Divided We Rule

In Divided We Govern, David Mayhew challenges conventional theories regarding both the ability of divided government to legislate effectively and the propensity for divided government to launch high salience investigations.  Mayhew concludes that based on an examination of the amount and type of legislation passed, as well as the number and type of investigations opened, there is in practice little to no difference in these areas between divided and united government.  The following is a short test to confirm (or negate) Mayhew’s claims.

The interpretive difficulty with Mayhew’s theory is that as sound as his examination appears, it still remains a bit too much to chew, at least from the policy making perspective.  As Mayhew states in the introduction, a common belief regarding divided government is that it deepens the balance of political power in Washington (in the same manner as separation of powers) thus keeping the parties in check, the result of which is legislation that is more often inconsequential.  In theory this sounds like it should be true, but in practice Mayhew demonstrates that there really is no significant difference either in the amount of legislation passed or the quality of legislation passed.

However, quality of legislation or importance is a fairly opinionated standard.  In general, Mayhew appears to qualify quality legislation as such if it represents a significant development (for instance a new agency) or greatly reforms something that already exists.  Even though Mayhew attempts to combine journalistic accounts and books on the subject to determine as accurately as possible public opinion of the legislation’s importance at the time it was passed, there is still a huge amount of personal opinion involved.  For instance, take the quote regarding the 1951 session of Congress during Truman’s presidency: “legislatively, there has been nothing in the way of initiation, no new ideas” (Mayhew 38).  While likely true, this journalistic observation also goes to show that importance was judged by the newness of the thing and not necessarily by its quality.

There should be a method of testing Mayhew’s assumptions without relying too heavily on subjective standards.  One such test is the Presidential Box Score, “which divides [the] number of Presidential requests enacted by the number of Presidential requests” (Mayhew 35).  Mayhew discards the Presidential Box Score’s viability as a test because first it focuses too much on the President’s ideas (Congress doesn’t need to rely on the President to come up with important legislation) and second it weighs them all the same (big or small) such that there is no critical measure of importance.
In many ways, although it is not stated, the items Mayhew eventually chooses are items that significantly alter either government’s expenditure or government’s size (which also affects expenditure).  If this is the case, then Mayhew’s theory can be tested using a completely different standard of criteria, federal expenditure: the advantage being that money is far less opinionated that a journalist.  To accomplish this standard measure, a simple dollar amount will not work as it disregards both inflation and the natural growth of the Gross Domestic Product (GDP) (and hence growth of tax revenues).  Thus, we convert the total sum of federal spending for a given fiscal year into a percentage of the GDP for that year to obtain a far more comparable standard by which to judge the performance of each legislative cycle.  Because Mayhew focuses between 1946 and 2002, our data will include fiscal years 1946 through 2003 (2003 is included for good measure as the overlap of fiscal years makes it such that part of the year’s spending is set by the previous congressional session).

The following chart shows first the year, followed by the arrangement of party control D for Democrat R for Republican in the order President, Senate, House.  The next column is total federal spending for that year while the last column is the U.S. GDP for that year (raw, not real.  For purposes of comparison inflationary adjustments are unnecessary because spending is presented as a percentage and is not adjusted for inflation).  The spending and GDP data comes directly from USGovernmentSpending.com .  In order to better see the difference not only between divided and united party control but also between various arrangements of divided party control the data has been sorted first by the arrangement of party control, next by spending, and finally by Fiscal Year.

1957    RDD    17.74    461.1
1973    RDD    17.77    1,382.70
1974    RDD    17.96    1,500.00
1958    RDD    18.42    467.2
1959    RDD    18.46    506.6
1960    RDD    18.48    526.4
1972    RDD    18.63    1,238.30
1969    RDD    18.65    984.6
1971    RDD    18.65    1,127.10
1970    RDD    18.84    1,038.50
1975    RDD    20.29    1,638.30
1976    RDD    20.37    1,825.30
1988    RDD    20.86    5,103.80
1989    RDD    20.86    5484.4
1987    RDD    21.19    4,739.50
1990    RDD    21.59    5,803.10
1992    RDD    21.8    6,337.70
1991    RDD    22.09    5,995.90

2001    RDR    18.4    10,128.00
2002    RDR    19.21    10,469.60

1984    RRD    21.66    3,933.20
1981    RRD    21.68    3,128.40
1986    RRD    22.19    4,462.80
1985    RRD    22.42    4,220.30
1983    RRD    22.86    3,536.70
1982    RRD    22.91    3,255.00

2003    RRR    19.71    10,960.80
1954    RRR    20.42    380.4
1953    RRR    21.08    379.4

Now for the statistics.
In the time period, the average expenditure was 19.46% while the median expenditure was 19.25% and the mode expenditure (included for interests’ sake) was a tie (at 2 each) between 18.65% and 20.86% (interestingly the average of the modes is 19.76%).  Our extreme low is 13.22% in 1948 while our extreme high is 29.93% in 1946 (fallout from World War II and great depression policies).  The average mode is about a quarter of a percent above the overall average while the median is about a quarter of a percent below it, which tells us our average is a fairly good midpoint.
Looking at just the averages for each of the respective party arrangements, we see that with just democrats in office spending averages 19.10%.  With a Democratic President and Republican control of both House and Senate, the average falls significantly to 18.17%. With a Republican President and Democratic control of both the House and Senate, spending is slightly below the overall average at 19.40%.  With a Republican President and House and a Democratic Senate (2001-2002) the average is 18.81%.  The two arrangements with the largest amount of spending are a Republican President and Senate with a Democratic House (22.29%) and unified Republican government (all three branches), 20.40% (not including 2003, the average becomes 20.75%).
What do these numbers tell us?  Lets look at them in a chart, the first column in the arrangement (President Senate House), the second column is the average spending percentage, while the third column is the number of times this arrangement appeared between 1946 and 2003.

Arrangement (PHS)    Spending (%)    Occurred
DDD                                 19.1                         19
DRR                                 18.17                        8
RDD                                 19.4                        20
RDR                                18.81                         2
RRD                                22.29                        6
RRR                                 20.4                         3
Unified                           19.28                       22
Divided                           19.58                       36

It should be noted that the Democrats had a platform policy shift away from sound fiscal policy in the late 60s, and that a split average beginning in 1967 gives the Democrats an average of 20.26% through 8 fiscal years with an all Democrat arrangement and an average (not including 1946) of 17.16% over 10 fiscal years.  That withstanding, the data shows that from a purely spending perspective there is little to no difference between united and divided government, which would suggest that Mayhew is in fact correct.  However, the data also suggests that certain divided arrangements are more fiscally constrained than others, specifically the arrangements DRR and RDR.  In other words, divided government may have an impact upon policy making in the cases of certain arrangements and not others.
In conclusion, Mayhew’s theory is supported by an elementary examination of federal spending data compared on a year-by-year basis to the GDP.  As Mayhew pointed out, there are 8 possible permutations of divided government; it may be the case that some of these affect policy making more than others.  A comparison, such as this, is less biased that a test designed to measure the importance of legislation passed, while at the same time it accomplishes the same purpose because important legislation in most cases will measurably affect government expenditure.  At the same time, a potential drawback in my approach is that I made no effort to distinguish between domestic and military/foreign spending while Mayhew constrains himself to domestic legislation.

Popularity: -0% [?]