central bank can technically set the benchmark rate to be whatever it pleases, in practice it acts in response to economic conditions. This response is described by the central bank reaction function. It generally increases the rate when capacity utilization is high, unemployment is low, or inflation is rising, and decreases it under the opposite conditions.

3.3 The relationship between banks and firms

Lines of credit are a contract between a bank and a borrower which specify the maximum amount that can be borrowed, the rate at which it is borrowed, and any conditions on the loan. When a firm has higher debt, the risk to the bank of default is higher, so they charge higher rates to compensate for this. This is illustrated by Kalecki’s principle of increasing risk, as mentioned in the last article and shown below.

As mentioned earlier, although money is endogenous, some people will still be left out because banks only lend to creditworthy borrowers. We can therefore separate out the notional demand for credit (which is the demand of everyone) from the effective demand for credit (which only counts creditworthy borrowers). As interest rates increase, notional demand for credit goes down. But effective demand goes down farther, since less borrowers will be deemed creditworthy. So, the gap between the two, and therefore the amount of credit rationing, increases with interest rates. Note that there can’t be an “oversupply” of money: banks will supply whatever creditworthy borrowers demand.

But what can cause an increase in interest rate? Well, the interest rate can be written like this:

Where i_b is the benchmark rate, and σ is the risk premium. So, either the central bank can increase the benchmark rate, or banks can increase the risk premium if they think loaning to a client has gotten riskier.

This relates to a key concept for agents in a monetary economy (first developed in Keynes’ “General Theory”): liquidity preference, or how badly savers want to hold their savings in a form that can be immediately spent. Of course, with more uncertainty, agents’ liquidity preference will go up. For banks specifically, higher uncertainty means an increase in the risk premium, so lending rates go up. At higher rates, fewer potential borrowers are found to be creditworthy, so lending goes down overall. But with lower uncertainty, the opposite occurs.

This leads us to Minsky’s financial instability hypothesis, or FIH for short. Minsky argued, by the logic above, that economic booms lead banks to make riskier loans. After all, the economy is in a good state, so they will mostly be repaid. But an economy full of riskier loans is more unstable, with more speculation and higher debt burdens. The central bank will likely raise the benchmark rate in response to the boom, thus increasing interest rates, making it harder to maintain those high debt burdens. This all adds up to a riskier environment: banks will now restrict lending more, perhaps leading to an economic crash if governments do not boost aggregate demand. The key insight of this process is the paradox of tranquility: economic stability leads to instability eventually.

3.4 A systematic view of the monetary economy

Here, we’ll systematize a mesoeconomic approach, which lies between the microeconomic analysis of individual agents, and the macroeconomic analysis of effective demand. This approach focuses on sectoral balance sheets and financial flows, displayed with matrices. It’s important to start with a model that’s stock-flow consistent, where every movement of money has a source and destination, and overall balance sheets always sum to zero. Such a model should include both real assets (like inventory and real estate) and financial assets (like loans).

Below is a highly simplified version of such a model in a closed economy, without a central bank or government (which would collect taxes, give out welfare, and employ people), without households borrowing from banks, and without intermediate goods. [You can see an example of a more complete model here.] Note that this is representing what households, firms, and banks do in sum across the WHOLE economy, with banks and firms divided into their current and capital accounts.

Note that each row sums to zero: there is a balance between inflows and outflows. Flows here move from boxes with a negative sign to those with a positive sign on the same row. Each column sums to zero as well: all money a sector gets ends up somewhere, whether it’s spent on consumption or put into banks as deposits. Current accounts show the actual transactions of banks and firms, while the capital accounts show assets and liabilities. The subscript of -1 denotes the last period: for example, at the beginning of every period, households receive interest on deposits, equal to the interest rate on deposits times the total deposits last period.

The shaded boxes are all involved in initial finance. First, banks make loans to firms. Then, firms use loans to pay wages, and/or bring in inventory to sell. Before it can be spent, wages are deposited at banks, closing out the initial finance stage.

Chapter 4: The Short Period: Effective Demand and the Labor Market

A core idea in post-Keynesianism is that economies are demand-led. In other words, increases in demand lead to increases in supply. One result of this, contrary to neoclassical economics, is that a decrease in real wages does not increase the demand for labor. In face, it’s the opposite: lower real wages mean lower consumption, so firms will hire less labor. This chapter expands on this idea by focusing on the short period, defined here as ignoring the effect of investment on the capital stock, and assuming the goods market is in equilibrium.

4.1 Effective demand and its components

In neoclassical models, aggregate demand is a basically a function of fiscal policy and the money supply. However, money supply is endogenous to post-Keynesians. Simplifying by assuming a closed economy with no government, what then determines aggregate demand?

Aggregate demand, and therefore income, is:

With these simple equations, we can now solve for profits, which are of course a core part of a capitalist economy:

Note that while individual capitalists can choose how much to consume and invest, they cannot choose to make an arbitrary amount of profits. So, the causality of this equation runs from right to left. In the words of Kaldor: “Capitalists earn what they spend, workers spend what they earn.”

We can contrast these ideas with the neoclassical approach when it comes to the role of deficits. To finance deficits, the government basically takes loans from the private sector by selling Treasury bills. Mainstream theory says that more deficit spending drives up the interest rate of these T-bills, and therefore the interest rate across the economy. This, in turn, “crowds out” private investment, as less investment is made at this higher interest rate.

But look back to our interest rate equation: the interest rate a bank charges is based on the benchmark rate and the risk premium. The interest on T-bills is not the benchmark rate, and more deficit spending could even decrease risk. Plus, if deficits that end up in the pocket of workers are spent like wages are, deficits actually end up INCREASING profits! So if anything, there is actually a crowding-in effect in post-Keynesianism.

4.2 The Kaleckian Model

Let’s rewrite that profit equation a bit. Let P = pa, where P are profits, p is price, and a is real autonomous expenditures (real consumption out of profits plus real investment). Furthermore, let wages equal the average annual wage rate w times the number of workers, N. In this simplified model without government, income is again just wages plus profits, which is equal to aggregate demand.

Using our knowledge from chapter 2 about firms, we’ll add a utilization function with constant returns when utilization is below 100%. In other words, production with a constant capital stock can be increased by hiring more labor that utilize machines more. “At the macroeconomic level, therefore, modern firms face only one constraint: effective demand”. However, individual firms are constrained by their market share.

The total quantity of goods produced, then, is the yearly output per worker (T) times the number of workers. In practice, firms can and will guess wrong about the level of demand. But here, we’ll assume they’re in a stable equilibrium overall. Since supply is determined by demand, the quantity of goods q = T*N = RAD. Rearranging this equation gives us one of two equations measuring the effective demand curve:

This is illustrated below with a graph of the effective demand curve. Employment is on the x-axis and real wages are on the y-axis.

A few things to note. First, there is a thick line drawn at T because it’s the maximum for real wages. For firms to make a profit, T must be greater than real wages. Second, (w/p)_fe and N_fe respectively denote real wages and total employment at full employment. The labels a_1/T and a_2/T denote the effective demand curve at two different levels of autonomous expenditures, as we’ll discuss shortly.

For an individual firm, cutting wages while keeping prices the same (in other words, lowering their workers’ real wages) would increase their markup, and therefore their profits. But as the graph shows, a decrease in real wages means a decrease in employment too, as it decreases demand. This is in stark contrast to neoclassical economics, where the relationship is switched.

Of course, an increase in autonomous expenditures can increase demand instead, in which case the effective demand curve will shift right. For example, at a constant real wage (w/p)_1, an increase in autonomous expenditures of a_1 to a_2 would bring the level of employment from N_1 to N_fe. But why would such an increase happen? As Keynes noted, it could be induced by lowering the benchmark rate: but this might not be enough, especially in a recession. That’s why he favored public employment programs, so governments could simply increase investment directly.

Let’s now add the savings rate out of profits (s_c) to our equations. Note that a_cc is real capitalist consumption, and a_i is real investment, so p*a_cc + p*a_i = pa = P.

This is known as the Cambridge short-period profit equation. Here, we can see that total profits don’t directly depend on real wages (at least in the short term). So if capitalists as a whole decrease real wages and keep prices constant, thus increasing their profit rate, total profits will NOT change. Instead, the demand decrease exactly offsets the profit rate increase, though employment falls. This is known as the paradox of costs, first noted by Kalecki.

Using what we have learned so far, we can derive equations for the employment level and total output in terms of productivity, real investment, real wages, and the savings rate out of profits.

From the employment equation, we see that if real wages decrease while the savings rate and real investment are constant, employment decreases. From the final equation for output, we see that if the savings rate increases, output decreases: this is the paradox of thrift, first noted by Keynes.

4.3 Further Developments of the Kaleckian model

Many labor economists think that the labor supply curve is “backward-bending”. At first, higher wages induce more people to join the workforce. But as wages get high, workers will be richer and decide to take more leisure time. Below, we can see this implies two equilibrium employment positions, denoted by H and L.

Which equilibrium will the economy tend towards? Let’s take an arbitrary point (denoted 0) between positions L and H to see. At this and all points, savings equal investment and the goods market is in equilibrium. But the labor market is not in equilibrium: there is an excess of labor supply. Absent any market rules or norms, the tendency would be for prices to stay the same as nominal wages decrease, until L is reached.

Put differently, market forces would push the economy into a worse equilibrium. Price flexibility here has clear negative effects. But luckily, we have ways to keep real wages from falling: unions, minimum wage laws, and public employment can all push wages up, acting against the market to get us to the high equilibrium.

We now turn to a classic question in economics: how does technical progress affect the level of employment? “Neoclassical authors are unanimous in arguing that technical progress can only have positive effects on employment, or at least that negative effects are only sector-specific.” Given how many normal folks are concerned about this issue, and how many economists have strong opinions on it, let’s examine it using the Kaleckian model.

Let’s start with an economy at full employment, with T = T_1. Next, we’ll move to T = T_2, with T_2 > T_1, and assume capitalist savings and real investment remain constant.

As we can see, real wages must be increased in order to maintain full employment. Otherwise, employment will fall to N_2. In fact, real wages must increase more than proportionally to maintain full employment. To see why, we return to this equation for employment:

If real investment and capitalist saving are constant, it’s the DIFFERENCE between T and real wages that must remain constant for employment to remain constant, not the ratio between them. In practice, it’s unlikely that wages will increase faster than T, so increased autonomous expenditures are needed to prop up aggregate output.

Crucially, output decreases too. To see this, start with our earlier equation for total output.

Given a change in T as before, when will q decrease? In other words, when will output at a productivity of T_2 be less than at T_1?

In other words: in this scenario, output at a productivity of T_2 is less than at T_1, if T_2 > T_1.

Assumptions aside, how technological progress plays out in real life depends on the business cycle. In an economic boom, unemployment is low, so workers will be able to bargain better for higher wages. Plus, the good times will likely make capitalists invest more (further increasing productivity), and may make households more comfortable spending more. But in a recession, cost-cutting measures are pursued to increase profit margins without lowering prices. Workers aren’t able to increase their real wages, so this higher profit lowers effective demand.

Let’s turn to a prescient example where this model is useful: the policy of work sharing, “whereby workers reduce their hours of work with the objective of increasing overall employment”. The idea is that if firms require a certain number of labor hours to meet their goals, and current workers reduce hours worked, they will have to hire more workers. But this ignores macroeconomic effects. Let’s reformulate some earlier equations in terms of hours to see why.

As is clear from the last equation, if a reduction in hours is accompanied by no change in productivity or the hourly wage, then employment will increase as intended. Suppose instead that a decrease in hours is exactly offset by an increase in productivity, such that T stays constant. If hourly wages remain the same, then annual real wages and therefore employment decrease. But if hourly wages were increased with productivity such that annual wages and productivity both stayed the same, employment would stay the same, yet workers would be paid a higher hourly wage. Thus, “Post-Keynesians only endorse work-sharing programmes and their reduced working week when they are accompanied by an increase in the hourly real wage”.

In the next article, we’ll finish off with chapter 5 on growth, and the conclusion of chapter 6.