This was something Keynes said in a 1942 BBC address. By 'we' he meant the government, and by 'actually do' he meant the physical resources (e.g. labour) were available. He was talking about what he hoped would happen after the war, and against the austerity policies that had seen mass unemployment in the late 1920s and 1930s. He says in the same address:
“With a big programme carried out at a regulated pace we can hope to keep employment good for many years to come. We shall, in fact, have built our New Jerusalem out of the labour which in our former vain folly we were keeping unused and unhappy in enforced idleness.”
It was a policy the UK pursued in the next few decades, after a few years producing full employment on average along with rapid growth and, initially at least, with little or no inflation. Keynes was proved correct. However in the 1960s and 1970s that was no longer the case, and we learnt through double digit inflation, and in the UK widespread disruption, that the real constraint on what you can do is inflation rather than full employment. It was always clear, however, at least to UK and other European economists, that how close you could get to full employment before inflation set in was governed by a complex and variable set of institutional factors. So Keynes’ statement became ‘governments can do anything as long as inflation remains stable’.
If inflation is the ultimate constraint on what we can actually do, why did governments start to worry about their own finances? Why did bodies like the IMF start advocating fiscal rules to limit government borrowing and later independent fiscal institutions to monitor government deficits? Some may have wanted to reduce the growth of government spending, but the official reason was real enough: deficit bias.
Deficit bias has been long forgotten as a result of the macroeconomic disaster that began in 2010, but it is crucial to understanding the origin of financial considerations influencing what we can actually do. To understand deficit bias you also need to understand another change that began in the 1970s and has now become dominant, which is how demand management is done.
In the UK in the decades after WWII, fiscal policy was used to stabilise the economy at full employment. Interest rates played a minor role, and arbitrary limits on personal borrowing were common. (This is similar to the policy proposal associated with MMT.) This made sense under the system of fixed exchange rates created by Bretton Woods. When that came to an end in the early 1970s, interest rate changes became more powerful because of their impact on exchange rates, and that was enhanced as credit controls were gradually abolished.
The world moved, often erratically, towards a system we have today: macroeconomic stabilisation using interest rates set by central banks trying to hit inflation targets. This in turn created the problem of deficit bias. In a world where fiscal policy stabilises the economy, deficit bias isn’t an issue because deficits increase in recessions but fall in booms. The economic cycle and the need for fiscal policy stabilisation means that government debt broadly looks after itself. To put it another way, labour shortages and high inflation are the cost an irresponsible government pays for spending to much or taxing too little.
In contrast, in a world where interest rates are varied to stabilise the economy, inflation is no longer a constraint on fiscal policy. In this world, Keynes’ statement of the title again becomes true, because central banks will look after inflation. The only cost of spending a lot or taxing too little would be high interest rates, but as these rates were set by someone else, the political cost to governments of running large deficits is more opaque as long as the increase in rates wasn’t too rapid.
This is perhaps the most important point of this post. Deficit targets or more sophisticated fiscal rules only make sense in a world where interest rates are able to be used to target inflation. It follows automatically that fiscal rules make no sense when rates are stuck at their lower bound. These two sentences are sufficient to show that 2010 austerity makes no sense. The reason MMTers don’t like fiscal rules follows automatically from their wish that fiscal policy rather than interest rates target inflation.
In the three decades after the fall in Bretton Woods, governments in the US and Europe (not the UK) took advantage of this new freedom monetary policy stabilisation had given them. Global debt to GDP ratios almost doubled. This is deficit bias. 
To what extent was this deficit bias a problem? In the short and medium term for the major economies (treating the Eurozone as a country) not much, but in the longer term (by which I mean centuries rather than decades) there is bound to be a limit on how large government debt could be in relation to GDP.  So for that reason alone it makes sense to try and make it hard, through rules and institutions, for governments to increase debt at that rate. It should also increase welfare if governments are discouraged from increasing debt for no good reason beyond buying elections.
I think the evidence that inflation stabilisation by independent central banks was highly successful compared to all the other stabilisation regimes that preceded it is overwhelming. So having good fiscal rules that make it difficult for governments to try and win elections by increasing debt is also worth having. This is why I support good fiscal rules.
However, I also think bad fiscal rules are far worse than no fiscal rules at all. It is vitally important that government debt should be allowed to rise in a number of situations. Fiscal rules that prevent that do far more harm than good. There are two clear situations today where debt needs to rise. The first is in severe recessions, where interest rates get stuck at their lower bound . The second is where large investments are required to produce a large future benefit. The most obvious example of the latter is climate change.
Why are there so many bad fiscal rules around today? I think one important reason is that the political right has seen them as a way of reducing the size of the state. The media have not helped, by grossly exaggerating the cost of higher debt and using any departure from a (bad) fiscal rule as a sign of government irresponsibility.
Which brings me to what inspired this post, Adam Tooze’s Chartbook No.34. This takes an MMT perspective, and ignores all the points I make about interest rate stabilisation and inflation. Which is understandable, when interest rates have been at or close to their lower bound for over a decade and inflation has not been an issue. The puzzle he addresses is how can we describe a government project (project X) as crowding out something else, when the amount the government could borrow before hitting an inflation constraint far exceeds what it actually borrows. You cannot say doing X stopped you doing Y and Z, because you could have done X, Y and Z.
I suspect this is usually an artificial question, because for me at least doing X, Y and Z would bring the economy to the point where it either hits the inflation constraint (in an MMT world) or where interest rates start rising (in today’s world). This would be a certainty if Y was greening the economy and Z were the fiscal transfers to make a carbon tax (or equivalent) politically acceptable. In both cases there is a clear opportunity cost of doing X.
An example of X where this arose recently was Trump’s tax cuts for the rich. I am happy to say the main problem with this is it was a transfer of income from the many to the few. The rich were better off, and that money either comes from the rest of the population now or in the future. Most people don’t see that because Trump paid for it by borrowing. For me , that means it’s also an example of deficit bias: raising the deficit just to pay off those who gave you money to win the election. 
I agree with Adam Tooze when he says there is no fixed pot of money. The government is the owner of the magic money tree, and in the US, UK and collective Eurozone today can borrow freely. I disagree on three points. First, I think the constraint in MMT type economies (where only fiscal policy does macro stabilisation) is normally inflation and not available resources, and second I think we can talk about the opportunity cost of bad policies even when the inflation constraint does not bite for the reason I gave above. Third, in today’s economies, it does make sense to ask whether deficits are justified or not, just as it did over the period of deficit bias and subsequently.
 The term deficit bias is bound to involve value judgments. No one should describe the rise in government debt after the GFC and the pandemic as deficit bias, because debt rose for very good reasons. Deficit bias is a rising government debt to GDP ratio over decades for no obviously good reason, and what a good reason is has to be a value judgement.
 Debt around 100% of GDP is fine. But if it doubles every 30 years for no good reason, you are over 1000% a century later and over 10,000% after that.
 This is why the fiscal rule that Jonathan Portes and I proposed here contained a knock-out where the rule no longer applies when interest rates hit the lower bound. Any fiscal rule today that fails to have something similar is a bad fiscal rule. It also focused on the current balance, putting no constraint on investment to green the economy, for example.
 I admit that a Republican who thought the economy was better off if taxes on the rich were reduced would not call this deficit bias. In that case all you can do is point out that there is little sign his belief is correct (e.g.)
 Does the relationship between r (the interest rate) and g (the nominal growth rate) matter here? I don’t think so, because we are talking about a permanent flow (tax cuts) rather than a one-off project.