Demand-side policies — fiscal and monetary — are powerful, but they work on a single variable, and that is the source of both their strength and their limits.
What they can do is considerable. By raising or lowering total demand a government can move output, employment and the price level together. In a recession, expansionary fiscal or monetary policy raises spending, firms increase production and unemployment falls; when demand is growing faster than the economy can produce, contractionary policy restrains it and demand-pull inflation eases. The instruments are also flexible: interest rates can be changed at any time, and tax rates reach every household at once. For an economy whose problem is that total demand is at the wrong level, demand-side policy is precisely the right tool.
The first limit is that the aims conflict when approached from the demand side. Raising demand to cut unemployment also raises the price level, so the government trades inflation for jobs; restraining demand to control inflation costs output and employment. It also worsens the current account, because a large part of the extra spending goes on imports. A government using demand alone is therefore choosing which aim to sacrifice, not achieving all of them.
The second limit is more fundamental: some problems are not demand problems at all. Structural unemployment exists because workers have skills an economy no longer needs and live where the new jobs are not. Raising total demand does not give a redundant miner the skills of a software engineer — it simply bids up wages and prices in the industries that are already busy, producing inflation alongside the unemployment rather than instead of it. Cost-push inflation, caused by rising import prices or wage costs, is the same case in reverse: contractionary demand policy can suppress it only by crushing output, because the cause was never excess demand. Low productivity, poor infrastructure and inadequate training limit how much the economy can produce, and no amount of demand can buy output that the economy is incapable of producing.
My judgement is that demand-side policy alone cannot achieve all the aims, and the reason is a matter of category rather than of degree. Demand-side policy determines how fully an economy's existing capacity is used; it does nothing to change what that capacity is. So it can close the gap between actual and potential output, and it can do that well — but once the gap is closed, every further attempt to raise demand produces inflation and nothing else.
That gives a clear division of labour rather than a ranking. Demand-side policy is the right instrument for a cyclical problem, and supply-side policy — education, training, infrastructure, incentives to invest — is the only instrument for a structural one, because only it raises the capacity that demand-side policy can then bring into use. The two are complements, not rivals: supply-side measures without sufficient demand leave new capacity idle, and demand without capacity produces only rising prices.
The practical conclusion is that a government facing high unemployment must first diagnose which kind it is, because the two require different medicine and the wrong one makes the situation worse.