I'd like to posit a few things:
1) It's impossible to understand fiscal policy without understanding finance - not banking finance, but capital markets finance.
2) It's very difficult to understand fiscal policy without understanding industrial organization
3) It's impossible to understand fiscal policy without understanding monetary policy
4) It's impossible to understand monetary policy without understanding trade
5) It's impossible to understand trade without understanding industrial organization and finance
Walking through each of these will hopefully be revealing.
1) The link between fiscal policy and finance:
I've written pieces here: http://tfideas.blogspot.com/2010/07/hypersimplistic-way-of-thinking-about.html
Summarized and generalized: A government should spend on something or regulate something if the economic impact (call them the economic returns) of that spending or regulation will eventually exceed the cost borne by the society. This cost is a blend of interest rates (government rates in the case of spending, private rates in the case of regulation) at infinite term, economic activity displaced by spending (crowding out), economic activity displaced by current and future taxation and regulation to pay off cost of the initial government spending and regulation.
To think about sensible stimulus, then, you need to think about what the stimulus means.
Interest rates on the short and intermediate end are low, but these rates are short-term rates. They'll need to be rolled over, and given how "sticky" tax rates are, you need to consider short rates in the distant future. Thus, while low interest rates mean the rate of spending can go up, it shouldn't be by very much. Having an understanding of applied finance and having spent time with discount rates on different securities is really important - it's easy for an academic economist to select an interest rate empirically, but until you've worked with discount rates and understand them intuitively as well as academically, you're not going to make a good choice.
I use a projected interest rate of about 7% - 8% moving forward. This was 10% when rates were higher. I do work in finance, but I'm young. I don't claim to know the answer - just that a 4% discount rate (as I see often) is ludicrously low, given that stimulus projects are higher risk projects to not "pay off" than most corporate bonds, given poor quality of project assessment in government.
This means that EVEN IN RECESSION, ignoring crowding out, your spending needs to "return" 7% - 8% to GDP without even looking at crowding out and tax impacts, just based on interest rates.
As a risk control metric, you also need to understand what happens when people lose faith in your ability to repay. As investors in sovereign debt in the PIIGS are finding out, this can make interest rates go up very, very rapidly, and very uncontrollably. So from a risk control perspective, you probably should incorporate that into your assessment, as well.
Of course, at some point, you can ignore interest rates... but that's cuz you've paid off the debt, and you've done that with taxes. So if you ever have any intention of paying down your debt, you need to look at deadweight loss of taxes, as well.
2) Note that I've ignored crowding out - this is because it falls into the "fiscal policy - industrial organization" category.
Many people argue that crowding out isn't of consequence in recessions, because there's so much extra capacity.
The issue with this blanket statement is that there are plenty of frictions based on training. This is widely recognized when talking about unemployment (from people in favor of job retraining programs, etc) but a lot of people seem to forget it when it comes time to talk stimulus. This is a danger of fiscal policy - Spending lots of money on electronic health records can cause a great deal of crowding out, because one job where we really are very capacity-constrained is computer programmers. I've heard a number of issues with a lot of the ARRA infrastructure projects, because laid-off residential construction workers can't do a lot of the jobs required by infrastructure building, so the generally low-quality ARRA projects ("shovel ready" tends to mean "considered and previously rejected as not being worth it") end up taking qualified people off of better projects. Crowding out still happens, so when you're designing fiscal policy, you need to look at the very specific places you have excess labor supply, which requires understanding competitive substitutability, market power, and a number of other factors.
3) As Scott Sumner likes to point out, if a central bank has an inflation target (or, as he'd point out, even better an NGDP target), then fiscal policy is endogenous anyway - as you do more fiscal policy, you do less monetary policy, so all fiscal policy ends up doing is redistributing from the most efficient potential projects on the margin (who benefit most from a small drop in interest rates and increase in inflation expectations, because they can get started) towards projects linked to congressional campaign donors. Even if a central bank isn't perfect at targeting, they're still moving directionally - hence, as Sumner has pointed out, the fiscal multiplier has been like 0.3 or 0.4 since the late 70s, when the central bank started targeting. Even at zero interest rates, there are plenty of monetary options remaining - QE, IOR reductions, NGDP commitments, level targeting - there are plenty of options, and while some of them would need to get through Congress, none of them would be hugely controversial as long as the dual mandate stays in place. Understanding the mechanics and empirics of liquidity traps is also essential to understanding the limits of monetary stimulus and when fiscal becomes more important (and no, I don't believe we're at that point, or have been yet during this crisis). Without understanding how monetary policy works, you'll have a significantly inflated view of the power of fiscal stimulus.
4) Of course, if you don't understand currency and international interest rates, you'll have an inflated view of monetary stimulus. Many of our trading partners - China, Japan, even Switzerland now - peg their exchange rates, which means that monetary policy gets partially sterilized. The US imported about 10 trillion dollars in goods and services in 2008, and we're close to that as an annual run-rate now. Exports run about 8 trillion dollars, off the top of my head, I believe. Monetary policy gets sterilized proportional to some combination of the trade deficit and the overall import number (for a fixed trade deficit it would always be proportional to the trade deficit, but price sensitivity and price stickiness should mean that it's fuzzier than that). That means that you're taking at least a 15-20% haircut on the effectiveness of monetary stimulus, bringing down THAT multiplier.
5) Of course, you could fix the monetary policy sterilization issue of number 4 with tariffs, but those create huge deadweight losses. An import certificates strategy would create smaller deadweight losses (you'd only get the deadweight losses on the products on which we were creating the least producer surplus to begin with, so it'd be way smaller for an equivalent currency move)... but you'd also throw China into a colossal recession because their investment-intensive model, driven by exceptionally cheap capital (driven down by government mandate) would explode (in this opinion, it's going to explode anyway, it's just a matter of when and how badly - how badly being a product of how long they wait before readjusting and how decisively they can move to improve consumption... but anyway), and that would cause problems here in the short run, as well. So moving slowly on trade has benefits, which means you need to size and time your monetary policy accordingly. Additionally, while a readjustment would slow down the rate of job transfer from the US to China, China has a number of structural advantages - their uneducated labor is much lower cost, their land is largely cheaper and their regulatory structure, while still vast, is smaller and less restrictive than the US, which means that to really normalize trade, you have to fix a number of regulatory issues in the US and watch the Chinese banking system to avoid the overly rapid collapse of their investment/property bubble (yes, it's a bubble). You also need to look at all of China's trading partners, many of whom rely on natural resources and will be thrown into major recessions when China pops, because that would offset a lot of the US currency benefits you'd get from a Yuan revaluation.
Thus, it's important to understand input costs, bank lending dynamics, natural and regulatory competitive advantages, resource dependence and behavioral, public choice and financial effects of price swings, etc, when looking at trade.
Note how complicated and convoluted this post has been. It's supposed to be that way. This stuff is not simple, and the argument shouldn't be simple. There are a lot of pieces to understanding economic policy, and when people address these in part without a full acknowledgment, they're doing a disservice.
I hope this somewhat explains my positions on issues as a marginal libertarian - simpler regulation (not even necessarily less regulation, just faster, clearer, shorter, more understandable regulation that's more flexible in the face of innovation and a changing world), lower taxes, much lower government spending and bureaucracy, more aggressive monetary policy (but transparent and indexed to either price level of NGDP level), free trade but with a semi-managed currency to offset foreign currency management, and significant and stable international capital investment by the US to diversify economies. I'm all for job retraining and short term bridge unemployment benefits, but against anything that saps those incentives (like 99 week UI). I'm pro- universal healthcare, but think the route to that is significant deregulation of health insurance coupled with a fixed voucher, not single payer. Etc.
Incentives matter, but it's important to work through the logic. Very few economists I read anywhere actually think about all these areas when putting forward opinions.
