University Towns and Recession risk

The time has come for me to start looking for new apartments in the US. The logical question has

appeared: What is the best area to rent/buy an apartment?

The simplest answer is “Go call your realtor and ask this question to him/her!.” However, I decided to think a bit independently. Here is the first part of my housing price analysis research.

First thing first, I feel lucky to get a job not far from one big university’s campus. It means if I get an apartment in university town it’ll be a convenient to drive to work and home without traffic and in shortest time.

But does it give any benefits to rent/buy an apartment in university city? One of the variables which should be disclosed is a correlation between price volatility and GDP. The 2008 crisis had teached us that it could pass a decade before the real estate price can come back to the pre-crisis level. As the real estate prices now near those all-time-high levels again it would be a shame to buy at the highest price just before the new recession.

Since price in university cities depend more on school’s costs and the vast majority of renters is students, the housing price supposes to be less sensitive to the entire real estate market.

Here is the primary hypothesis:

“University towns have their mean housing prices less affected by recessions.”

To prove or refute this hypothesis I’ll calculate a p-value for university and not-university cities price ratio correlation.

Price ratio calculated by this formula:

  
    Un_sity= pd.read_table('university_towns.txt', header=None)
    tmp = ''
    deldubl= False
    Un_sity[1] = [0 for i in range(len(Un_sity))]
    for i in range(len(Un_sity)):
        if '[edit]' in Un_sity.iloc[i,0]:
             tmp= Un_sity.iloc[i,0].split('[')[0]        
        Un_sity.iloc[i,1] = tmp
        if '(' in Un_sity.iloc[i,0]:
            Un_sity.iloc[i,0] = Un_sity.iloc[i,0].split('(')[0]
    Un_sity.columns = ["RegionName","State"]
    Un_sity['RegionName'] = Un_sity['RegionName'].apply(lambda x: x.strip(' '))   
    Un_sity = (Un_sity[Un_sity.RegionName.str.contains('[edit]') == False]
               .reset_index(drop=True))
    col = Un_sity.columns.tolist()
    col = col[-1:] + col[:-1]
    Un_sity = Un_sity[col]
    return Un_sity

get_list_of_university_towns()

def is_test_true():
        if p_val < 0.01:
            return True
        else:
            return False
    def is_better():
        if not_uni.mean() < is_uni.mean():
            return 'non-university town'
        else:
            return 'university town'
    df, not_uni, is_uni = cut_to_cohort('2006q4','2009q2')
    p_val = list(ttest_ind(not_uni,is_uni))[1]
    return (is_test_true(),p_val,is_better())

run_ttest()

(True, 1.7900059546810142e-05, 'university town')